bump

- Add support for Upper or Lower inputs.
- Add supports for sparse RHS
- Remove transposed cases, remove ordering method interface
- Add full access to PARDISO parameters

.hgeol

@@ -1,3 +1,8 @@
-[patterns]
-**.* = native
-eigen_autoexp_part.dat = CRLF
+[patterns]
+scripts/*.in = LF
+debug/msvc/*.dat = CRLF
+unsupported/test/mpreal/*.* = CRLF
+** = native
+
+[repository]
+native = LF

CMakeLists.txt

@@ -64,6 +64,10 @@ set(CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake)
 
 find_package(StandardMathLibrary)
 
+
+set(EIGEN_TEST_CUSTOM_LINKER_FLAGS  "" CACHE STRING "Additional linker flags when linking unit tests.")
+set(EIGEN_TEST_CUSTOM_CXX_FLAGS     "" CACHE STRING "Additional compiler flags when compiling unit tests.")
+
 set(EIGEN_STANDARD_LIBRARIES_TO_LINK_TO "")
 
 if(NOT STANDARD_MATH_LIBRARY_FOUND)
@@ -101,6 +105,10 @@ if(EIGEN_DEFAULT_TO_ROW_MAJOR)
   add_definitions("-DEIGEN_DEFAULT_TO_ROW_MAJOR")
 endif()
 
+add_definitions("-DEIGEN_PERMANENTLY_DISABLE_STUPID_WARNINGS")
+
+set(EIGEN_TEST_MAX_SIZE "320" CACHE STRING "Maximal matrix/vector size, default is 320")
+
 if(CMAKE_COMPILER_IS_GNUCXX)
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wnon-virtual-dtor -Wno-long-long -ansi -Wundef -Wcast-align -Wchar-subscripts -Wall -W -Wpointer-arith -Wwrite-strings -Wformat-security -fexceptions -fno-check-new -fno-common -fstrict-aliasing")
   set(CMAKE_CXX_FLAGS_DEBUG "-g3")
@@ -205,6 +213,7 @@ endif(MSVC)
 
 option(EIGEN_TEST_NO_EXPLICIT_VECTORIZATION "Disable explicit vectorization in tests/examples" OFF)
 option(EIGEN_TEST_X87 "Force using X87 instructions. Implies no vectorization." OFF)
+option(EIGEN_TEST_32BIT "Force generating 32bit code." OFF)
 
 if(EIGEN_TEST_X87)
   set(EIGEN_TEST_NO_EXPLICIT_VECTORIZATION ON)
@@ -216,6 +225,15 @@ if(EIGEN_TEST_X87)
   endif()
 endif()
 
+if(EIGEN_TEST_32BIT)
+  if(CMAKE_COMPILER_IS_GNUCXX)
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -m32")
+    message(STATUS "Forcing generation of 32-bit code in tests/examples")
+  else()
+    message(STATUS "EIGEN_TEST_32BIT ignored on your compiler")
+  endif()
+endif()
+
 if(EIGEN_TEST_NO_EXPLICIT_VECTORIZATION)
   add_definitions(-DEIGEN_DONT_VECTORIZE=1)
   message(STATUS "Disabling vectorization in tests/examples")
@@ -267,9 +285,21 @@ install(FILES
   )
 
 if(EIGEN_BUILD_PKGCONFIG)
+    SET(path_separator ":")
+    STRING(REPLACE ${path_separator} ";" pkg_config_libdir_search "$ENV{PKG_CONFIG_LIBDIR}")
+    message(STATUS "searching for 'pkgconfig' directory in PKG_CONFIG_LIBDIR ( $ENV{PKG_CONFIG_LIBDIR} ), ${CMAKE_INSTALL_PREFIX}/share, and ${CMAKE_INSTALL_PREFIX}/lib")
+    FIND_PATH(pkg_config_libdir pkgconfig ${pkg_config_libdir_search} ${CMAKE_INSTALL_PREFIX}/share ${CMAKE_INSTALL_PREFIX}/lib ${pkg_config_libdir_search})
+    if(pkg_config_libdir)
+        SET(pkg_config_install_dir ${pkg_config_libdir})
+        message(STATUS "found ${pkg_config_libdir}/pkgconfig" )
+    else(pkg_config_libdir)
+        SET(pkg_config_install_dir ${CMAKE_INSTALL_PREFIX}/share)
+        message(STATUS "pkgconfig not found; installing in ${pkg_config_install_dir}" )
+    endif(pkg_config_libdir)
+
     configure_file(eigen3.pc.in eigen3.pc)
     install(FILES ${CMAKE_CURRENT_BINARY_DIR}/eigen3.pc
-        DESTINATION share/pkgconfig
+        DESTINATION ${pkg_config_install_dir}/pkgconfig
         )
 endif(EIGEN_BUILD_PKGCONFIG)
 
@@ -277,10 +307,10 @@ add_subdirectory(Eigen)
 
 add_subdirectory(doc EXCLUDE_FROM_ALL)
 
-include(CTest)
+include(EigenConfigureTesting)
+# fixme, not sure this line is still needed:
 enable_testing() # must be called from the root CMakeLists, see man page
-include(EigenTesting)
-ei_init_testing()
+
 
 if(EIGEN_LEAVE_TEST_IN_ALL_TARGET)
   add_subdirectory(test) # can't do EXCLUDE_FROM_ALL here, breaks CTest
@@ -288,20 +318,18 @@ else()
   add_subdirectory(test EXCLUDE_FROM_ALL)
 endif()
 
+if(EIGEN_LEAVE_TEST_IN_ALL_TARGET)
+  add_subdirectory(blas)
+  add_subdirectory(lapack)
+else()
+  add_subdirectory(blas EXCLUDE_FROM_ALL)
+  add_subdirectory(lapack EXCLUDE_FROM_ALL)
+endif()
+
 add_subdirectory(unsupported)
 
 add_subdirectory(demos EXCLUDE_FROM_ALL)
 
-if(NOT MSVC)
-  if(EIGEN_LEAVE_TEST_IN_ALL_TARGET)
-    add_subdirectory(blas)
-    add_subdirectory(lapack)
-  else()
-    add_subdirectory(blas EXCLUDE_FROM_ALL)
-    add_subdirectory(lapack EXCLUDE_FROM_ALL)
-  endif()
-endif(NOT MSVC)
-
 # must be after test and unsupported, for configuring buildtests.in
 add_subdirectory(scripts EXCLUDE_FROM_ALL)
 

COPYING.BSD

@@ -0,0 +1,26 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/

CTestConfig.cmake

@@ -11,7 +11,3 @@ set(CTEST_DROP_METHOD "http")
 set(CTEST_DROP_SITE "eigen.tuxfamily.org")
 set(CTEST_DROP_LOCATION "/CDash/submit.php?project=Eigen")
 set(CTEST_DROP_SITE_CDASH TRUE)
-
-## A tribute to Dynamic!
-set(CTEST_CUSTOM_MAXIMUM_NUMBER_OF_WARNINGS "33331")
-set(CTEST_CUSTOM_MAXIMUM_NUMBER_OF_ERRORS "33331")

CTestCustom.cmake.in

@@ -0,0 +1,4 @@
+
+## A tribute to Dynamic!
+set(CTEST_CUSTOM_MAXIMUM_NUMBER_OF_WARNINGS "33331")
+set(CTEST_CUSTOM_MAXIMUM_NUMBER_OF_ERRORS "33331")

Eigen/Cholesky

@@ -3,7 +3,7 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 namespace Eigen {
 
@@ -24,10 +24,13 @@ namespace Eigen {
 #include "src/misc/Solve.h"
 #include "src/Cholesky/LLT.h"
 #include "src/Cholesky/LDLT.h"
+#ifdef EIGEN_USE_LAPACKE
+#include "src/Cholesky/LLT_MKL.h"
+#endif
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_CHOLESKY_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/CholmodSupport

@@ -0,0 +1,34 @@
+#ifndef EIGEN_CHOLMODSUPPORT_MODULE_H
+#define EIGEN_CHOLMODSUPPORT_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+extern "C" {
+  #include <cholmod.h>
+}
+
+namespace Eigen {
+
+/** \ingroup Support_modules
+  * \defgroup CholmodSupport_Module CholmodSupport module
+  *
+  *
+  * \code
+  * #include <Eigen/CholmodSupport>
+  * \endcode
+  */
+
+#include "src/misc/Solve.h"
+#include "src/misc/SparseSolve.h"
+
+#include "src/CholmodSupport/CholmodSupport.h"
+
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_CHOLMODSUPPORT_MODULE_H
+

Eigen/Core

@@ -26,14 +26,18 @@
 #ifndef EIGEN_CORE_H
 #define EIGEN_CORE_H
 
-// first thing Eigen does: prevent MSVC from committing suicide
-#include "src/Core/util/DisableMSVCWarnings.h"
+// first thing Eigen does: stop the compiler from committing suicide
+#include "src/Core/util/DisableStupidWarnings.h"
 
 // then include this file where all our macros are defined. It's really important to do it first because
 // it's where we do all the alignment settings (platform detection and honoring the user's will if he
 // defined e.g. EIGEN_DONT_ALIGN) so it needs to be done before we do anything with vectorization.
 #include "src/Core/util/Macros.h"
 
+// this include file manages BLAS and MKL related macros
+// and inclusion of their respective header files
+#include "src/Core/util/MKL_support.h"
+
 // if alignment is disabled, then disable vectorization. Note: EIGEN_ALIGN is the proper check, it takes into
 // account both the user's will (EIGEN_DONT_ALIGN) and our own platform checks
 #if !EIGEN_ALIGN
@@ -51,16 +55,16 @@
       #define EIGEN_SSE2_ON_MSVC_2008_OR_LATER
     #endif
   #endif
-#endif
-
-// Remember that usage of defined() in a #define is undefined by the standard
-#if (defined __SSE2__) && ( (!defined __GNUC__) || EIGEN_GNUC_AT_LEAST(4,2) )
-  #define EIGEN_SSE2_BUT_NOT_OLD_GCC
+#else
+  // Remember that usage of defined() in a #define is undefined by the standard
+  #if (defined __SSE2__) && ( (!defined __GNUC__) || EIGEN_GNUC_AT_LEAST(4,2) )
+    #define EIGEN_SSE2_ON_NON_MSVC_BUT_NOT_OLD_GCC
+  #endif
 #endif
 
 #ifndef EIGEN_DONT_VECTORIZE
 
-  #if defined (EIGEN_SSE2_BUT_NOT_OLD_GCC) || defined(EIGEN_SSE2_ON_MSVC_2008_OR_LATER)
+  #if defined (EIGEN_SSE2_ON_NON_MSVC_BUT_NOT_OLD_GCC) || defined(EIGEN_SSE2_ON_MSVC_2008_OR_LATER)
 
     // Defines symbols for compile-time detection of which instructions are
     // used.
@@ -143,6 +147,7 @@
 #ifdef EIGEN_HAS_ERRNO
 #include <cerrno>
 #endif
+#include <cstddef>
 #include <cstdlib>
 #include <cmath>
 #include <complex>
@@ -158,7 +163,7 @@
 
 // for outputting debug info
 #ifdef EIGEN_DEBUG_ASSIGN
-#include<iostream>
+#include <iostream>
 #endif
 
 // required for __cpuid, needs to be included after cmath
@@ -166,7 +171,7 @@
   #include <intrin.h>
 #endif
 
-#if (defined(_CPPUNWIND) || defined(__EXCEPTIONS)) && !defined(EIGEN_NO_EXCEPTIONS)
+#if defined(_CPPUNWIND) || defined(__EXCEPTIONS)
   #define EIGEN_EXCEPTIONS
 #endif
 
@@ -174,16 +179,7 @@
   #include <new>
 #endif
 
-// this needs to be done after all possible windows C header includes and before any Eigen source includes
-// (system C++ includes are supposed to be able to deal with this already):
-// windows.h defines min and max macros which would make Eigen fail to compile.
-#if defined(min) || defined(max)
-#error The preprocessor symbols 'min' or 'max' are defined. If you are compiling on Windows, do #define NOMINMAX to prevent windows.h from defining these symbols.
-#endif
-
-// defined in bits/termios.h
-#undef B0
-
+/** \brief Namespace containing all symbols from the %Eigen library. */
 namespace Eigen {
 
 inline static const char *SimdInstructionSetsInUse(void) {
@@ -239,6 +235,8 @@ inline static const char *SimdInstructionSetsInUse(void) {
 // we use size_t frequently and we'll never remember to prepend it with std:: everytime just to
 // ensure QNX/QCC support
 using std::size_t;
+// gcc 4.6.0 wants std:: for ptrdiff_t 
+using std::ptrdiff_t;
 
 /** \defgroup Core_Module Core module
   * This is the main module of Eigen providing dense matrix and vector support
@@ -250,6 +248,10 @@ using std::size_t;
   * \endcode
   */
 
+/** \defgroup Support_modules Support modules [category]
+  * Category of modules which add support for external libraries.
+  */
+
 #include "src/Core/util/Constants.h"
 #include "src/Core/util/ForwardDeclarations.h"
 #include "src/Core/util/Meta.h"
@@ -321,7 +323,7 @@ using std::size_t;
 #include "src/Core/CommaInitializer.h"
 #include "src/Core/Flagged.h"
 #include "src/Core/ProductBase.h"
-#include "src/Core/Product.h"
+#include "src/Core/GeneralProduct.h"
 #include "src/Core/TriangularMatrix.h"
 #include "src/Core/SelfAdjointView.h"
 #include "src/Core/SolveTriangular.h"
@@ -350,11 +352,26 @@ using std::size_t;
 #include "src/Core/ArrayBase.h"
 #include "src/Core/ArrayWrapper.h"
 
+#ifdef EIGEN_USE_BLAS
+#include "src/Core/products/GeneralMatrixMatrix_MKL.h"
+#include "src/Core/products/GeneralMatrixVector_MKL.h"
+#include "src/Core/products/GeneralMatrixMatrixTriangular_MKL.h"
+#include "src/Core/products/SelfadjointMatrixMatrix_MKL.h"
+#include "src/Core/products/SelfadjointMatrixVector_MKL.h"
+#include "src/Core/products/TriangularMatrixMatrix_MKL.h"
+#include "src/Core/products/TriangularMatrixVector_MKL.h"
+#include "src/Core/products/TriangularSolverMatrix_MKL.h"
+#endif // EIGEN_USE_BLAS
+
+#ifdef EIGEN_USE_MKL_VML
+#include "src/Core/Assign_MKL.h"
+#endif
+
 } // namespace Eigen
 
 #include "src/Core/GlobalFunctions.h"
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #ifdef EIGEN2_SUPPORT
 #include "Eigen2Support"

Eigen/Eigen2Support

@@ -29,11 +29,12 @@
 #error Eigen2 support must be enabled by defining EIGEN2_SUPPORT before including any Eigen header
 #endif
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 namespace Eigen {
 
-/** \defgroup Eigen2Support_Module Eigen2 support module
+/** \ingroup Support_modules
+  * \defgroup Eigen2Support_Module Eigen2 support module
   * This module provides a couple of deprecated functions improving the compatibility with Eigen2.
   *
   * To use it, define EIGEN2_SUPPORT before including any Eigen header
@@ -58,11 +59,29 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 // Eigen2 used to include iostream
 #include<iostream>
 
+#define EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, SizeSuffix) \
+using Eigen::Matrix##SizeSuffix##TypeSuffix; \
+using Eigen::Vector##SizeSuffix##TypeSuffix; \
+using Eigen::RowVector##SizeSuffix##TypeSuffix;
+
+#define EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(TypeSuffix) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, 2) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, 3) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, 4) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, X) \
+
+#define EIGEN_USING_MATRIX_TYPEDEFS \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(i) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(f) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(d) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(cf) \
+EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(cd)
+
 #define USING_PART_OF_NAMESPACE_EIGEN \
 EIGEN_USING_MATRIX_TYPEDEFS \
 using Eigen::Matrix; \

Eigen/Eigenvalues

@@ -3,12 +3,13 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 #include "Cholesky"
 #include "Jacobi"
 #include "Householder"
 #include "LU"
+#include "Geometry"
 
 namespace Eigen {
 
@@ -35,10 +36,15 @@ namespace Eigen {
 #include "src/Eigenvalues/ComplexSchur.h"
 #include "src/Eigenvalues/ComplexEigenSolver.h"
 #include "src/Eigenvalues/MatrixBaseEigenvalues.h"
+#ifdef EIGEN_USE_LAPACKE
+#include "src/Eigenvalues/RealSchur_MKL.h"
+#include "src/Eigenvalues/ComplexSchur_MKL.h"
+#include "src/Eigenvalues/SelfAdjointEigenSolver_MKL.h"
+#endif
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_EIGENVALUES_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/Geometry

@@ -3,7 +3,7 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 #include "SVD"
 #include "LU"
@@ -60,7 +60,7 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_GEOMETRY_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/Householder

@@ -3,7 +3,7 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 namespace Eigen {
 
@@ -21,7 +21,7 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_HOUSEHOLDER_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/IterativeLinearSolvers

@@ -0,0 +1,37 @@
+#ifndef EIGEN_ITERATIVELINEARSOLVERS_MODULE_H
+#define EIGEN_ITERATIVELINEARSOLVERS_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+namespace Eigen {
+
+/** \ingroup Sparse_modules
+  * \defgroup IterativeLinearSolvers_Module IterativeLinearSolvers module
+  *
+  * This module currently provides iterative methods to solve problems of the form \c A \c x = \c b, where \c A is a squared matrix, usually very large and sparse.
+  * Those solvers are accessible via the following classes:
+  *  - ConjugateGradient for selfadjoint (hermitian) matrices,
+  *  - BiCGSTAB for general square matrices.
+  *
+  * Such problems can also be solved using the direct sparse decomposition modules: SparseCholesky, CholmodSupport, UmfPackSupport, SuperLUSupport.
+  *
+  * \code
+  * #include <Eigen/IterativeLinearSolvers>
+  * \endcode
+  */
+
+#include "src/misc/Solve.h"
+#include "src/misc/SparseSolve.h"
+
+#include "src/IterativeLinearSolvers/IterativeSolverBase.h"
+#include "src/IterativeLinearSolvers/BasicPreconditioners.h"
+#include "src/IterativeLinearSolvers/ConjugateGradient.h"
+#include "src/IterativeLinearSolvers/BiCGSTAB.h"
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_ITERATIVELINEARSOLVERS_MODULE_H

Eigen/Jacobi

@@ -3,7 +3,7 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 namespace Eigen {
 
@@ -23,7 +23,7 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_JACOBI_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/LU

@@ -3,7 +3,7 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 namespace Eigen {
 
@@ -23,6 +23,9 @@ namespace Eigen {
 #include "src/misc/Image.h"
 #include "src/LU/FullPivLU.h"
 #include "src/LU/PartialPivLU.h"
+#ifdef EIGEN_USE_LAPACKE
+#include "src/LU/PartialPivLU_MKL.h"
+#endif
 #include "src/LU/Determinant.h"
 #include "src/LU/Inverse.h"
 
@@ -36,7 +39,7 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_LU_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/LeastSquares

@@ -5,9 +5,12 @@
 #error LeastSquares is only available in Eigen2 support mode (define EIGEN2_SUPPORT)
 #endif
 
+// exclude from normal eigen3-only documentation
+#ifdef EIGEN2_SUPPORT
+
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 #include "Eigenvalues"
 #include "Geometry"
@@ -26,6 +29,8 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN2_SUPPORT
 
 #endif // EIGEN_REGRESSION_MODULE_H

Eigen/OrderingMethods

@@ -0,0 +1,27 @@
+#ifndef EIGEN_ORDERINGMETHODS_MODULE_H
+#define EIGEN_ORDERINGMETHODS_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+namespace Eigen {
+
+/** \ingroup Sparse_modules
+  * \defgroup OrderingMethods_Module OrderingMethods module
+  *
+  * This module is currently for internal use only.
+  *
+  *
+  * \code
+  * #include <Eigen/OrderingMethods>
+  * \endcode
+  */
+
+#include "src/OrderingMethods/Amd.h"
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_ORDERINGMETHODS_MODULE_H

Eigen/PARDISOSupport

@@ -0,0 +1,30 @@
+#ifndef EIGEN_PARDISOSUPPORT_MODULE_H
+#define EIGEN_PARDISOSUPPORT_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+#include <mkl_pardiso.h>
+
+#include <unsupported/Eigen/SparseExtra>
+
+namespace Eigen {
+
+/** \ingroup Support_modules
+  * \defgroup PARDISOSupport_Module PARDISOSupport module
+  *
+  * This module brings support for the Intel(R) MKL PARDISO direct sparse solvers
+  *
+  * \code
+  * #include <Eigen/PARDISOSupport>
+  * \endcode
+  */
+
+#include "src/PARDISOSupport/PARDISOSupport.h"
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_PARDISOSUPPORT_MODULE_H

Eigen/QR

@@ -3,7 +3,7 @@
 
 #include "Core"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 #include "Cholesky"
 #include "Jacobi"
@@ -28,6 +28,10 @@ namespace Eigen {
 #include "src/QR/HouseholderQR.h"
 #include "src/QR/FullPivHouseholderQR.h"
 #include "src/QR/ColPivHouseholderQR.h"
+#ifdef EIGEN_USE_LAPACKE
+#include "src/QR/HouseholderQR_MKL.h"
+#include "src/QR/ColPivHouseholderQR_MKL.h"
+#endif
 
 #ifdef EIGEN2_SUPPORT
 #include "src/Eigen2Support/QR.h"
@@ -35,7 +39,7 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #ifdef EIGEN2_SUPPORT
 #include "Eigenvalues"

Eigen/QtAlignedMalloc

@@ -6,7 +6,7 @@
 
 #if (!EIGEN_MALLOC_ALREADY_ALIGNED)
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 void *qMalloc(size_t size)
 {
@@ -26,7 +26,7 @@ void *qRealloc(void *ptr, size_t size)
   return newPtr;
 }
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif
 

Eigen/SVD

@@ -5,7 +5,7 @@
 #include "Householder"
 #include "Jacobi"
 
-#include "src/Core/util/DisableMSVCWarnings.h"
+#include "src/Core/util/DisableStupidWarnings.h"
 
 namespace Eigen {
 
@@ -13,9 +13,9 @@ namespace Eigen {
   *
   *
   *
-  * This module provides SVD decomposition for (currently) real matrices.
+  * This module provides SVD decomposition for matrices (both real and complex).
   * This decomposition is accessible via the following MatrixBase method:
-  *  - MatrixBase::svd()
+  *  - MatrixBase::jacobiSvd()
   *
   * \code
   * #include <Eigen/SVD>
@@ -24,6 +24,9 @@ namespace Eigen {
 
 #include "src/misc/Solve.h"
 #include "src/SVD/JacobiSVD.h"
+#if defined(EIGEN_USE_LAPACKE) && !defined(EIGEN_USE_LAPACKE_STRICT)
+#include "src/SVD/JacobiSVD_MKL.h"
+#endif
 #include "src/SVD/UpperBidiagonalization.h"
 
 #ifdef EIGEN2_SUPPORT
@@ -32,7 +35,7 @@ namespace Eigen {
 
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "src/Core/util/ReenableStupidWarnings.h"
 
 #endif // EIGEN_SVD_MODULE_H
 /* vim: set filetype=cpp et sw=2 ts=2 ai: */

Eigen/Sparse

@@ -1,69 +1,27 @@
 #ifndef EIGEN_SPARSE_MODULE_H
 #define EIGEN_SPARSE_MODULE_H
 
-#include "Core"
-
-#include "src/Core/util/DisableMSVCWarnings.h"
-
-#include <vector>
-#include <map>
-#include <cstdlib>
-#include <cstring>
-#include <algorithm>
-
-#ifdef EIGEN2_SUPPORT
-#define EIGEN_YES_I_KNOW_SPARSE_MODULE_IS_NOT_STABLE_YET
-#endif
-
-#ifndef EIGEN_YES_I_KNOW_SPARSE_MODULE_IS_NOT_STABLE_YET
-#error The sparse module API is not stable yet. To use it anyway, please define the EIGEN_YES_I_KNOW_SPARSE_MODULE_IS_NOT_STABLE_YET preprocessor token.
-#endif
-
 namespace Eigen {
 
-/** \defgroup Sparse_Module Sparse module
+/** \defgroup Sparse_modules Sparse modules
   *
-  *
-  *
-  * See the \ref TutorialSparse "Sparse tutorial"
+  * Meta-module including all related modules:
+  * - SparseCore
+  * - OrderingMethods
+  * - SparseCholesky
+  * - IterativeLinearSolvers
   *
   * \code
   * #include <Eigen/Sparse>
   * \endcode
   */
 
-/** The type used to identify a general sparse storage. */
-struct Sparse {};
-
-#include "src/Sparse/SparseUtil.h"
-#include "src/Sparse/SparseMatrixBase.h"
-#include "src/Sparse/CompressedStorage.h"
-#include "src/Sparse/AmbiVector.h"
-#include "src/Sparse/SparseMatrix.h"
-#include "src/Sparse/DynamicSparseMatrix.h"
-#include "src/Sparse/MappedSparseMatrix.h"
-#include "src/Sparse/SparseVector.h"
-#include "src/Sparse/CoreIterators.h"
-#include "src/Sparse/SparseBlock.h"
-#include "src/Sparse/SparseTranspose.h"
-#include "src/Sparse/SparseCwiseUnaryOp.h"
-#include "src/Sparse/SparseCwiseBinaryOp.h"
-#include "src/Sparse/SparseDot.h"
-#include "src/Sparse/SparseAssign.h"
-#include "src/Sparse/SparseRedux.h"
-#include "src/Sparse/SparseFuzzy.h"
-#include "src/Sparse/SparseProduct.h"
-#include "src/Sparse/SparseSparseProduct.h"
-#include "src/Sparse/SparseDenseProduct.h"
-#include "src/Sparse/SparseDiagonalProduct.h"
-#include "src/Sparse/SparseTriangularView.h"
-#include "src/Sparse/SparseSelfAdjointView.h"
-#include "src/Sparse/TriangularSolver.h"
-#include "src/Sparse/SparseView.h"
-
 } // namespace Eigen
 
-#include "src/Core/util/EnableMSVCWarnings.h"
+#include "SparseCore"
+#include "OrderingMethods"
+#include "SparseCholesky"
+#include "IterativeLinearSolvers"
 
 #endif // EIGEN_SPARSE_MODULE_H
 

Eigen/SparseCholesky

@@ -0,0 +1,34 @@
+#ifndef EIGEN_SPARSECHOLESKY_MODULE_H
+#define EIGEN_SPARSECHOLESKY_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+namespace Eigen {
+
+/** \ingroup Sparse_modules
+  * \defgroup SparseCholesky_Module SparseCholesky module
+  *
+  * This module currently provides two variants of the direct sparse Cholesky decomposition for selfadjoint (hermitian) matrices.
+  * Those decompositions are accessible via the following classes:
+  *  - SimplicialLLt,
+  *  - SimplicialLDLt
+  *
+  * Such problems can also be solved using the ConjugateGradient solver from the IterativeLinearSolvers module.
+  *
+  * \code
+  * #include <Eigen/SparseCholesky>
+  * \endcode
+  */
+
+#include "src/misc/Solve.h"
+#include "src/misc/SparseSolve.h"
+
+#include "src/SparseCholesky/SimplicialCholesky.h"
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_SPARSECHOLESKY_MODULE_H

Eigen/SparseCore

@@ -0,0 +1,65 @@
+#ifndef EIGEN_SPARSECORE_MODULE_H
+#define EIGEN_SPARSECORE_MODULE_H
+
+#include "Core"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+#include <vector>
+#include <map>
+#include <cstdlib>
+#include <cstring>
+#include <algorithm>
+
+namespace Eigen {
+
+/** \ingroup Sparse_modules
+  * \defgroup SparseCore_Module SparseCore module
+  *
+  * This module provides a sparse matrix representation, and basic associatd matrix manipulations
+  * and operations.
+  *
+  * See the \ref TutorialSparse "Sparse tutorial"
+  *
+  * \code
+  * #include <Eigen/SparseCore>
+  * \endcode
+  *
+  * This module depends on: Core.
+  */
+
+/** The type used to identify a general sparse storage. */
+struct Sparse {};
+
+#include "src/SparseCore/SparseUtil.h"
+#include "src/SparseCore/SparseMatrixBase.h"
+#include "src/SparseCore/CompressedStorage.h"
+#include "src/SparseCore/AmbiVector.h"
+#include "src/SparseCore/SparseMatrix.h"
+#include "src/SparseCore/MappedSparseMatrix.h"
+#include "src/SparseCore/SparseVector.h"
+#include "src/SparseCore/CoreIterators.h"
+#include "src/SparseCore/SparseBlock.h"
+#include "src/SparseCore/SparseTranspose.h"
+#include "src/SparseCore/SparseCwiseUnaryOp.h"
+#include "src/SparseCore/SparseCwiseBinaryOp.h"
+#include "src/SparseCore/SparseDot.h"
+#include "src/SparseCore/SparseAssign.h"
+#include "src/SparseCore/SparseRedux.h"
+#include "src/SparseCore/SparseFuzzy.h"
+#include "src/SparseCore/ConservativeSparseSparseProduct.h"
+#include "src/SparseCore/SparseSparseProductWithPruning.h"
+#include "src/SparseCore/SparseProduct.h"
+#include "src/SparseCore/SparseDenseProduct.h"
+#include "src/SparseCore/SparseDiagonalProduct.h"
+#include "src/SparseCore/SparseTriangularView.h"
+#include "src/SparseCore/SparseSelfAdjointView.h"
+#include "src/SparseCore/TriangularSolver.h"
+#include "src/SparseCore/SparseView.h"
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_SPARSECORE_MODULE_H
+

Eigen/SuperLUSupport

@@ -0,0 +1,53 @@
+#ifndef EIGEN_SUPERLUSUPPORT_MODULE_H
+#define EIGEN_SUPERLUSUPPORT_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+#ifdef EMPTY
+#define EIGEN_EMPTY_WAS_ALREADY_DEFINED
+#endif
+
+typedef int int_t;
+#include <slu_Cnames.h>
+#include <supermatrix.h>
+#include <slu_util.h>
+
+// slu_util.h defines a preprocessor token named EMPTY which is really polluting,
+// so we remove it in favor of a SUPERLU_EMPTY token.
+// If EMPTY was already, defined then we don't undef it.
+
+#if defined(EIGEN_EMPTY_WAS_ALREADY_DEFINED)
+# undef EIGEN_EMPTY_WAS_ALREADY_DEFINED
+#elif defined(EMPTY)
+# undef EMPTY
+#endif
+
+#define SUPERLU_EMPTY (-1)
+
+namespace Eigen { struct SluMatrix; }
+
+namespace Eigen {
+
+/** \ingroup Support_modules
+  * \defgroup SuperLUSupport_Module SuperLUSupport module
+  *
+  * \warning When including this module, you have to use SUPERLU_EMPTY instead of EMPTY which is no longer defined because it is too polluting.
+  *
+  * \code
+  * #include <Eigen/SuperLUSupport>
+  * \endcode
+  */
+
+#include "src/misc/Solve.h"
+#include "src/misc/SparseSolve.h"
+
+#include "src/SuperLUSupport/SuperLUSupport.h"
+
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_SUPERLUSUPPORT_MODULE_H

Eigen/UmfPackSupport

@@ -0,0 +1,34 @@
+#ifndef EIGEN_UMFPACKSUPPORT_MODULE_H
+#define EIGEN_UMFPACKSUPPORT_MODULE_H
+
+#include "SparseCore"
+
+#include "src/Core/util/DisableStupidWarnings.h"
+
+extern "C" {
+#include <umfpack.h>
+}
+
+namespace Eigen {
+
+/** \ingroup Support_modules
+  * \defgroup UmfPackSupport_Module UmfPackSupport module
+  *
+  *
+  *
+  *
+  * \code
+  * #include <Eigen/UmfPackSupport>
+  * \endcode
+  */
+
+#include "src/misc/Solve.h"
+#include "src/misc/SparseSolve.h"
+
+#include "src/UmfPackSupport/UmfPackSupport.h"
+
+} // namespace Eigen
+
+#include "src/Core/util/ReenableStupidWarnings.h"
+
+#endif // EIGEN_UMFPACKSUPPORT_MODULE_H

Eigen/src/Cholesky/LDLT.h

@@ -1,9 +1,10 @@
 // This file is part of Eigen, a lightweight C++ template library
 // for linear algebra.
 //
-// Copyright (C) 2008-2010 Gael Guennebaud <gael.guennebaud@inria.fr>
+// Copyright (C) 2008-2011 Gael Guennebaud <gael.guennebaud@inria.fr>
 // Copyright (C) 2009 Keir Mierle <mierle@gmail.com>
 // Copyright (C) 2009 Benoit Jacob <jacob.benoit.1@gmail.com>
+// Copyright (C) 2011 Timothy E. Holy <tim.holy@gmail.com >
 //
 // Eigen is free software; you can redistribute it and/or
 // modify it under the terms of the GNU Lesser General Public
@@ -31,13 +32,15 @@ namespace internal {
 template<typename MatrixType, int UpLo> struct LDLT_Traits;
 }
 
-/** \ingroup cholesky_Module
+/** \ingroup Cholesky_Module
   *
   * \class LDLT
   *
   * \brief Robust Cholesky decomposition of a matrix with pivoting
   *
   * \param MatrixType the type of the matrix of which to compute the LDL^T Cholesky decomposition
+  * \param UpLo the triangular part that will be used for the decompositon: Lower (default) or Upper.
+  *             The other triangular part won't be read.
   *
   * Perform a robust Cholesky decomposition of a positive semidefinite or negative semidefinite
   * matrix \f$ A \f$ such that \f$ A =  P^TLDL^*P \f$, where P is a permutation matrix, L
@@ -48,14 +51,10 @@ template<typename MatrixType, int UpLo> struct LDLT_Traits;
   * on D also stabilizes the computation.
   *
   * Remember that Cholesky decompositions are not rank-revealing. Also, do not use a Cholesky
-	* decomposition to determine whether a system of equations has a solution.
+  * decomposition to determine whether a system of equations has a solution.
   *
   * \sa MatrixBase::ldlt(), class LLT
   */
- /* THIS PART OF THE DOX IS CURRENTLY DISABLED BECAUSE INACCURATE BECAUSE OF BUG IN THE DECOMPOSITION CODE
-  * Note that during the decomposition, only the upper triangular part of A is considered. Therefore,
-  * the strict lower part does not have to store correct values.
-  */
 template<typename _MatrixType, int _UpLo> class LDLT
 {
   public:
@@ -98,6 +97,11 @@ template<typename _MatrixType, int _UpLo> class LDLT
         m_isInitialized(false)
     {}
 
+    /** \brief Constructor with decomposition
+      *
+      * This calculates the decomposition for the input \a matrix.
+      * \sa LDLT(Index size)
+      */
     LDLT(const MatrixType& matrix)
       : m_matrix(matrix.rows(), matrix.cols()),
         m_transpositions(matrix.rows()),
@@ -107,6 +111,14 @@ template<typename _MatrixType, int _UpLo> class LDLT
       compute(matrix);
     }
 
+    /** Clear any existing decomposition
+     * \sa rankUpdate(w,sigma)
+     */
+    void setZero()
+    {
+      m_isInitialized = false;
+    }
+
     /** \returns a view of the upper triangular matrix U */
     inline typename Traits::MatrixU matrixU() const
     {
@@ -130,14 +142,14 @@ template<typename _MatrixType, int _UpLo> class LDLT
     }
 
     /** \returns the coefficients of the diagonal matrix D */
-    inline Diagonal<const MatrixType> vectorD(void) const
+    inline Diagonal<const MatrixType> vectorD() const
     {
       eigen_assert(m_isInitialized && "LDLT is not initialized.");
       return m_matrix.diagonal();
     }
 
     /** \returns true if the matrix is positive (semidefinite) */
-    inline bool isPositive(void) const
+    inline bool isPositive() const
     {
       eigen_assert(m_isInitialized && "LDLT is not initialized.");
       return m_sign == 1;
@@ -158,10 +170,19 @@ template<typename _MatrixType, int _UpLo> class LDLT
     }
 
     /** \returns a solution x of \f$ A x = b \f$ using the current decomposition of A.
+      *
+      * This function also supports in-place solves using the syntax <tt>x = decompositionObject.solve(x)</tt> .
       *
       * \note_about_checking_solutions
       *
-      * \sa solveInPlace(), MatrixBase::ldlt()
+      * More precisely, this method solves \f$ A x = b \f$ using the decomposition \f$ A = P^T L D L^* P \f$
+      * by solving the systems \f$ P^T y_1 = b \f$, \f$ L y_2 = y_1 \f$, \f$ D y_3 = y_2 \f$, 
+      * \f$ L^* y_4 = y_3 \f$ and \f$ P x = y_4 \f$ in succession. If the matrix \f$ A \f$ is singular, then
+      * \f$ D \f$ will also be singular (all the other matrices are invertible). In that case, the
+      * least-square solution of \f$ D y_3 = y_2 \f$ is computed. This does not mean that this function
+      * computes the least-square solution of \f$ A x = b \f$ is \f$ A \f$ is singular.
+      *
+      * \sa MatrixBase::ldlt()
       */
     template<typename Rhs>
     inline const internal::solve_retval<LDLT, Rhs>
@@ -187,6 +208,9 @@ template<typename _MatrixType, int _UpLo> class LDLT
 
     LDLT& compute(const MatrixType& matrix);
 
+    template <typename Derived>
+    LDLT& rankUpdate(const MatrixBase<Derived>& w,RealScalar alpha=1);
+
     /** \returns the internal LDLT decomposition matrix
       *
       * TODO: document the storage layout
@@ -202,6 +226,17 @@ template<typename _MatrixType, int _UpLo> class LDLT
     inline Index rows() const { return m_matrix.rows(); }
     inline Index cols() const { return m_matrix.cols(); }
 
+    /** \brief Reports whether previous computation was successful.
+      *
+      * \returns \c Success if computation was succesful,
+      *          \c NumericalIssue if the matrix.appears to be negative.
+      */
+    ComputationInfo info() const
+    {
+      eigen_assert(m_isInitialized && "LDLT is not initialized.");
+      return Success;
+    }
+
   protected:
 
     /** \internal
@@ -240,7 +275,7 @@ template<> struct ldlt_inplace<Lower>
       return true;
     }
 
-    RealScalar cutoff = 0, biggest_in_corner;
+    RealScalar cutoff(0), biggest_in_corner;
 
     for (Index k = 0; k < size; ++k)
     {
@@ -308,6 +343,61 @@ template<> struct ldlt_inplace<Lower>
 
     return true;
   }
+
+  // Reference for the algorithm: Davis and Hager, "Multiple Rank
+  // Modifications of a Sparse Cholesky Factorization" (Algorithm 1)
+  // Trivial rearrangements of their computations (Timothy E. Holy)
+  // allow their algorithm to work for rank-1 updates even if the
+  // original matrix is not of full rank.
+  // Here only rank-1 updates are implemented, to reduce the
+  // requirement for intermediate storage and improve accuracy
+  template<typename MatrixType, typename WDerived>
+  static bool updateInPlace(MatrixType& mat, MatrixBase<WDerived>& w, typename MatrixType::RealScalar sigma=1)
+  {
+    using internal::isfinite;
+    typedef typename MatrixType::Scalar Scalar;
+    typedef typename MatrixType::RealScalar RealScalar;
+    typedef typename MatrixType::Index Index;
+
+    const Index size = mat.rows();
+    eigen_assert(mat.cols() == size && w.size()==size);
+
+    RealScalar alpha = 1;
+
+    // Apply the update
+    for (Index j = 0; j < size; j++)
+    {
+      // Check for termination due to an original decomposition of low-rank
+      if (!isfinite(alpha))
+        break;
+
+      // Update the diagonal terms
+      RealScalar dj = real(mat.coeff(j,j));
+      Scalar wj = w.coeff(j);
+      RealScalar swj2 = sigma*abs2(wj);
+      RealScalar gamma = dj*alpha + swj2;
+
+      mat.coeffRef(j,j) += swj2/alpha;
+      alpha += swj2/dj;
+
+
+      // Update the terms of L
+      Index rs = size-j-1;
+      w.tail(rs) -= wj * mat.col(j).tail(rs);
+      if(gamma != 0)
+        mat.col(j).tail(rs) += (sigma*conj(wj)/gamma)*w.tail(rs);
+    }
+    return true;
+  }
+
+  template<typename MatrixType, typename TranspositionType, typename Workspace, typename WType>
+  static bool update(MatrixType& mat, const TranspositionType& transpositions, Workspace& tmp, const WType& w, typename MatrixType::RealScalar sigma=1)
+  {
+    // Apply the permutation to the input w
+    tmp = transpositions * w;
+
+    return ldlt_inplace<Lower>::updateInPlace(mat,tmp,sigma);
+  }
 };
 
 template<> struct ldlt_inplace<Upper>
@@ -318,22 +408,29 @@ template<> struct ldlt_inplace<Upper>
     Transpose<MatrixType> matt(mat);
     return ldlt_inplace<Lower>::unblocked(matt, transpositions, temp, sign);
   }
+
+  template<typename MatrixType, typename TranspositionType, typename Workspace, typename WType>
+  static EIGEN_STRONG_INLINE bool update(MatrixType& mat, TranspositionType& transpositions, Workspace& tmp, WType& w, typename MatrixType::RealScalar sigma=1)
+  {
+    Transpose<MatrixType> matt(mat);
+    return ldlt_inplace<Lower>::update(matt, transpositions, tmp, w.conjugate(), sigma);
+  }
 };
 
 template<typename MatrixType> struct LDLT_Traits<MatrixType,Lower>
 {
-  typedef TriangularView<MatrixType, UnitLower> MatrixL;
-  typedef TriangularView<typename MatrixType::AdjointReturnType, UnitUpper> MatrixU;
-  inline static MatrixL getL(const MatrixType& m) { return m; }
-  inline static MatrixU getU(const MatrixType& m) { return m.adjoint(); }
+  typedef TriangularView<const MatrixType, UnitLower> MatrixL;
+  typedef TriangularView<const typename MatrixType::AdjointReturnType, UnitUpper> MatrixU;
+  static inline MatrixL getL(const MatrixType& m) { return m; }
+  static inline MatrixU getU(const MatrixType& m) { return m.adjoint(); }
 };
 
 template<typename MatrixType> struct LDLT_Traits<MatrixType,Upper>
 {
-  typedef TriangularView<typename MatrixType::AdjointReturnType, UnitLower> MatrixL;
-  typedef TriangularView<MatrixType, UnitUpper> MatrixU;
-  inline static MatrixL getL(const MatrixType& m) { return m.adjoint(); }
-  inline static MatrixU getU(const MatrixType& m) { return m; }
+  typedef TriangularView<const typename MatrixType::AdjointReturnType, UnitLower> MatrixL;
+  typedef TriangularView<const MatrixType, UnitUpper> MatrixU;
+  static inline MatrixL getL(const MatrixType& m) { return m.adjoint(); }
+  static inline MatrixU getU(const MatrixType& m) { return m; }
 };
 
 } // end namespace internal
@@ -358,6 +455,37 @@ LDLT<MatrixType,_UpLo>& LDLT<MatrixType,_UpLo>::compute(const MatrixType& a)
   return *this;
 }
 
+/** Update the LDLT decomposition:  given A = L D L^T, efficiently compute the decomposition of A + sigma w w^T.
+ * \param w a vector to be incorporated into the decomposition.
+ * \param sigma a scalar, +1 for updates and -1 for "downdates," which correspond to removing previously-added column vectors. Optional; default value is +1.
+ * \sa setZero()
+  */
+template<typename MatrixType, int _UpLo>
+template<typename Derived>
+LDLT<MatrixType,_UpLo>& LDLT<MatrixType,_UpLo>::rankUpdate(const MatrixBase<Derived>& w,typename NumTraits<typename MatrixType::Scalar>::Real sigma)
+{
+  const Index size = w.rows();
+  if (m_isInitialized)
+  {
+    eigen_assert(m_matrix.rows()==size);
+  }
+  else
+  {    
+    m_matrix.resize(size,size);
+    m_matrix.setZero();
+    m_transpositions.resize(size);
+    for (Index i = 0; i < size; i++)
+      m_transpositions.coeffRef(i) = i;
+    m_temporary.resize(size);
+    m_sign = sigma;
+    m_isInitialized = true;
+  }
+
+  internal::ldlt_inplace<UpLo>::update(m_matrix, m_transpositions, m_temporary, w, sigma);
+
+  return *this;
+}
+
 namespace internal {
 template<typename _MatrixType, int _UpLo, typename Rhs>
 struct solve_retval<LDLT<_MatrixType,_UpLo>, Rhs>
@@ -376,7 +504,21 @@ struct solve_retval<LDLT<_MatrixType,_UpLo>, Rhs>
     dec().matrixL().solveInPlace(dst);
 
     // dst = D^-1 (L^-1 P b)
-    dst = dec().vectorD().asDiagonal().inverse() * dst;
+    // more precisely, use pseudo-inverse of D (see bug 241)
+    using std::abs;
+    using std::max;
+    typedef typename LDLTType::MatrixType MatrixType;
+    typedef typename LDLTType::Scalar Scalar;
+    typedef typename LDLTType::RealScalar RealScalar;
+    const Diagonal<const MatrixType> vectorD = dec().vectorD();
+    RealScalar tolerance = (max)(vectorD.array().abs().maxCoeff() * NumTraits<Scalar>::epsilon(),
+				 RealScalar(1) / NumTraits<RealScalar>::highest()); // motivated by LAPACK's xGELSS
+    for (Index i = 0; i < vectorD.size(); ++i) {
+      if(abs(vectorD(i)) > tolerance)
+	dst.row(i) /= vectorD(i);
+      else
+	dst.row(i).setZero();
+    }
 
     // dst = L^-T (D^-1 L^-1 P b)
     dec().matrixU().solveInPlace(dst);

Eigen/src/Cholesky/LLT.h

@@ -29,13 +29,15 @@ namespace internal{
 template<typename MatrixType, int UpLo> struct LLT_Traits;
 }
 
-/** \ingroup cholesky_Module
+/** \ingroup Cholesky_Module
   *
   * \class LLT
   *
   * \brief Standard Cholesky decomposition (LL^T) of a matrix and associated features
   *
   * \param MatrixType the type of the matrix of which we are computing the LL^T Cholesky decomposition
+  * \param UpLo the triangular part that will be used for the decompositon: Lower (default) or Upper.
+  *             The other triangular part won't be read.
   *
   * This class performs a LL^T Cholesky decomposition of a symmetric, positive definite
   * matrix A such that A = LL^* = U^*U, where L is lower triangular.
@@ -49,6 +51,9 @@ template<typename MatrixType, int UpLo> struct LLT_Traits;
   * use LDLT instead for the semidefinite case. Also, do not use a Cholesky decomposition to determine whether a system of equations
   * has a solution.
   *
+  * Example: \include LLT_example.cpp
+  * Output: \verbinclude LLT_example.out
+  *    
   * \sa MatrixBase::llt(), class LDLT
   */
  /* HEY THIS DOX IS DISABLED BECAUSE THERE's A BUG EITHER HERE OR IN LDLT ABOUT THAT (OR BOTH)
@@ -178,6 +183,9 @@ template<typename _MatrixType, int _UpLo> class LLT
     inline Index rows() const { return m_matrix.rows(); }
     inline Index cols() const { return m_matrix.cols(); }
 
+    template<typename VectorType>
+    LLT& rankUpdate(const VectorType& vec, const RealScalar& sigma = 1);
+
   protected:
     /** \internal
       * Used to compute and store L
@@ -190,16 +198,15 @@ template<typename _MatrixType, int _UpLo> class LLT
 
 namespace internal {
 
-template<int UpLo> struct llt_inplace;
+template<typename Scalar, int UpLo> struct llt_inplace;
 
-template<> struct llt_inplace<Lower>
+template<typename Scalar> struct llt_inplace<Scalar, Lower>
 {
+  typedef typename NumTraits<Scalar>::Real RealScalar;
   template<typename MatrixType>
   static typename MatrixType::Index unblocked(MatrixType& mat)
   {
     typedef typename MatrixType::Index Index;
-    typedef typename MatrixType::Scalar Scalar;
-    typedef typename MatrixType::RealScalar RealScalar;
     
     eigen_assert(mat.rows()==mat.cols());
     const Index size = mat.rows();
@@ -233,7 +240,7 @@ template<> struct llt_inplace<Lower>
 
     Index blockSize = size/8;
     blockSize = (blockSize/16)*16;
-    blockSize = std::min(std::max(blockSize,Index(8)), Index(128));
+    blockSize = (std::min)((std::max)(blockSize,Index(8)), Index(128));
 
     for (Index k=0; k<size; k+=blockSize)
     {
@@ -241,7 +248,7 @@ template<> struct llt_inplace<Lower>
       //       A00 |  -  |  -
       // lu  = A10 | A11 |  -
       //       A20 | A21 | A22
-      Index bs = std::min(blockSize, size-k);
+      Index bs = (std::min)(blockSize, size-k);
       Index rs = size - k - bs;
       Block<MatrixType,Dynamic,Dynamic> A11(m,k,   k,   bs,bs);
       Block<MatrixType,Dynamic,Dynamic> A21(m,k+bs,k,   rs,bs);
@@ -254,55 +261,133 @@ template<> struct llt_inplace<Lower>
     }
     return -1;
   }
+
+  template<typename MatrixType, typename VectorType>
+  static typename MatrixType::Index rankUpdate(MatrixType& mat, const VectorType& vec, const RealScalar& sigma)
+  {
+    typedef typename MatrixType::Index Index;
+    typedef typename MatrixType::ColXpr ColXpr;
+    typedef typename internal::remove_all<ColXpr>::type ColXprCleaned;
+    typedef typename ColXprCleaned::SegmentReturnType ColXprSegment;
+    typedef Matrix<Scalar,Dynamic,1> TempVectorType;
+    typedef typename TempVectorType::SegmentReturnType TempVecSegment;
+
+    int n = mat.cols();
+    eigen_assert(mat.rows()==n && vec.size()==n);
+
+    TempVectorType temp;
+
+    if(sigma>0)
+    {
+      // This version is based on Givens rotations.
+      // It is faster than the other one below, but only works for updates,
+      // i.e., for sigma > 0
+      temp = sqrt(sigma) * vec;
+
+      for(int i=0; i<n; ++i)
+      {
+        JacobiRotation<Scalar> g;
+        g.makeGivens(mat(i,i), -temp(i), &mat(i,i));
+
+        int rs = n-i-1;
+        if(rs>0)
+        {
+          ColXprSegment x(mat.col(i).tail(rs));
+          TempVecSegment y(temp.tail(rs));
+          apply_rotation_in_the_plane(x, y, g);
+        }
+      }
+    }
+    else
+    {
+      temp = vec;
+      RealScalar beta = 1;
+      for(int j=0; j<n; ++j)
+      {
+        RealScalar Ljj = real(mat.coeff(j,j));
+        RealScalar dj = abs2(Ljj);
+        Scalar wj = temp.coeff(j);
+        RealScalar swj2 = sigma*abs2(wj);
+        RealScalar gamma = dj*beta + swj2;
+
+        RealScalar x = dj + swj2/beta;
+        if (x<=RealScalar(0))
+          return j;
+        RealScalar nLjj = sqrt(x);
+        mat.coeffRef(j,j) = nLjj;
+        beta += swj2/dj;
+
+        // Update the terms of L
+        Index rs = n-j-1;
+        if(rs)
+        {
+          temp.tail(rs) -= (wj/Ljj) * mat.col(j).tail(rs);
+          if(gamma != 0)
+            mat.col(j).tail(rs) = (nLjj/Ljj) * mat.col(j).tail(rs) + (nLjj * sigma*conj(wj)/gamma)*temp.tail(rs);
+        }
+      }
+    }
+    return -1;
+  }
 };
 
-template<> struct llt_inplace<Upper>
+template<typename Scalar> struct llt_inplace<Scalar, Upper>
 {
+  typedef typename NumTraits<Scalar>::Real RealScalar;
+
   template<typename MatrixType>
   static EIGEN_STRONG_INLINE typename MatrixType::Index unblocked(MatrixType& mat)
   {
     Transpose<MatrixType> matt(mat);
-    return llt_inplace<Lower>::unblocked(matt);
+    return llt_inplace<Scalar, Lower>::unblocked(matt);
   }
   template<typename MatrixType>
   static EIGEN_STRONG_INLINE typename MatrixType::Index blocked(MatrixType& mat)
   {
     Transpose<MatrixType> matt(mat);
-    return llt_inplace<Lower>::blocked(matt);
+    return llt_inplace<Scalar, Lower>::blocked(matt);
+  }
+  template<typename MatrixType, typename VectorType>
+  static typename MatrixType::Index rankUpdate(MatrixType& mat, const VectorType& vec, const RealScalar& sigma)
+  {
+    Transpose<MatrixType> matt(mat);
+    return llt_inplace<Scalar, Lower>::rankUpdate(matt, vec.conjugate(), sigma);
   }
 };
 
 template<typename MatrixType> struct LLT_Traits<MatrixType,Lower>
 {
-  typedef TriangularView<MatrixType, Lower> MatrixL;
-  typedef TriangularView<typename MatrixType::AdjointReturnType, Upper> MatrixU;
-  inline static MatrixL getL(const MatrixType& m) { return m; }
-  inline static MatrixU getU(const MatrixType& m) { return m.adjoint(); }
+  typedef TriangularView<const MatrixType, Lower> MatrixL;
+  typedef TriangularView<const typename MatrixType::AdjointReturnType, Upper> MatrixU;
+  static inline MatrixL getL(const MatrixType& m) { return m; }
+  static inline MatrixU getU(const MatrixType& m) { return m.adjoint(); }
   static bool inplace_decomposition(MatrixType& m)
-  { return llt_inplace<Lower>::blocked(m)==-1; }
+  { return llt_inplace<typename MatrixType::Scalar, Lower>::blocked(m)==-1; }
 };
 
 template<typename MatrixType> struct LLT_Traits<MatrixType,Upper>
 {
-  typedef TriangularView<typename MatrixType::AdjointReturnType, Lower> MatrixL;
-  typedef TriangularView<MatrixType, Upper> MatrixU;
-  inline static MatrixL getL(const MatrixType& m) { return m.adjoint(); }
-  inline static MatrixU getU(const MatrixType& m) { return m; }
+  typedef TriangularView<const typename MatrixType::AdjointReturnType, Lower> MatrixL;
+  typedef TriangularView<const MatrixType, Upper> MatrixU;
+  static inline MatrixL getL(const MatrixType& m) { return m.adjoint(); }
+  static inline MatrixU getU(const MatrixType& m) { return m; }
   static bool inplace_decomposition(MatrixType& m)
-  { return llt_inplace<Upper>::blocked(m)==-1; }
+  { return llt_inplace<typename MatrixType::Scalar, Upper>::blocked(m)==-1; }
 };
 
 } // end namespace internal
 
 /** Computes / recomputes the Cholesky decomposition A = LL^* = U^*U of \a matrix
-  *
   *
   * \returns a reference to *this
+  *
+  * Example: \include TutorialLinAlgComputeTwice.cpp
+  * Output: \verbinclude TutorialLinAlgComputeTwice.out
   */
 template<typename MatrixType, int _UpLo>
 LLT<MatrixType,_UpLo>& LLT<MatrixType,_UpLo>::compute(const MatrixType& a)
 {
-  assert(a.rows()==a.cols());
+  eigen_assert(a.rows()==a.cols());
   const Index size = a.rows();
   m_matrix.resize(size, size);
   m_matrix = a;
@@ -314,6 +399,26 @@ LLT<MatrixType,_UpLo>& LLT<MatrixType,_UpLo>::compute(const MatrixType& a)
   return *this;
 }
 
+/** Performs a rank one update (or dowdate) of the current decomposition.
+  * If A = LL^* before the rank one update,
+  * then after it we have LL^* = A + sigma * v v^* where \a v must be a vector
+  * of same dimension.
+  */
+template<typename MatrixType, int _UpLo>
+template<typename VectorType>
+LLT<MatrixType,_UpLo>& LLT<MatrixType,_UpLo>::rankUpdate(const VectorType& v, const RealScalar& sigma)
+{
+  EIGEN_STATIC_ASSERT_VECTOR_ONLY(VectorType);
+  eigen_assert(v.size()==m_matrix.cols());
+  eigen_assert(m_isInitialized);
+  if(internal::llt_inplace<typename MatrixType::Scalar, UpLo>::rankUpdate(m_matrix,v,sigma)>=0)
+    m_info = NumericalIssue;
+  else
+    m_info = Success;
+
+  return *this;
+}
+    
 namespace internal {
 template<typename _MatrixType, int UpLo, typename Rhs>
 struct solve_retval<LLT<_MatrixType, UpLo>, Rhs>
@@ -384,3 +489,4 @@ SelfAdjointView<MatrixType, UpLo>::llt() const
 }
 
 #endif // EIGEN_LLT_H
+

Eigen/src/Cholesky/LLT_MKL.h

@@ -0,0 +1,123 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *     LLt decomposition based on LAPACKE_?potrf function.
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_LLT_MKL_H
+#define EIGEN_LLT_MKL_H
+
+#include "Eigen/src/Core/util/MKL_support.h"
+#include <iostream>
+
+namespace internal {
+
+template<typename Scalar> struct mkl_llt;
+
+#define EIGEN_MKL_LLT(EIGTYPE, MKLTYPE, MKLPREFIX) \
+template<> struct mkl_llt<EIGTYPE> \
+{ \
+  template<typename MatrixType> \
+  static inline typename MatrixType::Index potrf(MatrixType& m, char uplo) \
+  { \
+    lapack_int matrix_order; \
+    lapack_int size, lda, info, StorageOrder; \
+    EIGTYPE* a; \
+    eigen_assert(m.rows()==m.cols()); \
+/* Set up parameters for ?potrf */ \
+    size = m.rows(); \
+    StorageOrder = MatrixType::Flags&RowMajorBit?RowMajor:ColMajor; \
+    matrix_order = StorageOrder==RowMajor ? LAPACK_ROW_MAJOR : LAPACK_COL_MAJOR; \
+    a = &(m.coeffRef(0,0)); \
+    lda = m.outerStride(); \
+\
+    info = LAPACKE_##MKLPREFIX##potrf( matrix_order, uplo, size, (MKLTYPE*)a, lda ); \
+    info = (info==0) ? Success : NumericalIssue; \
+    return info; \
+  } \
+}; \
+template<> struct llt_inplace<EIGTYPE, Lower> \
+{ \
+  template<typename MatrixType> \
+  static typename MatrixType::Index blocked(MatrixType& m) \
+  { \
+    return mkl_llt<EIGTYPE>::potrf(m, 'L'); \
+  } \
+  template<typename MatrixType, typename VectorType> \
+  static void rankUpdate(MatrixType& mat, const VectorType& vec) \
+  { \
+    typedef typename MatrixType::ColXpr ColXpr; \
+    typedef typename internal::remove_all<ColXpr>::type ColXprCleaned; \
+    typedef typename ColXprCleaned::SegmentReturnType ColXprSegment; \
+    typedef typename MatrixType::Scalar Scalar; \
+    typedef Matrix<Scalar,Dynamic,1> TempVectorType; \
+    typedef typename TempVectorType::SegmentReturnType TempVecSegment; \
+\
+    int n = mat.cols(); \
+    eigen_assert(mat.rows()==n && vec.size()==n); \
+    TempVectorType temp(vec); \
+\
+    for(int i=0; i<n; ++i) \
+    { \
+      JacobiRotation<Scalar> g; \
+      g.makeGivens(mat(i,i), -temp(i), &mat(i,i)); \
+\
+      int rs = n-i-1; \
+      if(rs>0) \
+      { \
+        ColXprSegment x(mat.col(i).tail(rs)); \
+        TempVecSegment y(temp.tail(rs)); \
+        apply_rotation_in_the_plane(x, y, g); \
+      } \
+    } \
+  } \
+}; \
+template<> struct llt_inplace<EIGTYPE, Upper> \
+{ \
+  template<typename MatrixType> \
+  static typename MatrixType::Index blocked(MatrixType& m) \
+  { \
+    return mkl_llt<EIGTYPE>::potrf(m, 'U'); \
+  } \
+  template<typename MatrixType, typename VectorType> \
+  static void rankUpdate(MatrixType& mat, const VectorType& vec) \
+  { \
+    Transpose<MatrixType> matt(mat); \
+    return llt_inplace<EIGTYPE, Lower>::rankUpdate(matt, vec.conjugate()); \
+  } \
+};
+
+EIGEN_MKL_LLT(double, double, d)
+EIGEN_MKL_LLT(float, float, s)
+EIGEN_MKL_LLT(dcomplex, MKL_Complex16, z)
+EIGEN_MKL_LLT(scomplex, MKL_Complex8, c)
+
+}
+
+#endif // EIGEN_LLT_MKL_H

Eigen/src/CholmodSupport/CMakeLists.txt

@@ -0,0 +1,6 @@
+FILE(GLOB Eigen_CholmodSupport_SRCS "*.h")
+
+INSTALL(FILES 
+  ${Eigen_CholmodSupport_SRCS}
+  DESTINATION ${INCLUDE_INSTALL_DIR}/Eigen/src/CholmodSupport COMPONENT Devel
+  )

Eigen/src/CholmodSupport/CholmodSupport.h

@@ -69,11 +69,20 @@ cholmod_sparse viewAsCholmod(SparseMatrix<_Scalar,_Options,_Index>& mat)
   res.nzmax   = mat.nonZeros();
   res.nrow    = mat.rows();;
   res.ncol    = mat.cols();
-  res.p       = mat._outerIndexPtr();
-  res.i       = mat._innerIndexPtr();
-  res.x       = mat._valuePtr();
+  res.p       = mat.outerIndexPtr();
+  res.i       = mat.innerIndexPtr();
+  res.x       = mat.valuePtr();
   res.sorted  = 1;
-  res.packed  = 1;
+  if(mat.isCompressed())
+  {
+    res.packed  = 1;
+  }
+  else
+  {
+    res.packed  = 0;
+    res.nz = mat.innerNonZeroPtr();
+  }
+
   res.dtype   = 0;
   res.stype   = -1;
   
@@ -149,7 +158,9 @@ enum CholmodMode {
   CholmodAuto, CholmodSimplicialLLt, CholmodSupernodalLLt, CholmodLDLt
 };
 
-/** \brief A Cholesky factorization and solver based on Cholmod
+/** \ingroup CholmodSupport_Module
+  * \class CholmodDecomposition
+  * \brief A Cholesky factorization and solver based on Cholmod
   *
   * This class allows to solve for A.X = B sparse linear problems via a LL^T or LDL^T Cholesky factorization
   * using the Cholmod library. The sparse matrix A must be selfajoint and positive definite. The vectors or matrices
@@ -159,6 +170,9 @@ enum CholmodMode {
   * \tparam _UpLo the triangular part that will be used for the computations. It can be Lower
   *               or Upper. Default is Lower.
   *
+  * This class supports all kind of SparseMatrix<>: row or column major; upper, lower, or both; isCompressed() or unisCompressed().
+  *
+  * \sa \ref TutorialSparseDirectSolvers
   */
 template<typename _MatrixType, int _UpLo = Lower>
 class CholmodDecomposition

Eigen/src/Core/Array.h

@@ -37,6 +37,9 @@
   * API for the %Matrix class provides easy access to linear-algebra
   * operations.
   *
+  * This class can be extended with the help of the plugin mechanism described on the page
+  * \ref TopicCustomizingEigen by defining the preprocessor symbol \c EIGEN_ARRAY_PLUGIN.
+  *
   * \sa \ref TutorialArrayClass, \ref TopicClassHierarchy
   */
 namespace internal {
@@ -65,10 +68,8 @@ class Array
     friend struct internal::conservative_resize_like_impl;
 
     using Base::m_storage;
+
   public:
-    enum { NeedsToAlign = (!(Options&DontAlign))
-                          && SizeAtCompileTime!=Dynamic && ((static_cast<int>(sizeof(Scalar))*SizeAtCompileTime)%16)==0 };
-    EIGEN_MAKE_ALIGNED_OPERATOR_NEW_IF(NeedsToAlign)
 
     using Base::base;
     using Base::coeff;

Eigen/src/Core/ArrayBase.h

@@ -42,7 +42,10 @@ template<typename ExpressionType> class MatrixWrapper;
   *
   * This class is the base that is inherited by all array expression types.
   *
-  * \param Derived is the derived type, e.g., an array or an expression type.
+  * \tparam Derived is the derived type, e.g., an array or an expression type.
+  *
+  * This class can be extended with the help of the plugin mechanism described on the page
+  * \ref TopicCustomizingEigen by defining the preprocessor symbol \c EIGEN_ARRAYBASE_PLUGIN.
   *
   * \sa class MatrixBase, \ref TopicClassHierarchy
   */
@@ -156,7 +159,7 @@ template<typename Derived> class ArrayBase
     /** \returns an \link MatrixBase Matrix \endlink expression of this array
       * \sa MatrixBase::array() */
     MatrixWrapper<Derived> matrix() { return derived(); }
-    const MatrixWrapper<Derived> matrix() const { return derived(); }
+    const MatrixWrapper<const Derived> matrix() const { return derived(); }
 
 //     template<typename Dest>
 //     inline void evalTo(Dest& dst) const { dst = matrix(); }
@@ -171,10 +174,10 @@ template<typename Derived> class ArrayBase
   protected:
     // mixing arrays and matrices is not legal
     template<typename OtherDerived> Derived& operator+=(const MatrixBase<OtherDerived>& )
-    {EIGEN_STATIC_ASSERT(sizeof(typename OtherDerived::Scalar)==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES);}
+    {EIGEN_STATIC_ASSERT(std::ptrdiff_t(sizeof(typename OtherDerived::Scalar))==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES); return *this;}
     // mixing arrays and matrices is not legal
     template<typename OtherDerived> Derived& operator-=(const MatrixBase<OtherDerived>& )
-    {EIGEN_STATIC_ASSERT(sizeof(typename OtherDerived::Scalar)==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES);}
+    {EIGEN_STATIC_ASSERT(std::ptrdiff_t(sizeof(typename OtherDerived::Scalar))==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES); return *this;}
 };
 
 /** replaces \c *this by \c *this - \a other.

Eigen/src/Core/ArrayWrapper.h

@@ -53,16 +53,25 @@ class ArrayWrapper : public ArrayBase<ArrayWrapper<ExpressionType> >
     EIGEN_DENSE_PUBLIC_INTERFACE(ArrayWrapper)
     EIGEN_INHERIT_ASSIGNMENT_OPERATORS(ArrayWrapper)
 
+    typedef typename internal::conditional<
+                       internal::is_lvalue<ExpressionType>::value,
+                       Scalar,
+                       const Scalar
+                     >::type ScalarWithConstIfNotLvalue;
+
     typedef typename internal::nested<ExpressionType>::type NestedExpressionType;
 
-    inline ArrayWrapper(const ExpressionType& matrix) : m_expression(matrix) {}
+    inline ArrayWrapper(ExpressionType& matrix) : m_expression(matrix) {}
 
     inline Index rows() const { return m_expression.rows(); }
     inline Index cols() const { return m_expression.cols(); }
     inline Index outerStride() const { return m_expression.outerStride(); }
     inline Index innerStride() const { return m_expression.innerStride(); }
 
-    inline const CoeffReturnType coeff(Index row, Index col) const
+    inline ScalarWithConstIfNotLvalue* data() { return m_expression.data(); }
+    inline const Scalar* data() const { return m_expression.data(); }
+
+    inline CoeffReturnType coeff(Index row, Index col) const
     {
       return m_expression.coeff(row, col);
     }
@@ -77,7 +86,7 @@ class ArrayWrapper : public ArrayBase<ArrayWrapper<ExpressionType> >
       return m_expression.const_cast_derived().coeffRef(row, col);
     }
 
-    inline const CoeffReturnType coeff(Index index) const
+    inline CoeffReturnType coeff(Index index) const
     {
       return m_expression.coeff(index);
     }
@@ -119,8 +128,14 @@ class ArrayWrapper : public ArrayBase<ArrayWrapper<ExpressionType> >
     template<typename Dest>
     inline void evalTo(Dest& dst) const { dst = m_expression; }
 
+    const typename internal::remove_all<NestedExpressionType>::type& 
+    nestedExpression() const 
+    {
+      return m_expression;
+    }
+
   protected:
-    const NestedExpressionType m_expression;
+    NestedExpressionType m_expression;
 };
 
 /** \class MatrixWrapper
@@ -151,16 +166,25 @@ class MatrixWrapper : public MatrixBase<MatrixWrapper<ExpressionType> >
     EIGEN_DENSE_PUBLIC_INTERFACE(MatrixWrapper)
     EIGEN_INHERIT_ASSIGNMENT_OPERATORS(MatrixWrapper)
 
+    typedef typename internal::conditional<
+                       internal::is_lvalue<ExpressionType>::value,
+                       Scalar,
+                       const Scalar
+                     >::type ScalarWithConstIfNotLvalue;
+
     typedef typename internal::nested<ExpressionType>::type NestedExpressionType;
 
-    inline MatrixWrapper(const ExpressionType& matrix) : m_expression(matrix) {}
+    inline MatrixWrapper(ExpressionType& matrix) : m_expression(matrix) {}
 
     inline Index rows() const { return m_expression.rows(); }
     inline Index cols() const { return m_expression.cols(); }
     inline Index outerStride() const { return m_expression.outerStride(); }
     inline Index innerStride() const { return m_expression.innerStride(); }
 
-    inline const CoeffReturnType coeff(Index row, Index col) const
+    inline ScalarWithConstIfNotLvalue* data() { return m_expression.data(); }
+    inline const Scalar* data() const { return m_expression.data(); }
+
+    inline CoeffReturnType coeff(Index row, Index col) const
     {
       return m_expression.coeff(row, col);
     }
@@ -175,7 +199,7 @@ class MatrixWrapper : public MatrixBase<MatrixWrapper<ExpressionType> >
       return m_expression.derived().coeffRef(row, col);
     }
 
-    inline const CoeffReturnType coeff(Index index) const
+    inline CoeffReturnType coeff(Index index) const
     {
       return m_expression.coeff(index);
     }
@@ -214,8 +238,14 @@ class MatrixWrapper : public MatrixBase<MatrixWrapper<ExpressionType> >
       m_expression.const_cast_derived().template writePacket<LoadMode>(index, x);
     }
 
+    const typename internal::remove_all<NestedExpressionType>::type& 
+    nestedExpression() const 
+    {
+      return m_expression;
+    }
+
   protected:
-    const NestedExpressionType m_expression;
+    NestedExpressionType m_expression;
 };
 
 #endif // EIGEN_ARRAYWRAPPER_H

Eigen/src/Core/Assign.h

@@ -41,7 +41,7 @@ public:
     DstIsAligned = Derived::Flags & AlignedBit,
     DstHasDirectAccess = Derived::Flags & DirectAccessBit,
     SrcIsAligned = OtherDerived::Flags & AlignedBit,
-    JointAlignment = DstIsAligned && SrcIsAligned ? Aligned : Unaligned
+    JointAlignment = bool(DstIsAligned) && bool(SrcIsAligned) ? Aligned : Unaligned
   };
 
 private:
@@ -106,9 +106,9 @@ public:
                                              : int(NoUnrolling)
                   )
               : int(Traversal) == int(LinearVectorizedTraversal)
-                ? ( int(MayUnrollCompletely) && int(DstIsAligned) ? int(CompleteUnrolling) : int(NoUnrolling) )
+                ? ( bool(MayUnrollCompletely) && bool(DstIsAligned) ? int(CompleteUnrolling) : int(NoUnrolling) )
               : int(Traversal) == int(LinearTraversal)
-                ? ( int(MayUnrollCompletely) ? int(CompleteUnrolling) : int(NoUnrolling) )
+                ? ( bool(MayUnrollCompletely) ? int(CompleteUnrolling) : int(NoUnrolling) )
               : int(NoUnrolling)
   };
 
@@ -152,7 +152,7 @@ struct assign_DefaultTraversal_CompleteUnrolling
     inner = Index % Derived1::InnerSizeAtCompileTime
   };
 
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     dst.copyCoeffByOuterInner(outer, inner, src);
     assign_DefaultTraversal_CompleteUnrolling<Derived1, Derived2, Index+1, Stop>::run(dst, src);
@@ -162,13 +162,13 @@ struct assign_DefaultTraversal_CompleteUnrolling
 template<typename Derived1, typename Derived2, int Stop>
 struct assign_DefaultTraversal_CompleteUnrolling<Derived1, Derived2, Stop, Stop>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &, const Derived2 &) {}
+  static EIGEN_STRONG_INLINE void run(Derived1 &, const Derived2 &) {}
 };
 
 template<typename Derived1, typename Derived2, int Index, int Stop>
 struct assign_DefaultTraversal_InnerUnrolling
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src, int outer)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src, int outer)
   {
     dst.copyCoeffByOuterInner(outer, Index, src);
     assign_DefaultTraversal_InnerUnrolling<Derived1, Derived2, Index+1, Stop>::run(dst, src, outer);
@@ -178,7 +178,7 @@ struct assign_DefaultTraversal_InnerUnrolling
 template<typename Derived1, typename Derived2, int Stop>
 struct assign_DefaultTraversal_InnerUnrolling<Derived1, Derived2, Stop, Stop>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &, const Derived2 &, int) {}
+  static EIGEN_STRONG_INLINE void run(Derived1 &, const Derived2 &, int) {}
 };
 
 /***********************
@@ -188,7 +188,7 @@ struct assign_DefaultTraversal_InnerUnrolling<Derived1, Derived2, Stop, Stop>
 template<typename Derived1, typename Derived2, int Index, int Stop>
 struct assign_LinearTraversal_CompleteUnrolling
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     dst.copyCoeff(Index, src);
     assign_LinearTraversal_CompleteUnrolling<Derived1, Derived2, Index+1, Stop>::run(dst, src);
@@ -198,7 +198,7 @@ struct assign_LinearTraversal_CompleteUnrolling
 template<typename Derived1, typename Derived2, int Stop>
 struct assign_LinearTraversal_CompleteUnrolling<Derived1, Derived2, Stop, Stop>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &, const Derived2 &) {}
+  static EIGEN_STRONG_INLINE void run(Derived1 &, const Derived2 &) {}
 };
 
 /**************************
@@ -214,7 +214,7 @@ struct assign_innervec_CompleteUnrolling
     JointAlignment = assign_traits<Derived1,Derived2>::JointAlignment
   };
 
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     dst.template copyPacketByOuterInner<Derived2, Aligned, JointAlignment>(outer, inner, src);
     assign_innervec_CompleteUnrolling<Derived1, Derived2,
@@ -225,13 +225,13 @@ struct assign_innervec_CompleteUnrolling
 template<typename Derived1, typename Derived2, int Stop>
 struct assign_innervec_CompleteUnrolling<Derived1, Derived2, Stop, Stop>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &, const Derived2 &) {}
+  static EIGEN_STRONG_INLINE void run(Derived1 &, const Derived2 &) {}
 };
 
 template<typename Derived1, typename Derived2, int Index, int Stop>
 struct assign_innervec_InnerUnrolling
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src, int outer)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src, int outer)
   {
     dst.template copyPacketByOuterInner<Derived2, Aligned, Aligned>(outer, Index, src);
     assign_innervec_InnerUnrolling<Derived1, Derived2,
@@ -242,7 +242,7 @@ struct assign_innervec_InnerUnrolling
 template<typename Derived1, typename Derived2, int Stop>
 struct assign_innervec_InnerUnrolling<Derived1, Derived2, Stop, Stop>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &, const Derived2 &, int) {}
+  static EIGEN_STRONG_INLINE void run(Derived1 &, const Derived2 &, int) {}
 };
 
 /***************************************************************************
@@ -251,24 +251,25 @@ struct assign_innervec_InnerUnrolling<Derived1, Derived2, Stop, Stop>
 
 template<typename Derived1, typename Derived2,
          int Traversal = assign_traits<Derived1, Derived2>::Traversal,
-         int Unrolling = assign_traits<Derived1, Derived2>::Unrolling>
+         int Unrolling = assign_traits<Derived1, Derived2>::Unrolling,
+         int Version = Specialized>
 struct assign_impl;
 
 /************************
 *** Default traversal ***
 ************************/
 
-template<typename Derived1, typename Derived2, int Unrolling>
-struct assign_impl<Derived1, Derived2, InvalidTraversal, Unrolling>
+template<typename Derived1, typename Derived2, int Unrolling, int Version>
+struct assign_impl<Derived1, Derived2, InvalidTraversal, Unrolling, Version>
 {
-  inline static void run(Derived1 &, const Derived2 &) { }
+  static inline void run(Derived1 &, const Derived2 &) { }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, DefaultTraversal, NoUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, DefaultTraversal, NoUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     const Index innerSize = dst.innerSize();
     const Index outerSize = dst.outerSize();
@@ -278,21 +279,21 @@ struct assign_impl<Derived1, Derived2, DefaultTraversal, NoUnrolling>
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, DefaultTraversal, CompleteUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, DefaultTraversal, CompleteUnrolling, Version>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     assign_DefaultTraversal_CompleteUnrolling<Derived1, Derived2, 0, Derived1::SizeAtCompileTime>
       ::run(dst, src);
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, DefaultTraversal, InnerUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, DefaultTraversal, InnerUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     const Index outerSize = dst.outerSize();
     for(Index outer = 0; outer < outerSize; ++outer)
@@ -305,11 +306,11 @@ struct assign_impl<Derived1, Derived2, DefaultTraversal, InnerUnrolling>
 *** Linear traversal ***
 ***********************/
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, LinearTraversal, NoUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, LinearTraversal, NoUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     const Index size = dst.size();
     for(Index i = 0; i < size; ++i)
@@ -317,10 +318,10 @@ struct assign_impl<Derived1, Derived2, LinearTraversal, NoUnrolling>
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, LinearTraversal, CompleteUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, LinearTraversal, CompleteUnrolling, Version>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     assign_LinearTraversal_CompleteUnrolling<Derived1, Derived2, 0, Derived1::SizeAtCompileTime>
       ::run(dst, src);
@@ -331,11 +332,11 @@ struct assign_impl<Derived1, Derived2, LinearTraversal, CompleteUnrolling>
 *** Inner vectorization ***
 **************************/
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, NoUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, NoUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     const Index innerSize = dst.innerSize();
     const Index outerSize = dst.outerSize();
@@ -346,21 +347,21 @@ struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, NoUnrolling>
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, CompleteUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, CompleteUnrolling, Version>
 {
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     assign_innervec_CompleteUnrolling<Derived1, Derived2, 0, Derived1::SizeAtCompileTime>
       ::run(dst, src);
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, InnerUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, InnerVectorizedTraversal, InnerUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     const Index outerSize = dst.outerSize();
     for(Index outer = 0; outer < outerSize; ++outer)
@@ -398,11 +399,11 @@ struct unaligned_assign_impl<false>
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, NoUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, NoUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     const Index size = dst.size();
     typedef packet_traits<typename Derived1::Scalar> PacketTraits;
@@ -412,7 +413,7 @@ struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, NoUnrolling>
       srcAlignment = assign_traits<Derived1,Derived2>::JointAlignment
     };
     const Index alignedStart = assign_traits<Derived1,Derived2>::DstIsAligned ? 0
-                             : first_aligned(&dst.coeffRef(0), size);
+                             : internal::first_aligned(&dst.coeffRef(0), size);
     const Index alignedEnd = alignedStart + ((size-alignedStart)/packetSize)*packetSize;
 
     unaligned_assign_impl<assign_traits<Derived1,Derived2>::DstIsAligned!=0>::run(src,dst,0,alignedStart);
@@ -426,11 +427,11 @@ struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, NoUnrolling>
   }
 };
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, CompleteUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, CompleteUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  EIGEN_STRONG_INLINE static void run(Derived1 &dst, const Derived2 &src)
+  static EIGEN_STRONG_INLINE void run(Derived1 &dst, const Derived2 &src)
   {
     enum { size = Derived1::SizeAtCompileTime,
            packetSize = packet_traits<typename Derived1::Scalar>::size,
@@ -445,11 +446,11 @@ struct assign_impl<Derived1, Derived2, LinearVectorizedTraversal, CompleteUnroll
 *** Slice vectorization ***
 ***************************/
 
-template<typename Derived1, typename Derived2>
-struct assign_impl<Derived1, Derived2, SliceVectorizedTraversal, NoUnrolling>
+template<typename Derived1, typename Derived2, int Version>
+struct assign_impl<Derived1, Derived2, SliceVectorizedTraversal, NoUnrolling, Version>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     typedef packet_traits<typename Derived1::Scalar> PacketTraits;
     enum {
@@ -463,7 +464,7 @@ struct assign_impl<Derived1, Derived2, SliceVectorizedTraversal, NoUnrolling>
     const Index outerSize = dst.outerSize();
     const Index alignedStep = alignable ? (packetSize - dst.outerStride() % packetSize) & packetAlignedMask : 0;
     Index alignedStart = ((!alignable) || assign_traits<Derived1,Derived2>::DstIsAligned) ? 0
-                       : first_aligned(&dst.coeffRef(0,0), innerSize);
+                       : internal::first_aligned(&dst.coeffRef(0,0), innerSize);
 
     for(Index outer = 0; outer < outerSize; ++outer)
     {
@@ -474,7 +475,7 @@ struct assign_impl<Derived1, Derived2, SliceVectorizedTraversal, NoUnrolling>
 
       // do the vectorizable part of the assignment
       for(Index inner = alignedStart; inner<alignedEnd; inner+=packetSize)
-        dst.template copyPacketByOuterInner<Derived2, Aligned, Unaligned>(outer, inner, src);
+        dst.template copyPacketByOuterInner<Derived2, dstAlignment, Unaligned>(outer, inner, src);
 
       // do the non-vectorizable part of the assignment
       for(Index inner = alignedEnd; inner<innerSize ; ++inner)
@@ -531,19 +532,19 @@ struct assign_selector;
 
 template<typename Derived, typename OtherDerived>
 struct assign_selector<Derived,OtherDerived,false,false> {
-  EIGEN_STRONG_INLINE static Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.derived()); }
+  static EIGEN_STRONG_INLINE Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.derived()); }
 };
 template<typename Derived, typename OtherDerived>
 struct assign_selector<Derived,OtherDerived,true,false> {
-  EIGEN_STRONG_INLINE static Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.eval()); }
+  static EIGEN_STRONG_INLINE Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.eval()); }
 };
 template<typename Derived, typename OtherDerived>
 struct assign_selector<Derived,OtherDerived,false,true> {
-  EIGEN_STRONG_INLINE static Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.transpose()); }
+  static EIGEN_STRONG_INLINE Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.transpose()); }
 };
 template<typename Derived, typename OtherDerived>
 struct assign_selector<Derived,OtherDerived,true,true> {
-  EIGEN_STRONG_INLINE static Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.transpose().eval()); }
+  static EIGEN_STRONG_INLINE Derived& run(Derived& dst, const OtherDerived& other) { return dst.lazyAssign(other.transpose().eval()); }
 };
 
 } // end namespace internal

Eigen/src/Core/Assign_MKL.h

@@ -0,0 +1,217 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *   MKL VML support for coefficient-wise unary Eigen expressions like a=b.sin()
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_ASSIGN_VML_H
+#define EIGEN_ASSIGN_VML_H
+
+namespace internal {
+
+template<typename Op> struct vml_call
+{ enum { IsSupported = 0 }; };
+
+template<typename Dst, typename Src, typename UnaryOp>
+class vml_assign_traits
+{
+  private:
+    enum {
+      DstHasDirectAccess = Dst::Flags & DirectAccessBit,
+      SrcHasDirectAccess = Src::Flags & DirectAccessBit,
+
+      StorageOrdersAgree = (int(Dst::IsRowMajor) == int(Src::IsRowMajor)),
+      InnerSize = int(Dst::IsVectorAtCompileTime) ? int(Dst::SizeAtCompileTime)
+                : int(Dst::Flags)&RowMajorBit ? int(Dst::ColsAtCompileTime)
+                : int(Dst::RowsAtCompileTime),
+      InnerMaxSize  = int(Dst::IsVectorAtCompileTime) ? int(Dst::MaxSizeAtCompileTime)
+                    : int(Dst::Flags)&RowMajorBit ? int(Dst::MaxColsAtCompileTime)
+                    : int(Dst::MaxRowsAtCompileTime),
+      MaxSizeAtCompileTime = Dst::SizeAtCompileTime,
+
+      MightEnableVml =  vml_call<UnaryOp>::IsSupported && StorageOrdersAgree && DstHasDirectAccess && SrcHasDirectAccess
+                     && Src::InnerStrideAtCompileTime==1 && Dst::InnerStrideAtCompileTime==1,
+      MightLinearize = MightEnableVml && (int(Dst::Flags) & int(Src::Flags) & LinearAccessBit),
+      VmlSize = MightLinearize ? MaxSizeAtCompileTime : InnerMaxSize,
+      LargeEnough = VmlSize==Dynamic || VmlSize>=EIGEN_MKL_VML_THRESHOLD,
+      MayEnableVml = MightEnableVml && LargeEnough,
+      MayLinearize = MayEnableVml && MightLinearize
+    };
+  public:
+    enum {
+      Traversal = MayLinearize ? LinearVectorizedTraversal
+                : MayEnableVml ? InnerVectorizedTraversal
+                : DefaultTraversal
+    };
+};
+
+template<typename Derived1, typename Derived2, typename UnaryOp, int Traversal, int Unrolling,
+         int VmlTraversal = vml_assign_traits<Derived1, Derived2, UnaryOp>::Traversal >
+struct vml_assign_impl
+  : assign_impl<Derived1, Eigen::CwiseUnaryOp<UnaryOp, Derived2>,Traversal,Unrolling,BuiltIn>
+{
+};
+
+template<typename Derived1, typename Derived2, typename UnaryOp, int Traversal, int Unrolling>
+struct vml_assign_impl<Derived1, Derived2, UnaryOp, Traversal, Unrolling, InnerVectorizedTraversal>
+{
+  typedef typename Derived1::Scalar Scalar;
+  typedef typename Derived1::Index Index;
+  static inline void run(Derived1& dst, const CwiseUnaryOp<UnaryOp, Derived2>& src)
+  {
+    // in case we want to (or have to) skip VML at runtime we can call:
+    // assign_impl<Derived1,Eigen::CwiseUnaryOp<UnaryOp, Derived2>,Traversal,Unrolling,BuiltIn>::run(dst,src);
+    const Index innerSize = dst.innerSize();
+    const Index outerSize = dst.outerSize();
+    for(Index outer = 0; outer < outerSize; ++outer) {
+      const Scalar *src_ptr = src.IsRowMajor ?  &(src.nestedExpression().coeffRef(outer,0)) :
+                                                &(src.nestedExpression().coeffRef(0, outer));
+      Scalar *dst_ptr = dst.IsRowMajor ? &(dst.coeffRef(outer,0)) : &(dst.coeffRef(0, outer));
+      vml_call<UnaryOp>::run(src.functor(), innerSize, src_ptr, dst_ptr );
+    }
+  }
+};
+
+template<typename Derived1, typename Derived2, typename UnaryOp, int Traversal, int Unrolling>
+struct vml_assign_impl<Derived1, Derived2, UnaryOp, Traversal, Unrolling, LinearVectorizedTraversal>
+{
+  static inline void run(Derived1& dst, const CwiseUnaryOp<UnaryOp, Derived2>& src)
+  {
+    // in case we want to (or have to) skip VML at runtime we can call:
+    // assign_impl<Derived1,Eigen::CwiseUnaryOp<UnaryOp, Derived2>,Traversal,Unrolling,BuiltIn>::run(dst,src);
+    vml_call<UnaryOp>::run(src.functor(), dst.size(), src.nestedExpression().data(), dst.data() );
+  }
+};
+
+// Macroses
+
+#define EIGEN_MKL_VML_SPECIALIZE_ASSIGN(TRAVERSAL,UNROLLING) \
+  template<typename Derived1, typename Derived2, typename UnaryOp> \
+  struct assign_impl<Derived1, Eigen::CwiseUnaryOp<UnaryOp, Derived2>, TRAVERSAL, UNROLLING, Specialized>  {  \
+    static inline void run(Derived1 &dst, const Eigen::CwiseUnaryOp<UnaryOp, Derived2> &src) { \
+      vml_assign_impl<Derived1,Derived2,UnaryOp,TRAVERSAL,UNROLLING>::run(dst, src); \
+    } \
+  };
+
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(DefaultTraversal,NoUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(DefaultTraversal,CompleteUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(DefaultTraversal,InnerUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(LinearTraversal,NoUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(LinearTraversal,CompleteUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(InnerVectorizedTraversal,NoUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(InnerVectorizedTraversal,CompleteUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(InnerVectorizedTraversal,InnerUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(LinearVectorizedTraversal,CompleteUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(LinearVectorizedTraversal,NoUnrolling)
+EIGEN_MKL_VML_SPECIALIZE_ASSIGN(SliceVectorizedTraversal,NoUnrolling)
+
+
+#if !defined (EIGEN_FAST_MATH) || (EIGEN_FAST_MATH != 1)
+#define  EIGEN_MKL_VML_MODE VML_HA
+#else
+#define  EIGEN_MKL_VML_MODE VML_LA
+#endif
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALL(EIGENOP, VMLOP, EIGENTYPE, VMLTYPE)     \
+  template<> struct vml_call< scalar_##EIGENOP##_op<EIGENTYPE> > {               \
+    enum { IsSupported = 1 };                                                    \
+    static inline void run( const scalar_##EIGENOP##_op<EIGENTYPE>& func,        \
+                            int size, const EIGENTYPE* src, EIGENTYPE* dst) {    \
+      VMLOP(size, (const VMLTYPE*)src, (VMLTYPE*)dst);                           \
+    }                                                                            \
+  };
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALL_LA(EIGENOP, VMLOP, EIGENTYPE, VMLTYPE)  \
+  template<> struct vml_call< scalar_##EIGENOP##_op<EIGENTYPE> > {               \
+    enum { IsSupported = 1 };                                                    \
+    static inline void run( const scalar_##EIGENOP##_op<EIGENTYPE>& func,        \
+                            int size, const EIGENTYPE* src, EIGENTYPE* dst) {    \
+      MKL_INT64 vmlMode = EIGEN_MKL_VML_MODE;                                    \
+      VMLOP(size, (const VMLTYPE*)src, (VMLTYPE*)dst, vmlMode);                  \
+    }                                                                            \
+  };
+
+#define EIGEN_MKL_VML_DECLARE_POW_CALL(EIGENOP, VMLOP, EIGENTYPE, VMLTYPE)       \
+  template<> struct vml_call< scalar_##EIGENOP##_op<EIGENTYPE> > {               \
+    enum { IsSupported = 1 };                                                    \
+    static inline void run( const scalar_##EIGENOP##_op<EIGENTYPE>& func,        \
+                          int size, const EIGENTYPE* src, EIGENTYPE* dst) {      \
+      EIGENTYPE exponent = func.m_exponent;                                      \
+      MKL_INT64 vmlMode = EIGEN_MKL_VML_MODE;                                    \
+      VMLOP(&size, (const VMLTYPE*)src, (const VMLTYPE*)&exponent,               \
+                        (VMLTYPE*)dst, &vmlMode);                                \
+    }                                                                            \
+  };
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALLS_REAL(EIGENOP, VMLOP)                   \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL(EIGENOP, vs##VMLOP, float, float)             \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL(EIGENOP, vd##VMLOP, double, double)
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALLS_COMPLEX(EIGENOP, VMLOP)                \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL(EIGENOP, vc##VMLOP, scomplex, MKL_Complex8)   \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL(EIGENOP, vz##VMLOP, dcomplex, MKL_Complex16)
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALLS(EIGENOP, VMLOP)                        \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALLS_REAL(EIGENOP, VMLOP)                         \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALLS_COMPLEX(EIGENOP, VMLOP)
+
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALLS_REAL_LA(EIGENOP, VMLOP)                \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL_LA(EIGENOP, vms##VMLOP, float, float)         \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL_LA(EIGENOP, vmd##VMLOP, double, double)
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALLS_COMPLEX_LA(EIGENOP, VMLOP)             \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL_LA(EIGENOP, vmc##VMLOP, scomplex, MKL_Complex8)  \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALL_LA(EIGENOP, vmz##VMLOP, dcomplex, MKL_Complex16)
+
+#define EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(EIGENOP, VMLOP)                     \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALLS_REAL_LA(EIGENOP, VMLOP)                      \
+  EIGEN_MKL_VML_DECLARE_UNARY_CALLS_COMPLEX_LA(EIGENOP, VMLOP)
+
+
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(sin,  Sin)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(asin, Asin)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(cos,  Cos)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(acos, Acos)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(tan,  Tan)
+//EIGEN_MKL_VML_DECLARE_UNARY_CALLS(abs,  Abs)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(exp,  Exp)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(log,  Ln)
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_LA(sqrt, Sqrt)
+
+EIGEN_MKL_VML_DECLARE_UNARY_CALLS_REAL(square, Sqr)
+
+EIGEN_MKL_VML_DECLARE_POW_CALL(pow, vmspowx_, float, float)
+EIGEN_MKL_VML_DECLARE_POW_CALL(pow, vmdpowx_, double, double)
+EIGEN_MKL_VML_DECLARE_POW_CALL(pow, vmcpowx_, scomplex, MKL_Complex8)
+EIGEN_MKL_VML_DECLARE_POW_CALL(pow, vmzpowx_, dcomplex, MKL_Complex16)
+
+} // end namespace internal
+
+#endif // EIGEN_ASSIGN_VML_H

Eigen/src/Core/BandMatrix.h

@@ -87,7 +87,7 @@ class BandMatrixBase : public EigenBase<Derived>
       if (i<=supers())
       {
         start = supers()-i;
-        len = std::min(rows(),std::max<Index>(0,coeffs().rows() - (supers()-i)));
+        len = (std::min)(rows(),std::max<Index>(0,coeffs().rows() - (supers()-i)));
       }
       else if (i>=rows()-subs())
         len = std::max<Index>(0,coeffs().rows() - (i + 1 - rows() + subs()));
@@ -96,11 +96,11 @@ class BandMatrixBase : public EigenBase<Derived>
 
     /** \returns a vector expression of the main diagonal */
     inline Block<CoefficientsType,1,SizeAtCompileTime> diagonal()
-    { return Block<CoefficientsType,1,SizeAtCompileTime>(coeffs(),supers(),0,1,std::min(rows(),cols())); }
+    { return Block<CoefficientsType,1,SizeAtCompileTime>(coeffs(),supers(),0,1,(std::min)(rows(),cols())); }
 
     /** \returns a vector expression of the main diagonal (const version) */
     inline const Block<const CoefficientsType,1,SizeAtCompileTime> diagonal() const
-    { return Block<const CoefficientsType,1,SizeAtCompileTime>(coeffs(),supers(),0,1,std::min(rows(),cols())); }
+    { return Block<const CoefficientsType,1,SizeAtCompileTime>(coeffs(),supers(),0,1,(std::min)(rows(),cols())); }
 
     template<int Index> struct DiagonalIntReturnType {
       enum {
@@ -122,13 +122,13 @@ class BandMatrixBase : public EigenBase<Derived>
     /** \returns a vector expression of the \a N -th sub or super diagonal */
     template<int N> inline typename DiagonalIntReturnType<N>::Type diagonal()
     {
-      return typename DiagonalIntReturnType<N>::BuildType(coeffs(), supers()-N, std::max(0,N), 1, diagonalLength(N));
+      return typename DiagonalIntReturnType<N>::BuildType(coeffs(), supers()-N, (std::max)(0,N), 1, diagonalLength(N));
     }
 
     /** \returns a vector expression of the \a N -th sub or super diagonal */
     template<int N> inline const typename DiagonalIntReturnType<N>::Type diagonal() const
     {
-      return typename DiagonalIntReturnType<N>::BuildType(coeffs(), supers()-N, std::max(0,N), 1, diagonalLength(N));
+      return typename DiagonalIntReturnType<N>::BuildType(coeffs(), supers()-N, (std::max)(0,N), 1, diagonalLength(N));
     }
 
     /** \returns a vector expression of the \a i -th sub or super diagonal */
@@ -166,7 +166,7 @@ class BandMatrixBase : public EigenBase<Derived>
   protected:
 
     inline Index diagonalLength(Index i) const
-    { return i<0 ? std::min(cols(),rows()+i) : std::min(rows(),cols()-i); }
+    { return i<0 ? (std::min)(cols(),rows()+i) : (std::min)(rows(),cols()-i); }
 };
 
 /**
@@ -180,7 +180,7 @@ class BandMatrixBase : public EigenBase<Derived>
   * \param Cols Number of columns, or \b Dynamic
   * \param Supers Number of super diagonal
   * \param Subs Number of sub diagonal
-  * \param _Options A combination of either \b RowMajor or \b ColMajor, and of \b SelfAdjoint
+  * \param _Options A combination of either \b #RowMajor or \b #ColMajor, and of \b #SelfAdjoint
   *                 The former controls \ref TopicStorageOrders "storage order", and defaults to
   *                 column-major. The latter controls whether the matrix represents a selfadjoint 
   *                 matrix in which case either Supers of Subs have to be null.
@@ -284,6 +284,7 @@ class BandMatrixWrapper : public BandMatrixBase<BandMatrixWrapper<_CoefficientsT
       : m_coeffs(coeffs),
         m_rows(rows), m_supers(supers), m_subs(subs)
     {
+      EIGEN_UNUSED_VARIABLE(cols);
       //internal::assert(coeffs.cols()==cols() && (supers()+subs()+1)==coeffs.rows());
     }
 

Eigen/src/Core/Block.h

@@ -94,7 +94,7 @@ struct traits<Block<XprType, BlockRows, BlockCols, InnerPanel, HasDirectAccess>
     MaskPacketAccessBit = (InnerSize == Dynamic || (InnerSize % packet_traits<Scalar>::size) == 0)
                        && (InnerStrideAtCompileTime == 1)
                         ? PacketAccessBit : 0,
-    MaskAlignedBit = (InnerPanel && (OuterStrideAtCompileTime!=Dynamic) && ((OuterStrideAtCompileTime % packet_traits<Scalar>::size) == 0)) ? AlignedBit : 0,
+    MaskAlignedBit = (InnerPanel && (OuterStrideAtCompileTime!=Dynamic) && (((OuterStrideAtCompileTime * int(sizeof(Scalar))) % 16) == 0)) ? AlignedBit : 0,
     FlagsLinearAccessBit = (RowsAtCompileTime == 1 || ColsAtCompileTime == 1) ? LinearAccessBit : 0,
     FlagsLvalueBit = is_lvalue<XprType>::value ? LvalueBit : 0,
     FlagsRowMajorBit = IsRowMajor ? RowMajorBit : 0,
@@ -242,6 +242,21 @@ template<typename XprType, int BlockRows, int BlockCols, bool InnerPanel, bool H
     inline Index outerStride() const;
     #endif
 
+    const typename internal::remove_all<typename XprType::Nested>::type& nestedExpression() const 
+    { 
+      return m_xpr; 
+    }
+      
+    Index startRow() const 
+    { 
+      return m_startRow.value(); 
+    }
+      
+    Index startCol() const 
+    { 
+      return m_startCol.value(); 
+    }
+
   protected:
 
     const typename XprType::Nested m_xpr;
@@ -304,6 +319,11 @@ class Block<XprType,BlockRows,BlockCols, InnerPanel,true>
       init();
     }
 
+    const typename internal::remove_all<typename XprType::Nested>::type& nestedExpression() const 
+    { 
+      return m_xpr; 
+    }
+      
     /** \sa MapBase::innerStride() */
     inline Index innerStride() const
     {
@@ -341,7 +361,7 @@ class Block<XprType,BlockRows,BlockCols, InnerPanel,true>
                     : m_xpr.innerStride();
     }
 
-    const typename XprType::Nested m_xpr;
+    typename XprType::Nested m_xpr;
     int m_outerStride;
 };
 

Eigen/src/Core/BooleanRedux.h

@@ -35,7 +35,7 @@ struct all_unroller
     row = (UnrollCount-1) % Derived::RowsAtCompileTime
   };
 
-  inline static bool run(const Derived &mat)
+  static inline bool run(const Derived &mat)
   {
     return all_unroller<Derived, UnrollCount-1>::run(mat) && mat.coeff(row, col);
   }
@@ -44,13 +44,13 @@ struct all_unroller
 template<typename Derived>
 struct all_unroller<Derived, 1>
 {
-  inline static bool run(const Derived &mat) { return mat.coeff(0, 0); }
+  static inline bool run(const Derived &mat) { return mat.coeff(0, 0); }
 };
 
 template<typename Derived>
 struct all_unroller<Derived, Dynamic>
 {
-  inline static bool run(const Derived &) { return false; }
+  static inline bool run(const Derived &) { return false; }
 };
 
 template<typename Derived, int UnrollCount>
@@ -61,7 +61,7 @@ struct any_unroller
     row = (UnrollCount-1) % Derived::RowsAtCompileTime
   };
 
-  inline static bool run(const Derived &mat)
+  static inline bool run(const Derived &mat)
   {
     return any_unroller<Derived, UnrollCount-1>::run(mat) || mat.coeff(row, col);
   }
@@ -70,13 +70,13 @@ struct any_unroller
 template<typename Derived>
 struct any_unroller<Derived, 1>
 {
-  inline static bool run(const Derived &mat) { return mat.coeff(0, 0); }
+  static inline bool run(const Derived &mat) { return mat.coeff(0, 0); }
 };
 
 template<typename Derived>
 struct any_unroller<Derived, Dynamic>
 {
-  inline static bool run(const Derived &) { return false; }
+  static inline bool run(const Derived &) { return false; }
 };
 
 } // end namespace internal

Eigen/src/Core/CwiseBinaryOp.h

@@ -167,8 +167,8 @@ class CwiseBinaryOp : internal::no_assignment_operator,
     const BinaryOp& functor() const { return m_functor; }
 
   protected:
-    const LhsNested m_lhs;
-    const RhsNested m_rhs;
+    LhsNested m_lhs;
+    RhsNested m_rhs;
     const BinaryOp m_functor;
 };
 

Eigen/src/Core/CwiseNullaryOp.h

@@ -101,6 +101,9 @@ class CwiseNullaryOp : internal::no_assignment_operator,
       return m_functor.packetOp(index);
     }
 
+    /** \returns the functor representing the nullary operation */
+    const NullaryOp& functor() const { return m_functor; }
+
   protected:
     const internal::variable_if_dynamic<Index, RowsAtCompileTime> m_rows;
     const internal::variable_if_dynamic<Index, ColsAtCompileTime> m_cols;
@@ -742,7 +745,7 @@ struct setIdentity_impl<Derived, true>
   static EIGEN_STRONG_INLINE Derived& run(Derived& m)
   {
     m.setZero();
-    const Index size = std::min(m.rows(), m.cols());
+    const Index size = (std::min)(m.rows(), m.cols());
     for(Index i = 0; i < size; ++i) m.coeffRef(i,i) = typename Derived::Scalar(1);
     return m;
   }

Eigen/src/Core/CwiseUnaryOp.h

@@ -95,7 +95,7 @@ class CwiseUnaryOp : internal::no_assignment_operator,
     nestedExpression() { return m_xpr.const_cast_derived(); }
 
   protected:
-    const typename XprType::Nested m_xpr;
+    typename XprType::Nested m_xpr;
     const UnaryOp m_functor;
 };
 

Eigen/src/Core/CwiseUnaryView.h

@@ -97,7 +97,7 @@ class CwiseUnaryView : internal::no_assignment_operator,
 
   protected:
     // FIXME changed from MatrixType::Nested because of a weird compilation error with sun CC
-    const typename internal::nested<MatrixType>::type m_matrix;
+    typename internal::nested<MatrixType>::type m_matrix;
     ViewOp m_functor;
 };
 

Eigen/src/Core/DenseBase.h

@@ -34,7 +34,10 @@
   * This class is the base that is inherited by all dense objects (matrix, vector, arrays,
   * and related expression types). The common Eigen API for dense objects is contained in this class.
   *
-  * \param Derived is the derived type, e.g., a matrix type or an expression.
+  * \tparam Derived is the derived type, e.g., a matrix type or an expression.
+  *
+  * This class can be extended with the help of the plugin mechanism described on the page
+  * \ref TopicCustomizingEigen by defining the preprocessor symbol \c EIGEN_DENSEBASE_PLUGIN.
   *
   * \sa \ref TopicClassHierarchy
   */
@@ -53,7 +56,13 @@ template<typename Derived> class DenseBase
     class InnerIterator;
 
     typedef typename internal::traits<Derived>::StorageKind StorageKind;
-    typedef typename internal::traits<Derived>::Index Index; /**< The type of indices */
+
+    /** \brief The type of indices 
+      * \details To change this, \c \#define the preprocessor symbol \c EIGEN_DEFAULT_DENSE_INDEX_TYPE.
+      * \sa \ref TopicPreprocessorDirectives.
+      */
+    typedef typename internal::traits<Derived>::Index Index; 
+
     typedef typename internal::traits<Derived>::Scalar Scalar;
     typedef typename internal::packet_traits<Scalar>::type PacketScalar;
     typedef typename NumTraits<Scalar>::Real RealScalar;
@@ -367,12 +376,13 @@ template<typename Derived> class DenseBase
     inline Derived& operator*=(const Scalar& other);
     inline Derived& operator/=(const Scalar& other);
 
+    typedef typename internal::add_const_on_value_type<typename internal::eval<Derived>::type>::type EvalReturnType;
     /** \returns the matrix or vector obtained by evaluating this expression.
       *
       * Notice that in the case of a plain matrix or vector (not an expression) this function just returns
       * a const reference, in order to avoid a useless copy.
       */
-    EIGEN_STRONG_INLINE const typename internal::eval<Derived>::type eval() const
+    EIGEN_STRONG_INLINE EvalReturnType eval() const
     {
       // Even though MSVC does not honor strong inlining when the return type
       // is a dynamic matrix, we desperately need strong inlining for fixed

Eigen/src/Core/DenseCoeffsBase.h

@@ -35,7 +35,7 @@ template<typename T> struct add_const_on_value_type_if_arithmetic
 /** \brief Base class providing read-only coefficient access to matrices and arrays.
   * \ingroup Core_Module
   * \tparam Derived Type of the derived class
-  * \tparam ReadOnlyAccessors Constant indicating read-only access
+  * \tparam #ReadOnlyAccessors Constant indicating read-only access
   *
   * This class defines the \c operator() \c const function and friends, which can be used to read specific
   * entries of a matrix or array.
@@ -212,7 +212,7 @@ class DenseCoeffsBase<Derived,ReadOnlyAccessors> : public EigenBase<Derived>
       * to ensure that a packet really starts there. This method is only available on expressions having the
       * PacketAccessBit.
       *
-      * The \a LoadMode parameter may have the value \a Aligned or \a Unaligned. Its effect is to select
+      * The \a LoadMode parameter may have the value \a #Aligned or \a #Unaligned. Its effect is to select
       * the appropriate vectorization instruction. Aligned access is faster, but is only possible for packets
       * starting at an address which is a multiple of the packet size.
       */
@@ -239,7 +239,7 @@ class DenseCoeffsBase<Derived,ReadOnlyAccessors> : public EigenBase<Derived>
       * to ensure that a packet really starts there. This method is only available on expressions having the
       * PacketAccessBit and the LinearAccessBit.
       *
-      * The \a LoadMode parameter may have the value \a Aligned or \a Unaligned. Its effect is to select
+      * The \a LoadMode parameter may have the value \a #Aligned or \a #Unaligned. Its effect is to select
       * the appropriate vectorization instruction. Aligned access is faster, but is only possible for packets
       * starting at an address which is a multiple of the packet size.
       */
@@ -275,7 +275,7 @@ class DenseCoeffsBase<Derived,ReadOnlyAccessors> : public EigenBase<Derived>
 /** \brief Base class providing read/write coefficient access to matrices and arrays.
   * \ingroup Core_Module
   * \tparam Derived Type of the derived class
-  * \tparam WriteAccessors Constant indicating read/write access
+  * \tparam #WriteAccessors Constant indicating read/write access
   *
   * This class defines the non-const \c operator() function and friends, which can be used to write specific
   * entries of a matrix or array. This class inherits DenseCoeffsBase<Derived, ReadOnlyAccessors> which
@@ -433,7 +433,7 @@ class DenseCoeffsBase<Derived, WriteAccessors> : public DenseCoeffsBase<Derived,
       * to ensure that a packet really starts there. This method is only available on expressions having the
       * PacketAccessBit.
       *
-      * The \a LoadMode parameter may have the value \a Aligned or \a Unaligned. Its effect is to select
+      * The \a LoadMode parameter may have the value \a #Aligned or \a #Unaligned. Its effect is to select
       * the appropriate vectorization instruction. Aligned access is faster, but is only possible for packets
       * starting at an address which is a multiple of the packet size.
       */
@@ -567,7 +567,7 @@ class DenseCoeffsBase<Derived, WriteAccessors> : public DenseCoeffsBase<Derived,
 /** \brief Base class providing direct read-only coefficient access to matrices and arrays.
   * \ingroup Core_Module
   * \tparam Derived Type of the derived class
-  * \tparam DirectAccessors Constant indicating direct access
+  * \tparam #DirectAccessors Constant indicating direct access
   *
   * This class defines functions to work with strides which can be used to access entries directly. This class
   * inherits DenseCoeffsBase<Derived, ReadOnlyAccessors> which defines functions to access entries read-only using
@@ -637,7 +637,7 @@ class DenseCoeffsBase<Derived, DirectAccessors> : public DenseCoeffsBase<Derived
 /** \brief Base class providing direct read/write coefficient access to matrices and arrays.
   * \ingroup Core_Module
   * \tparam Derived Type of the derived class
-  * \tparam DirectAccessors Constant indicating direct access
+  * \tparam #DirectWriteAccessors Constant indicating direct access
   *
   * This class defines functions to work with strides which can be used to access entries directly. This class
   * inherits DenseCoeffsBase<Derived, WriteAccessors> which defines functions to access entries read/write using
@@ -710,16 +710,16 @@ namespace internal {
 template<typename Derived, bool JustReturnZero>
 struct first_aligned_impl
 {
-  inline static typename Derived::Index run(const Derived&)
+  static inline typename Derived::Index run(const Derived&)
   { return 0; }
 };
 
 template<typename Derived>
 struct first_aligned_impl<Derived, false>
 {
-  inline static typename Derived::Index run(const Derived& m)
+  static inline typename Derived::Index run(const Derived& m)
   {
-    return first_aligned(&m.const_cast_derived().coeffRef(0,0), m.size());
+    return internal::first_aligned(&m.const_cast_derived().coeffRef(0,0), m.size());
   }
 };
 
@@ -729,7 +729,7 @@ struct first_aligned_impl<Derived, false>
   * documentation.
   */
 template<typename Derived>
-inline static typename Derived::Index first_aligned(const Derived& m)
+static inline typename Derived::Index first_aligned(const Derived& m)
 {
   return first_aligned_impl
           <Derived, (Derived::Flags & AlignedBit) || !(Derived::Flags & DirectAccessBit)>

Eigen/src/Core/DenseStorage.h

@@ -58,14 +58,14 @@ struct plain_array
   #define EIGEN_MAKE_UNALIGNED_ARRAY_ASSERT(sizemask) \
     eigen_assert((reinterpret_cast<size_t>(array) & sizemask) == 0 \
               && "this assertion is explained here: " \
-              "http://eigen.tuxfamily.org/dox/UnalignedArrayAssert.html" \
+              "http://eigen.tuxfamily.org/dox-devel/TopicUnalignedArrayAssert.html" \
               " **** READ THIS WEB PAGE !!! ****");
 #endif
 
 template <typename T, int Size, int MatrixOrArrayOptions>
 struct plain_array<T, Size, MatrixOrArrayOptions, 16>
 {
-  EIGEN_ALIGN16 T array[Size];
+  EIGEN_USER_ALIGN16 T array[Size];
   plain_array() { EIGEN_MAKE_UNALIGNED_ARRAY_ASSERT(0xf) }
   plain_array(constructor_without_unaligned_array_assert) {}
 };
@@ -73,7 +73,7 @@ struct plain_array<T, Size, MatrixOrArrayOptions, 16>
 template <typename T, int MatrixOrArrayOptions, int Alignment>
 struct plain_array<T, 0, MatrixOrArrayOptions, Alignment>
 {
-  EIGEN_ALIGN16 T array[1];
+  EIGEN_USER_ALIGN16 T array[1];
   plain_array() {}
   plain_array(constructor_without_unaligned_array_assert) {}
 };
@@ -104,8 +104,8 @@ template<typename T, int Size, int _Rows, int _Cols, int _Options> class DenseSt
       : m_data(internal::constructor_without_unaligned_array_assert()) {}
     inline DenseStorage(DenseIndex,DenseIndex,DenseIndex) {}
     inline void swap(DenseStorage& other) { std::swap(m_data,other.m_data); }
-    inline static DenseIndex rows(void) {return _Rows;}
-    inline static DenseIndex cols(void) {return _Cols;}
+    static inline DenseIndex rows(void) {return _Rows;}
+    static inline DenseIndex cols(void) {return _Cols;}
     inline void conservativeResize(DenseIndex,DenseIndex,DenseIndex) {}
     inline void resize(DenseIndex,DenseIndex,DenseIndex) {}
     inline const T *data() const { return m_data.array; }
@@ -120,14 +120,24 @@ template<typename T, int _Rows, int _Cols, int _Options> class DenseStorage<T, 0
     inline DenseStorage(internal::constructor_without_unaligned_array_assert) {}
     inline DenseStorage(DenseIndex,DenseIndex,DenseIndex) {}
     inline void swap(DenseStorage& ) {}
-    inline static DenseIndex rows(void) {return _Rows;}
-    inline static DenseIndex cols(void) {return _Cols;}
+    static inline DenseIndex rows(void) {return _Rows;}
+    static inline DenseIndex cols(void) {return _Cols;}
     inline void conservativeResize(DenseIndex,DenseIndex,DenseIndex) {}
     inline void resize(DenseIndex,DenseIndex,DenseIndex) {}
     inline const T *data() const { return 0; }
     inline T *data() { return 0; }
 };
 
+// more specializations for null matrices; these are necessary to resolve ambiguities
+template<typename T, int _Options> class DenseStorage<T, 0, Dynamic, Dynamic, _Options>
+: public DenseStorage<T, 0, 0, 0, _Options> { };
+
+template<typename T, int _Rows, int _Options> class DenseStorage<T, 0, _Rows, Dynamic, _Options>
+: public DenseStorage<T, 0, 0, 0, _Options> { };
+
+template<typename T, int _Cols, int _Options> class DenseStorage<T, 0, Dynamic, _Cols, _Options>
+: public DenseStorage<T, 0, 0, 0, _Options> { };
+
 // dynamic-size matrix with fixed-size storage
 template<typename T, int Size, int _Options> class DenseStorage<T, Size, Dynamic, Dynamic, _Options>
 {
@@ -241,7 +251,7 @@ template<typename T, int _Rows, int _Options> class DenseStorage<T, Dynamic, _Ro
     { EIGEN_INTERNAL_DENSE_STORAGE_CTOR_PLUGIN }
     inline ~DenseStorage() { internal::conditional_aligned_delete_auto<T,(_Options&DontAlign)==0>(m_data, _Rows*m_cols); }
     inline void swap(DenseStorage& other) { std::swap(m_data,other.m_data); std::swap(m_cols,other.m_cols); }
-    inline static DenseIndex rows(void) {return _Rows;}
+    static inline DenseIndex rows(void) {return _Rows;}
     inline DenseIndex cols(void) const {return m_cols;}
     inline void conservativeResize(DenseIndex size, DenseIndex, DenseIndex cols)
     {
@@ -278,7 +288,7 @@ template<typename T, int _Cols, int _Options> class DenseStorage<T, Dynamic, Dyn
     inline ~DenseStorage() { internal::conditional_aligned_delete_auto<T,(_Options&DontAlign)==0>(m_data, _Cols*m_rows); }
     inline void swap(DenseStorage& other) { std::swap(m_data,other.m_data); std::swap(m_rows,other.m_rows); }
     inline DenseIndex rows(void) const {return m_rows;}
-    inline static DenseIndex cols(void) {return _Cols;}
+    static inline DenseIndex cols(void) {return _Cols;}
     inline void conservativeResize(DenseIndex size, DenseIndex rows, DenseIndex)
     {
       m_data = internal::conditional_aligned_realloc_new_auto<T,(_Options&DontAlign)==0>(m_data, size, m_rows*_Cols);

Eigen/src/Core/Diagonal.h

@@ -2,6 +2,7 @@
 // for linear algebra.
 //
 // Copyright (C) 2007-2009 Benoit Jacob <jacob.benoit.1@gmail.com>
+// Copyright (C) 2009-2010 Gael Guennebaud <gael.guennebaud@inria.fr>
 //
 // Eigen is free software; you can redistribute it and/or
 // modify it under the terms of the GNU Lesser General Public
@@ -87,7 +88,7 @@ template<typename MatrixType, int DiagIndex> class Diagonal
     EIGEN_INHERIT_ASSIGNMENT_OPERATORS(Diagonal)
 
     inline Index rows() const
-    { return m_index.value()<0 ? std::min(m_matrix.cols(),m_matrix.rows()+m_index.value()) : std::min(m_matrix.rows(),m_matrix.cols()-m_index.value()); }
+    { return m_index.value()<0 ? (std::min)(m_matrix.cols(),m_matrix.rows()+m_index.value()) : (std::min)(m_matrix.rows(),m_matrix.cols()-m_index.value()); }
 
     inline Index cols() const { return 1; }
 
@@ -101,6 +102,15 @@ template<typename MatrixType, int DiagIndex> class Diagonal
       return 0;
     }
 
+    typedef typename internal::conditional<
+                       internal::is_lvalue<MatrixType>::value,
+                       Scalar,
+                       const Scalar
+                     >::type ScalarWithConstIfNotLvalue;
+
+    inline ScalarWithConstIfNotLvalue* data() { return &(m_matrix.const_cast_derived().coeffRef(rowOffset(), colOffset())); }
+    inline const Scalar* data() const { return &(m_matrix.const_cast_derived().coeffRef(rowOffset(), colOffset())); }
+
     inline Scalar& coeffRef(Index row, Index)
     {
       EIGEN_STATIC_ASSERT_LVALUE(MatrixType)
@@ -133,8 +143,19 @@ template<typename MatrixType, int DiagIndex> class Diagonal
       return m_matrix.coeff(index+rowOffset(), index+colOffset());
     }
 
+    const typename internal::remove_all<typename MatrixType::Nested>::type& 
+    nestedExpression() const 
+    {
+      return m_matrix;
+    }
+
+    int index() const
+    {
+      return m_index.value();
+    }
+
   protected:
-    const typename MatrixType::Nested m_matrix;
+    typename MatrixType::Nested m_matrix;
     const internal::variable_if_dynamic<Index, DiagIndex> m_index;
 
   private:

Eigen/src/Core/DiagonalMatrix.h

@@ -72,7 +72,7 @@ class DiagonalBase : public EigenBase<Derived>
     const DiagonalProduct<MatrixDerived, Derived, OnTheLeft>
     operator*(const MatrixBase<MatrixDerived> &matrix) const;
 
-    inline const DiagonalWrapper<CwiseUnaryOp<internal::scalar_inverse_op<Scalar>, const DiagonalVectorType> >
+    inline const DiagonalWrapper<const CwiseUnaryOp<internal::scalar_inverse_op<Scalar>, const DiagonalVectorType> >
     inverse() const
     {
       return diagonal().cwiseInverse();
@@ -251,13 +251,13 @@ class DiagonalWrapper
     #endif
 
     /** Constructor from expression of diagonal coefficients to wrap. */
-    inline DiagonalWrapper(const DiagonalVectorType& diagonal) : m_diagonal(diagonal) {}
+    inline DiagonalWrapper(DiagonalVectorType& diagonal) : m_diagonal(diagonal) {}
 
     /** \returns a const reference to the wrapped expression of diagonal coefficients. */
     const DiagonalVectorType& diagonal() const { return m_diagonal; }
 
   protected:
-    const typename DiagonalVectorType::Nested m_diagonal;
+    typename DiagonalVectorType::Nested m_diagonal;
 };
 
 /** \returns a pseudo-expression of a diagonal matrix with *this as vector of diagonal coefficients

Eigen/src/Core/DiagonalProduct.h

@@ -107,8 +107,8 @@ class DiagonalProduct : internal::no_assignment_operator,
                      m_diagonal.diagonal().template packet<DiagonalVectorPacketLoadMode>(id));
     }
 
-    const typename MatrixType::Nested m_matrix;
-    const typename DiagonalType::Nested m_diagonal;
+    typename MatrixType::Nested m_matrix;
+    typename DiagonalType::Nested m_diagonal;
 };
 
 /** \returns the diagonal matrix product of \c *this by the diagonal matrix \a diagonal.

Eigen/src/Core/Dot.h

@@ -116,7 +116,9 @@ MatrixBase<Derived>::eigen2_dot(const MatrixBase<OtherDerived>& other) const
 
 //---------- implementation of L2 norm and related functions ----------
 
-/** \returns the squared \em l2 norm of *this, i.e., for vectors, the dot product of *this with itself.
+/** \returns, for vectors, the squared \em l2 norm of \c *this, and for matrices the Frobenius norm.
+  * In both cases, it consists in the sum of the square of all the matrix entries.
+  * For vectors, this is also equals to the dot product of \c *this with itself.
   *
   * \sa dot(), norm()
   */
@@ -126,7 +128,9 @@ EIGEN_STRONG_INLINE typename NumTraits<typename internal::traits<Derived>::Scala
   return internal::real((*this).cwiseAbs2().sum());
 }
 
-/** \returns the \em l2 norm of *this, i.e., for vectors, the square root of the dot product of *this with itself.
+/** \returns, for vectors, the \em l2 norm of \c *this, and for matrices the Frobenius norm.
+  * In both cases, it consists in the square root of the sum of the square of all the matrix entries.
+  * For vectors, this is also equals to the square root of the dot product of \c *this with itself.
   *
   * \sa dot(), squaredNorm()
   */
@@ -172,7 +176,7 @@ template<typename Derived, int p>
 struct lpNorm_selector
 {
   typedef typename NumTraits<typename traits<Derived>::Scalar>::Real RealScalar;
-  inline static RealScalar run(const MatrixBase<Derived>& m)
+  static inline RealScalar run(const MatrixBase<Derived>& m)
   {
     return pow(m.cwiseAbs().array().pow(p).sum(), RealScalar(1)/p);
   }
@@ -181,7 +185,7 @@ struct lpNorm_selector
 template<typename Derived>
 struct lpNorm_selector<Derived, 1>
 {
-  inline static typename NumTraits<typename traits<Derived>::Scalar>::Real run(const MatrixBase<Derived>& m)
+  static inline typename NumTraits<typename traits<Derived>::Scalar>::Real run(const MatrixBase<Derived>& m)
   {
     return m.cwiseAbs().sum();
   }
@@ -190,7 +194,7 @@ struct lpNorm_selector<Derived, 1>
 template<typename Derived>
 struct lpNorm_selector<Derived, 2>
 {
-  inline static typename NumTraits<typename traits<Derived>::Scalar>::Real run(const MatrixBase<Derived>& m)
+  static inline typename NumTraits<typename traits<Derived>::Scalar>::Real run(const MatrixBase<Derived>& m)
   {
     return m.norm();
   }
@@ -199,7 +203,7 @@ struct lpNorm_selector<Derived, 2>
 template<typename Derived>
 struct lpNorm_selector<Derived, Infinity>
 {
-  inline static typename NumTraits<typename traits<Derived>::Scalar>::Real run(const MatrixBase<Derived>& m)
+  static inline typename NumTraits<typename traits<Derived>::Scalar>::Real run(const MatrixBase<Derived>& m)
   {
     return m.cwiseAbs().maxCoeff();
   }

Eigen/src/Core/Functors.h

@@ -116,7 +116,7 @@ struct functor_traits<scalar_conj_product_op<LhsScalar,RhsScalar> > {
   */
 template<typename Scalar> struct scalar_min_op {
   EIGEN_EMPTY_STRUCT_CTOR(scalar_min_op)
-  EIGEN_STRONG_INLINE const Scalar operator() (const Scalar& a, const Scalar& b) const { return std::min(a, b); }
+  EIGEN_STRONG_INLINE const Scalar operator() (const Scalar& a, const Scalar& b) const { using std::min; return (min)(a, b); }
   template<typename Packet>
   EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
   { return internal::pmin(a,b); }
@@ -139,7 +139,7 @@ struct functor_traits<scalar_min_op<Scalar> > {
   */
 template<typename Scalar> struct scalar_max_op {
   EIGEN_EMPTY_STRUCT_CTOR(scalar_max_op)
-  EIGEN_STRONG_INLINE const Scalar operator() (const Scalar& a, const Scalar& b) const { return std::max(a, b); }
+  EIGEN_STRONG_INLINE const Scalar operator() (const Scalar& a, const Scalar& b) const { using std::max; return (max)(a, b); }
   template<typename Packet>
   EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
   { return internal::pmax(a,b); }
@@ -165,8 +165,10 @@ template<typename Scalar> struct scalar_hypot_op {
 //   typedef typename NumTraits<Scalar>::Real result_type;
   EIGEN_STRONG_INLINE const Scalar operator() (const Scalar& _x, const Scalar& _y) const
   {
-    Scalar p = std::max(_x, _y);
-    Scalar q = std::min(_x, _y);
+    using std::max;
+    using std::min;
+    Scalar p = (max)(_x, _y);
+    Scalar q = (min)(_x, _y);
     Scalar qp = q/p;
     return p * sqrt(Scalar(1) + qp*qp);
   }
@@ -218,6 +220,38 @@ struct functor_traits<scalar_quotient_op<Scalar> > {
   };
 };
 
+/** \internal
+  * \brief Template functor to compute the and of two booleans
+  *
+  * \sa class CwiseBinaryOp, ArrayBase::operator&&
+  */
+struct scalar_boolean_and_op {
+  EIGEN_EMPTY_STRUCT_CTOR(scalar_boolean_and_op)
+  EIGEN_STRONG_INLINE bool operator() (const bool& a, const bool& b) const { return a && b; }
+};
+template<> struct functor_traits<scalar_boolean_and_op> {
+  enum {
+    Cost = NumTraits<bool>::AddCost,
+    PacketAccess = false
+  };
+};
+
+/** \internal
+  * \brief Template functor to compute the or of two booleans
+  *
+  * \sa class CwiseBinaryOp, ArrayBase::operator||
+  */
+struct scalar_boolean_or_op {
+  EIGEN_EMPTY_STRUCT_CTOR(scalar_boolean_or_op)
+  EIGEN_STRONG_INLINE bool operator() (const bool& a, const bool& b) const { return a || b; }
+};
+template<> struct functor_traits<scalar_boolean_or_op> {
+  enum {
+    Cost = NumTraits<bool>::AddCost,
+    PacketAccess = false
+  };
+};
+
 // unary functors:
 
 /** \internal
@@ -605,7 +639,7 @@ template <typename Scalar, bool RandomAccess> struct linspaced_op
   EIGEN_STRONG_INLINE const Packet packetOp(Index row, Index col) const
   {
     eigen_assert(col==0 || row==0);
-    return impl(col + row);
+    return impl.packetOp(col + row);
   }
 
   // This proxy object handles the actual required temporaries, the different
@@ -669,7 +703,7 @@ struct functor_traits<scalar_sqrt_op<Scalar> >
 
 /** \internal
   * \brief Template functor to compute the cosine of a scalar
-  * \sa class CwiseUnaryOp, Cwise::cos()
+  * \sa class CwiseUnaryOp, ArrayBase::cos()
   */
 template<typename Scalar> struct scalar_cos_op {
   EIGEN_EMPTY_STRUCT_CTOR(scalar_cos_op)
@@ -688,7 +722,7 @@ struct functor_traits<scalar_cos_op<Scalar> >
 
 /** \internal
   * \brief Template functor to compute the sine of a scalar
-  * \sa class CwiseUnaryOp, Cwise::sin()
+  * \sa class CwiseUnaryOp, ArrayBase::sin()
   */
 template<typename Scalar> struct scalar_sin_op {
   EIGEN_EMPTY_STRUCT_CTOR(scalar_sin_op)
@@ -708,7 +742,7 @@ struct functor_traits<scalar_sin_op<Scalar> >
 
 /** \internal
   * \brief Template functor to compute the tan of a scalar
-  * \sa class CwiseUnaryOp, Cwise::tan()
+  * \sa class CwiseUnaryOp, ArrayBase::tan()
   */
 template<typename Scalar> struct scalar_tan_op {
   EIGEN_EMPTY_STRUCT_CTOR(scalar_tan_op)
@@ -725,6 +759,44 @@ struct functor_traits<scalar_tan_op<Scalar> >
   };
 };
 
+/** \internal
+  * \brief Template functor to compute the arc cosine of a scalar
+  * \sa class CwiseUnaryOp, ArrayBase::acos()
+  */
+template<typename Scalar> struct scalar_acos_op {
+  EIGEN_EMPTY_STRUCT_CTOR(scalar_acos_op)
+  inline const Scalar operator() (const Scalar& a) const { return acos(a); }
+  typedef typename packet_traits<Scalar>::type Packet;
+  inline Packet packetOp(const Packet& a) const { return internal::pacos(a); }
+};
+template<typename Scalar>
+struct functor_traits<scalar_acos_op<Scalar> >
+{
+  enum {
+    Cost = 5 * NumTraits<Scalar>::MulCost,
+    PacketAccess = packet_traits<Scalar>::HasACos
+  };
+};
+
+/** \internal
+  * \brief Template functor to compute the arc sine of a scalar
+  * \sa class CwiseUnaryOp, ArrayBase::asin()
+  */
+template<typename Scalar> struct scalar_asin_op {
+  EIGEN_EMPTY_STRUCT_CTOR(scalar_asin_op)
+  inline const Scalar operator() (const Scalar& a) const { return asin(a); }
+  typedef typename packet_traits<Scalar>::type Packet;
+  inline Packet packetOp(const Packet& a) const { return internal::pasin(a); }
+};
+template<typename Scalar>
+struct functor_traits<scalar_asin_op<Scalar> >
+{
+  enum {
+    Cost = 5 * NumTraits<Scalar>::MulCost,
+    PacketAccess = packet_traits<Scalar>::HasASin
+  };
+};
+
 /** \internal
   * \brief Template functor to raise a scalar to a power
   * \sa class CwiseUnaryOp, Cwise::pow

Eigen/src/Core/Fuzzy.h

@@ -34,9 +34,10 @@ struct isApprox_selector
 {
   static bool run(const Derived& x, const OtherDerived& y, typename Derived::RealScalar prec)
   {
-    const typename internal::nested<Derived,2>::type nested(x);
-    const typename internal::nested<OtherDerived,2>::type otherNested(y);
-    return (nested - otherNested).cwiseAbs2().sum() <= prec * prec * std::min(nested.cwiseAbs2().sum(), otherNested.cwiseAbs2().sum());
+    using std::min;
+    typename internal::nested<Derived,2>::type nested(x);
+    typename internal::nested<OtherDerived,2>::type otherNested(y);
+    return (nested - otherNested).cwiseAbs2().sum() <= prec * prec * (min)(nested.cwiseAbs2().sum(), otherNested.cwiseAbs2().sum());
   }
 };
 

Eigen/src/Core/GeneralProduct.h

@@ -0,0 +1,624 @@
+// This file is part of Eigen, a lightweight C++ template library
+// for linear algebra.
+//
+// Copyright (C) 2006-2008 Benoit Jacob <jacob.benoit.1@gmail.com>
+// Copyright (C) 2008-2011 Gael Guennebaud <gael.guennebaud@inria.fr>
+//
+// Eigen is free software; you can redistribute it and/or
+// modify it under the terms of the GNU Lesser General Public
+// License as published by the Free Software Foundation; either
+// version 3 of the License, or (at your option) any later version.
+//
+// Alternatively, you can redistribute it and/or
+// modify it under the terms of the GNU General Public License as
+// published by the Free Software Foundation; either version 2 of
+// the License, or (at your option) any later version.
+//
+// Eigen is distributed in the hope that it will be useful, but WITHOUT ANY
+// WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+// FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License or the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU Lesser General Public
+// License and a copy of the GNU General Public License along with
+// Eigen. If not, see <http://www.gnu.org/licenses/>.
+
+#ifndef EIGEN_GENERAL_PRODUCT_H
+#define EIGEN_GENERAL_PRODUCT_H
+
+/** \class GeneralProduct
+  * \ingroup Core_Module
+  *
+  * \brief Expression of the product of two general matrices or vectors
+  *
+  * \param LhsNested the type used to store the left-hand side
+  * \param RhsNested the type used to store the right-hand side
+  * \param ProductMode the type of the product
+  *
+  * This class represents an expression of the product of two general matrices.
+  * We call a general matrix, a dense matrix with full storage. For instance,
+  * This excludes triangular, selfadjoint, and sparse matrices.
+  * It is the return type of the operator* between general matrices. Its template
+  * arguments are determined automatically by ProductReturnType. Therefore,
+  * GeneralProduct should never be used direclty. To determine the result type of a
+  * function which involves a matrix product, use ProductReturnType::Type.
+  *
+  * \sa ProductReturnType, MatrixBase::operator*(const MatrixBase<OtherDerived>&)
+  */
+template<typename Lhs, typename Rhs, int ProductType = internal::product_type<Lhs,Rhs>::value>
+class GeneralProduct;
+
+enum {
+  Large = 2,
+  Small = 3
+};
+
+namespace internal {
+
+template<int Rows, int Cols, int Depth> struct product_type_selector;
+
+template<int Size, int MaxSize> struct product_size_category
+{
+  enum { is_large = MaxSize == Dynamic ||
+                    Size >= EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD,
+         value = is_large  ? Large
+               : Size == 1 ? 1
+                           : Small
+  };
+};
+
+template<typename Lhs, typename Rhs> struct product_type
+{
+  typedef typename remove_all<Lhs>::type _Lhs;
+  typedef typename remove_all<Rhs>::type _Rhs;
+  enum {
+    MaxRows  = _Lhs::MaxRowsAtCompileTime,
+    Rows  = _Lhs::RowsAtCompileTime,
+    MaxCols  = _Rhs::MaxColsAtCompileTime,
+    Cols  = _Rhs::ColsAtCompileTime,
+    MaxDepth = EIGEN_SIZE_MIN_PREFER_FIXED(_Lhs::MaxColsAtCompileTime,
+                                           _Rhs::MaxRowsAtCompileTime),
+    Depth = EIGEN_SIZE_MIN_PREFER_FIXED(_Lhs::ColsAtCompileTime,
+                                        _Rhs::RowsAtCompileTime),
+    LargeThreshold = EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD
+  };
+
+  // the splitting into different lines of code here, introducing the _select enums and the typedef below,
+  // is to work around an internal compiler error with gcc 4.1 and 4.2.
+private:
+  enum {
+    rows_select = product_size_category<Rows,MaxRows>::value,
+    cols_select = product_size_category<Cols,MaxCols>::value,
+    depth_select = product_size_category<Depth,MaxDepth>::value
+  };
+  typedef product_type_selector<rows_select, cols_select, depth_select> selector;
+
+public:
+  enum {
+    value = selector::ret
+  };
+#ifdef EIGEN_DEBUG_PRODUCT
+  static void debug()
+  {
+      EIGEN_DEBUG_VAR(Rows);
+      EIGEN_DEBUG_VAR(Cols);
+      EIGEN_DEBUG_VAR(Depth);
+      EIGEN_DEBUG_VAR(rows_select);
+      EIGEN_DEBUG_VAR(cols_select);
+      EIGEN_DEBUG_VAR(depth_select);
+      EIGEN_DEBUG_VAR(value);
+  }
+#endif
+};
+
+
+/* The following allows to select the kind of product at compile time
+ * based on the three dimensions of the product.
+ * This is a compile time mapping from {1,Small,Large}^3 -> {product types} */
+// FIXME I'm not sure the current mapping is the ideal one.
+template<int M, int N>  struct product_type_selector<M,N,1>              { enum { ret = OuterProduct }; };
+template<int Depth>     struct product_type_selector<1,    1,    Depth>  { enum { ret = InnerProduct }; };
+template<>              struct product_type_selector<1,    1,    1>      { enum { ret = InnerProduct }; };
+template<>              struct product_type_selector<Small,1,    Small>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<1,    Small,Small>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<Small,Small,Small>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<Small, Small, 1>    { enum { ret = LazyCoeffBasedProductMode }; };
+template<>              struct product_type_selector<Small, Large, 1>    { enum { ret = LazyCoeffBasedProductMode }; };
+template<>              struct product_type_selector<Large, Small, 1>    { enum { ret = LazyCoeffBasedProductMode }; };
+template<>              struct product_type_selector<1,    Large,Small>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<1,    Large,Large>  { enum { ret = GemvProduct }; };
+template<>              struct product_type_selector<1,    Small,Large>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<Large,1,    Small>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<Large,1,    Large>  { enum { ret = GemvProduct }; };
+template<>              struct product_type_selector<Small,1,    Large>  { enum { ret = CoeffBasedProductMode }; };
+template<>              struct product_type_selector<Small,Small,Large>  { enum { ret = GemmProduct }; };
+template<>              struct product_type_selector<Large,Small,Large>  { enum { ret = GemmProduct }; };
+template<>              struct product_type_selector<Small,Large,Large>  { enum { ret = GemmProduct }; };
+template<>              struct product_type_selector<Large,Large,Large>  { enum { ret = GemmProduct }; };
+template<>              struct product_type_selector<Large,Small,Small>  { enum { ret = GemmProduct }; };
+template<>              struct product_type_selector<Small,Large,Small>  { enum { ret = GemmProduct }; };
+template<>              struct product_type_selector<Large,Large,Small>  { enum { ret = GemmProduct }; };
+
+} // end namespace internal
+
+/** \class ProductReturnType
+  * \ingroup Core_Module
+  *
+  * \brief Helper class to get the correct and optimized returned type of operator*
+  *
+  * \param Lhs the type of the left-hand side
+  * \param Rhs the type of the right-hand side
+  * \param ProductMode the type of the product (determined automatically by internal::product_mode)
+  *
+  * This class defines the typename Type representing the optimized product expression
+  * between two matrix expressions. In practice, using ProductReturnType<Lhs,Rhs>::Type
+  * is the recommended way to define the result type of a function returning an expression
+  * which involve a matrix product. The class Product should never be
+  * used directly.
+  *
+  * \sa class Product, MatrixBase::operator*(const MatrixBase<OtherDerived>&)
+  */
+template<typename Lhs, typename Rhs, int ProductType>
+struct ProductReturnType
+{
+  // TODO use the nested type to reduce instanciations ????
+//   typedef typename internal::nested<Lhs,Rhs::ColsAtCompileTime>::type LhsNested;
+//   typedef typename internal::nested<Rhs,Lhs::RowsAtCompileTime>::type RhsNested;
+
+  typedef GeneralProduct<Lhs/*Nested*/, Rhs/*Nested*/, ProductType> Type;
+};
+
+template<typename Lhs, typename Rhs>
+struct ProductReturnType<Lhs,Rhs,CoeffBasedProductMode>
+{
+  typedef typename internal::nested<Lhs, Rhs::ColsAtCompileTime, typename internal::plain_matrix_type<Lhs>::type >::type LhsNested;
+  typedef typename internal::nested<Rhs, Lhs::RowsAtCompileTime, typename internal::plain_matrix_type<Rhs>::type >::type RhsNested;
+  typedef CoeffBasedProduct<LhsNested, RhsNested, EvalBeforeAssigningBit | EvalBeforeNestingBit> Type;
+};
+
+template<typename Lhs, typename Rhs>
+struct ProductReturnType<Lhs,Rhs,LazyCoeffBasedProductMode>
+{
+  typedef typename internal::nested<Lhs, Rhs::ColsAtCompileTime, typename internal::plain_matrix_type<Lhs>::type >::type LhsNested;
+  typedef typename internal::nested<Rhs, Lhs::RowsAtCompileTime, typename internal::plain_matrix_type<Rhs>::type >::type RhsNested;
+  typedef CoeffBasedProduct<LhsNested, RhsNested, NestByRefBit> Type;
+};
+
+// this is a workaround for sun CC
+template<typename Lhs, typename Rhs>
+struct LazyProductReturnType : public ProductReturnType<Lhs,Rhs,LazyCoeffBasedProductMode>
+{};
+
+/***********************************************************************
+*  Implementation of Inner Vector Vector Product
+***********************************************************************/
+
+// FIXME : maybe the "inner product" could return a Scalar
+// instead of a 1x1 matrix ??
+// Pro: more natural for the user
+// Cons: this could be a problem if in a meta unrolled algorithm a matrix-matrix
+// product ends up to a row-vector times col-vector product... To tackle this use
+// case, we could have a specialization for Block<MatrixType,1,1> with: operator=(Scalar x);
+
+namespace internal {
+
+template<typename Lhs, typename Rhs>
+struct traits<GeneralProduct<Lhs,Rhs,InnerProduct> >
+ : traits<Matrix<typename scalar_product_traits<typename Lhs::Scalar, typename Rhs::Scalar>::ReturnType,1,1> >
+{};
+
+}
+
+template<typename Lhs, typename Rhs>
+class GeneralProduct<Lhs, Rhs, InnerProduct>
+  : internal::no_assignment_operator,
+    public Matrix<typename internal::scalar_product_traits<typename Lhs::Scalar, typename Rhs::Scalar>::ReturnType,1,1>
+{
+    typedef Matrix<typename internal::scalar_product_traits<typename Lhs::Scalar, typename Rhs::Scalar>::ReturnType,1,1> Base;
+  public:
+    GeneralProduct(const Lhs& lhs, const Rhs& rhs)
+    {
+      EIGEN_STATIC_ASSERT((internal::is_same<typename Lhs::RealScalar, typename Rhs::RealScalar>::value),
+        YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
+
+      Base::coeffRef(0,0) = (lhs.transpose().cwiseProduct(rhs)).sum();
+    }
+
+    /** Convertion to scalar */
+    operator const typename Base::Scalar() const {
+      return Base::coeff(0,0);
+    }
+};
+
+/***********************************************************************
+*  Implementation of Outer Vector Vector Product
+***********************************************************************/
+
+namespace internal {
+template<int StorageOrder> struct outer_product_selector;
+
+template<typename Lhs, typename Rhs>
+struct traits<GeneralProduct<Lhs,Rhs,OuterProduct> >
+ : traits<ProductBase<GeneralProduct<Lhs,Rhs,OuterProduct>, Lhs, Rhs> >
+{};
+
+}
+
+template<typename Lhs, typename Rhs>
+class GeneralProduct<Lhs, Rhs, OuterProduct>
+  : public ProductBase<GeneralProduct<Lhs,Rhs,OuterProduct>, Lhs, Rhs>
+{
+  public:
+    EIGEN_PRODUCT_PUBLIC_INTERFACE(GeneralProduct)
+
+    GeneralProduct(const Lhs& lhs, const Rhs& rhs) : Base(lhs,rhs)
+    {
+      EIGEN_STATIC_ASSERT((internal::is_same<typename Lhs::RealScalar, typename Rhs::RealScalar>::value),
+        YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
+    }
+
+    template<typename Dest> void scaleAndAddTo(Dest& dest, Scalar alpha) const
+    {
+      internal::outer_product_selector<(int(Dest::Flags)&RowMajorBit) ? RowMajor : ColMajor>::run(*this, dest, alpha);
+    }
+};
+
+namespace internal {
+
+template<> struct outer_product_selector<ColMajor> {
+  template<typename ProductType, typename Dest>
+  static EIGEN_DONT_INLINE void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha) {
+    typedef typename Dest::Index Index;
+    // FIXME make sure lhs is sequentially stored
+    // FIXME not very good if rhs is real and lhs complex while alpha is real too
+    const Index cols = dest.cols();
+    for (Index j=0; j<cols; ++j)
+      dest.col(j) += (alpha * prod.rhs().coeff(j)) * prod.lhs();
+  }
+};
+
+template<> struct outer_product_selector<RowMajor> {
+  template<typename ProductType, typename Dest>
+  static EIGEN_DONT_INLINE void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha) {
+    typedef typename Dest::Index Index;
+    // FIXME make sure rhs is sequentially stored
+    // FIXME not very good if lhs is real and rhs complex while alpha is real too
+    const Index rows = dest.rows();
+    for (Index i=0; i<rows; ++i)
+      dest.row(i) += (alpha * prod.lhs().coeff(i)) * prod.rhs();
+  }
+};
+
+} // end namespace internal
+
+/***********************************************************************
+*  Implementation of General Matrix Vector Product
+***********************************************************************/
+
+/*  According to the shape/flags of the matrix we have to distinghish 3 different cases:
+ *   1 - the matrix is col-major, BLAS compatible and M is large => call fast BLAS-like colmajor routine
+ *   2 - the matrix is row-major, BLAS compatible and N is large => call fast BLAS-like rowmajor routine
+ *   3 - all other cases are handled using a simple loop along the outer-storage direction.
+ *  Therefore we need a lower level meta selector.
+ *  Furthermore, if the matrix is the rhs, then the product has to be transposed.
+ */
+namespace internal {
+
+template<typename Lhs, typename Rhs>
+struct traits<GeneralProduct<Lhs,Rhs,GemvProduct> >
+ : traits<ProductBase<GeneralProduct<Lhs,Rhs,GemvProduct>, Lhs, Rhs> >
+{};
+
+template<int Side, int StorageOrder, bool BlasCompatible>
+struct gemv_selector;
+
+} // end namespace internal
+
+template<typename Lhs, typename Rhs>
+class GeneralProduct<Lhs, Rhs, GemvProduct>
+  : public ProductBase<GeneralProduct<Lhs,Rhs,GemvProduct>, Lhs, Rhs>
+{
+  public:
+    EIGEN_PRODUCT_PUBLIC_INTERFACE(GeneralProduct)
+
+    typedef typename Lhs::Scalar LhsScalar;
+    typedef typename Rhs::Scalar RhsScalar;
+
+    GeneralProduct(const Lhs& lhs, const Rhs& rhs) : Base(lhs,rhs)
+    {
+//       EIGEN_STATIC_ASSERT((internal::is_same<typename Lhs::Scalar, typename Rhs::Scalar>::value),
+//         YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
+    }
+
+    enum { Side = Lhs::IsVectorAtCompileTime ? OnTheLeft : OnTheRight };
+    typedef typename internal::conditional<int(Side)==OnTheRight,_LhsNested,_RhsNested>::type MatrixType;
+
+    template<typename Dest> void scaleAndAddTo(Dest& dst, Scalar alpha) const
+    {
+      eigen_assert(m_lhs.rows() == dst.rows() && m_rhs.cols() == dst.cols());
+      internal::gemv_selector<Side,(int(MatrixType::Flags)&RowMajorBit) ? RowMajor : ColMajor,
+                       bool(internal::blas_traits<MatrixType>::HasUsableDirectAccess)>::run(*this, dst, alpha);
+    }
+};
+
+namespace internal {
+
+// The vector is on the left => transposition
+template<int StorageOrder, bool BlasCompatible>
+struct gemv_selector<OnTheLeft,StorageOrder,BlasCompatible>
+{
+  template<typename ProductType, typename Dest>
+  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
+  {
+    Transpose<Dest> destT(dest);
+    enum { OtherStorageOrder = StorageOrder == RowMajor ? ColMajor : RowMajor };
+    gemv_selector<OnTheRight,OtherStorageOrder,BlasCompatible>
+      ::run(GeneralProduct<Transpose<const typename ProductType::_RhsNested>,Transpose<const typename ProductType::_LhsNested>, GemvProduct>
+        (prod.rhs().transpose(), prod.lhs().transpose()), destT, alpha);
+  }
+};
+
+template<typename Scalar,int Size,int MaxSize,bool Cond> struct gemv_static_vector_if;
+
+template<typename Scalar,int Size,int MaxSize>
+struct gemv_static_vector_if<Scalar,Size,MaxSize,false>
+{
+  EIGEN_STRONG_INLINE  Scalar* data() { eigen_internal_assert(false && "should never be called"); return 0; }
+};
+
+template<typename Scalar,int Size>
+struct gemv_static_vector_if<Scalar,Size,Dynamic,true>
+{
+  EIGEN_STRONG_INLINE Scalar* data() { return 0; }
+};
+
+template<typename Scalar,int Size,int MaxSize>
+struct gemv_static_vector_if<Scalar,Size,MaxSize,true>
+{
+  #if EIGEN_ALIGN_STATICALLY
+  internal::plain_array<Scalar,EIGEN_SIZE_MIN_PREFER_FIXED(Size,MaxSize),0> m_data;
+  EIGEN_STRONG_INLINE Scalar* data() { return m_data.array; }
+  #else
+  // Some architectures cannot align on the stack,
+  // => let's manually enforce alignment by allocating more data and return the address of the first aligned element.
+  enum {
+    ForceAlignment  = internal::packet_traits<Scalar>::Vectorizable,
+    PacketSize      = internal::packet_traits<Scalar>::size
+  };
+  internal::plain_array<Scalar,EIGEN_SIZE_MIN_PREFER_FIXED(Size,MaxSize)+(ForceAlignment?PacketSize:0),0> m_data;
+  EIGEN_STRONG_INLINE Scalar* data() {
+    return ForceAlignment
+            ? reinterpret_cast<Scalar*>((reinterpret_cast<size_t>(m_data.array) & ~(size_t(15))) + 16)
+            : m_data.array;
+  }
+  #endif
+};
+
+template<> struct gemv_selector<OnTheRight,ColMajor,true>
+{
+  template<typename ProductType, typename Dest>
+  static inline void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
+  {
+    typedef typename ProductType::Index Index;
+    typedef typename ProductType::LhsScalar   LhsScalar;
+    typedef typename ProductType::RhsScalar   RhsScalar;
+    typedef typename ProductType::Scalar      ResScalar;
+    typedef typename ProductType::RealScalar  RealScalar;
+    typedef typename ProductType::ActualLhsType ActualLhsType;
+    typedef typename ProductType::ActualRhsType ActualRhsType;
+    typedef typename ProductType::LhsBlasTraits LhsBlasTraits;
+    typedef typename ProductType::RhsBlasTraits RhsBlasTraits;
+    typedef Map<Matrix<ResScalar,Dynamic,1>, Aligned> MappedDest;
+
+    ActualLhsType actualLhs = LhsBlasTraits::extract(prod.lhs());
+    ActualRhsType actualRhs = RhsBlasTraits::extract(prod.rhs());
+
+    ResScalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(prod.lhs())
+                                  * RhsBlasTraits::extractScalarFactor(prod.rhs());
+
+    enum {
+      // FIXME find a way to allow an inner stride on the result if packet_traits<Scalar>::size==1
+      // on, the other hand it is good for the cache to pack the vector anyways...
+      EvalToDestAtCompileTime = Dest::InnerStrideAtCompileTime==1,
+      ComplexByReal = (NumTraits<LhsScalar>::IsComplex) && (!NumTraits<RhsScalar>::IsComplex),
+      MightCannotUseDest = (Dest::InnerStrideAtCompileTime!=1) || ComplexByReal
+    };
+
+    gemv_static_vector_if<ResScalar,Dest::SizeAtCompileTime,Dest::MaxSizeAtCompileTime,MightCannotUseDest> static_dest;
+
+    bool alphaIsCompatible = (!ComplexByReal) || (imag(actualAlpha)==RealScalar(0));
+    bool evalToDest = EvalToDestAtCompileTime && alphaIsCompatible;
+    
+    RhsScalar compatibleAlpha = get_factor<ResScalar,RhsScalar>::run(actualAlpha);
+
+    ei_declare_aligned_stack_constructed_variable(ResScalar,actualDestPtr,dest.size(),
+                                                  evalToDest ? dest.data() : static_dest.data());
+    
+    if(!evalToDest)
+    {
+      #ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
+      int size = dest.size();
+      EIGEN_DENSE_STORAGE_CTOR_PLUGIN
+      #endif
+      if(!alphaIsCompatible)
+      {
+        MappedDest(actualDestPtr, dest.size()).setZero();
+        compatibleAlpha = RhsScalar(1);
+      }
+      else
+        MappedDest(actualDestPtr, dest.size()) = dest;
+    }
+
+    general_matrix_vector_product
+      <Index,LhsScalar,ColMajor,LhsBlasTraits::NeedToConjugate,RhsScalar,RhsBlasTraits::NeedToConjugate>::run(
+        actualLhs.rows(), actualLhs.cols(),
+        actualLhs.data(), actualLhs.outerStride(),
+        actualRhs.data(), actualRhs.innerStride(),
+        actualDestPtr, 1,
+        compatibleAlpha);
+
+    if (!evalToDest)
+    {
+      if(!alphaIsCompatible)
+        dest += actualAlpha * MappedDest(actualDestPtr, dest.size());
+      else
+        dest = MappedDest(actualDestPtr, dest.size());
+    }
+  }
+};
+
+template<> struct gemv_selector<OnTheRight,RowMajor,true>
+{
+  template<typename ProductType, typename Dest>
+  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
+  {
+    typedef typename ProductType::LhsScalar LhsScalar;
+    typedef typename ProductType::RhsScalar RhsScalar;
+    typedef typename ProductType::Scalar    ResScalar;
+    typedef typename ProductType::Index Index;
+    typedef typename ProductType::ActualLhsType ActualLhsType;
+    typedef typename ProductType::ActualRhsType ActualRhsType;
+    typedef typename ProductType::_ActualRhsType _ActualRhsType;
+    typedef typename ProductType::LhsBlasTraits LhsBlasTraits;
+    typedef typename ProductType::RhsBlasTraits RhsBlasTraits;
+
+    typename add_const<ActualLhsType>::type actualLhs = LhsBlasTraits::extract(prod.lhs());
+    typename add_const<ActualRhsType>::type actualRhs = RhsBlasTraits::extract(prod.rhs());
+
+    ResScalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(prod.lhs())
+                                  * RhsBlasTraits::extractScalarFactor(prod.rhs());
+
+    enum {
+      // FIXME find a way to allow an inner stride on the result if packet_traits<Scalar>::size==1
+      // on, the other hand it is good for the cache to pack the vector anyways...
+      DirectlyUseRhs = _ActualRhsType::InnerStrideAtCompileTime==1
+    };
+
+    gemv_static_vector_if<RhsScalar,_ActualRhsType::SizeAtCompileTime,_ActualRhsType::MaxSizeAtCompileTime,!DirectlyUseRhs> static_rhs;
+
+    ei_declare_aligned_stack_constructed_variable(RhsScalar,actualRhsPtr,actualRhs.size(),
+        DirectlyUseRhs ? const_cast<RhsScalar*>(actualRhs.data()) : static_rhs.data());
+
+    if(!DirectlyUseRhs)
+    {
+      #ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
+      int size = actualRhs.size();
+      EIGEN_DENSE_STORAGE_CTOR_PLUGIN
+      #endif
+      Map<typename _ActualRhsType::PlainObject>(actualRhsPtr, actualRhs.size()) = actualRhs;
+    }
+
+    general_matrix_vector_product
+      <Index,LhsScalar,RowMajor,LhsBlasTraits::NeedToConjugate,RhsScalar,RhsBlasTraits::NeedToConjugate>::run(
+        actualLhs.rows(), actualLhs.cols(),
+        actualLhs.data(), actualLhs.outerStride(),
+        actualRhsPtr, 1,
+        dest.data(), dest.innerStride(),
+        actualAlpha);
+  }
+};
+
+template<> struct gemv_selector<OnTheRight,ColMajor,false>
+{
+  template<typename ProductType, typename Dest>
+  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
+  {
+    typedef typename Dest::Index Index;
+    // TODO makes sure dest is sequentially stored in memory, otherwise use a temp
+    const Index size = prod.rhs().rows();
+    for(Index k=0; k<size; ++k)
+      dest += (alpha*prod.rhs().coeff(k)) * prod.lhs().col(k);
+  }
+};
+
+template<> struct gemv_selector<OnTheRight,RowMajor,false>
+{
+  template<typename ProductType, typename Dest>
+  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
+  {
+    typedef typename Dest::Index Index;
+    // TODO makes sure rhs is sequentially stored in memory, otherwise use a temp
+    const Index rows = prod.rows();
+    for(Index i=0; i<rows; ++i)
+      dest.coeffRef(i) += alpha * (prod.lhs().row(i).cwiseProduct(prod.rhs().transpose())).sum();
+  }
+};
+
+} // end namespace internal
+
+/***************************************************************************
+* Implementation of matrix base methods
+***************************************************************************/
+
+/** \returns the matrix product of \c *this and \a other.
+  *
+  * \note If instead of the matrix product you want the coefficient-wise product, see Cwise::operator*().
+  *
+  * \sa lazyProduct(), operator*=(const MatrixBase&), Cwise::operator*()
+  */
+template<typename Derived>
+template<typename OtherDerived>
+inline const typename ProductReturnType<Derived, OtherDerived>::Type
+MatrixBase<Derived>::operator*(const MatrixBase<OtherDerived> &other) const
+{
+  // A note regarding the function declaration: In MSVC, this function will sometimes
+  // not be inlined since DenseStorage is an unwindable object for dynamic
+  // matrices and product types are holding a member to store the result.
+  // Thus it does not help tagging this function with EIGEN_STRONG_INLINE.
+  enum {
+    ProductIsValid =  Derived::ColsAtCompileTime==Dynamic
+                   || OtherDerived::RowsAtCompileTime==Dynamic
+                   || int(Derived::ColsAtCompileTime)==int(OtherDerived::RowsAtCompileTime),
+    AreVectors = Derived::IsVectorAtCompileTime && OtherDerived::IsVectorAtCompileTime,
+    SameSizes = EIGEN_PREDICATE_SAME_MATRIX_SIZE(Derived,OtherDerived)
+  };
+  // note to the lost user:
+  //    * for a dot product use: v1.dot(v2)
+  //    * for a coeff-wise product use: v1.cwiseProduct(v2)
+  EIGEN_STATIC_ASSERT(ProductIsValid || !(AreVectors && SameSizes),
+    INVALID_VECTOR_VECTOR_PRODUCT__IF_YOU_WANTED_A_DOT_OR_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTIONS)
+  EIGEN_STATIC_ASSERT(ProductIsValid || !(SameSizes && !AreVectors),
+    INVALID_MATRIX_PRODUCT__IF_YOU_WANTED_A_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTION)
+  EIGEN_STATIC_ASSERT(ProductIsValid || SameSizes, INVALID_MATRIX_PRODUCT)
+#ifdef EIGEN_DEBUG_PRODUCT
+  internal::product_type<Derived,OtherDerived>::debug();
+#endif
+  return typename ProductReturnType<Derived,OtherDerived>::Type(derived(), other.derived());
+}
+
+/** \returns an expression of the matrix product of \c *this and \a other without implicit evaluation.
+  *
+  * The returned product will behave like any other expressions: the coefficients of the product will be
+  * computed once at a time as requested. This might be useful in some extremely rare cases when only
+  * a small and no coherent fraction of the result's coefficients have to be computed.
+  *
+  * \warning This version of the matrix product can be much much slower. So use it only if you know
+  * what you are doing and that you measured a true speed improvement.
+  *
+  * \sa operator*(const MatrixBase&)
+  */
+template<typename Derived>
+template<typename OtherDerived>
+const typename LazyProductReturnType<Derived,OtherDerived>::Type
+MatrixBase<Derived>::lazyProduct(const MatrixBase<OtherDerived> &other) const
+{
+  enum {
+    ProductIsValid =  Derived::ColsAtCompileTime==Dynamic
+                   || OtherDerived::RowsAtCompileTime==Dynamic
+                   || int(Derived::ColsAtCompileTime)==int(OtherDerived::RowsAtCompileTime),
+    AreVectors = Derived::IsVectorAtCompileTime && OtherDerived::IsVectorAtCompileTime,
+    SameSizes = EIGEN_PREDICATE_SAME_MATRIX_SIZE(Derived,OtherDerived)
+  };
+  // note to the lost user:
+  //    * for a dot product use: v1.dot(v2)
+  //    * for a coeff-wise product use: v1.cwiseProduct(v2)
+  EIGEN_STATIC_ASSERT(ProductIsValid || !(AreVectors && SameSizes),
+    INVALID_VECTOR_VECTOR_PRODUCT__IF_YOU_WANTED_A_DOT_OR_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTIONS)
+  EIGEN_STATIC_ASSERT(ProductIsValid || !(SameSizes && !AreVectors),
+    INVALID_MATRIX_PRODUCT__IF_YOU_WANTED_A_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTION)
+  EIGEN_STATIC_ASSERT(ProductIsValid || SameSizes, INVALID_MATRIX_PRODUCT)
+
+  return typename LazyProductReturnType<Derived,OtherDerived>::Type(derived(), other.derived());
+}
+
+#endif // EIGEN_PRODUCT_H

Eigen/src/Core/GenericPacketMath.h

@@ -134,12 +134,12 @@ pdiv(const Packet& a,
 /** \internal \returns the min of \a a and \a b  (coeff-wise) */
 template<typename Packet> inline Packet
 pmin(const Packet& a,
-        const Packet& b) { return std::min(a, b); }
+        const Packet& b) { using std::min; return (min)(a, b); }
 
 /** \internal \returns the max of \a a and \a b  (coeff-wise) */
 template<typename Packet> inline Packet
 pmax(const Packet& a,
-        const Packet& b) { return std::max(a, b); }
+        const Packet& b) { using std::max; return (max)(a, b); }
 
 /** \internal \returns the absolute value of \a a */
 template<typename Packet> inline Packet
@@ -225,15 +225,20 @@ template<typename Packet> inline typename unpacket_traits<Packet>::type predux_m
 template<typename Packet> inline Packet preverse(const Packet& a)
 { return a; }
 
+
+/** \internal \returns \a a with real and imaginary part flipped (for complex type only) */
+template<typename Packet> inline Packet pcplxflip(const Packet& a)
+{ return Packet(imag(a),real(a)); }
+
 /**************************
 * Special math functions
 ***************************/
 
-/** \internal \returns the sin of \a a (coeff-wise) */
+/** \internal \returns the sine of \a a (coeff-wise) */
 template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
 Packet psin(const Packet& a) { return sin(a); }
 
-/** \internal \returns the cos of \a a (coeff-wise) */
+/** \internal \returns the cosine of \a a (coeff-wise) */
 template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
 Packet pcos(const Packet& a) { return cos(a); }
 
@@ -241,6 +246,14 @@ Packet pcos(const Packet& a) { return cos(a); }
 template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
 Packet ptan(const Packet& a) { return tan(a); }
 
+/** \internal \returns the arc sine of \a a (coeff-wise) */
+template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
+Packet pasin(const Packet& a) { return asin(a); }
+
+/** \internal \returns the arc cosine of \a a (coeff-wise) */
+template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
+Packet pacos(const Packet& a) { return acos(a); }
+
 /** \internal \returns the exp of \a a (coeff-wise) */
 template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
 Packet pexp(const Packet& a) { return exp(a); }
@@ -257,6 +270,14 @@ Packet psqrt(const Packet& a) { return sqrt(a); }
 * The following functions might not have to be overwritten for vectorized types
 ***************************************************************************/
 
+/** \internal copy a packet with constant coeficient \a a (e.g., [a,a,a,a]) to \a *to. \a to must be 16 bytes aligned */
+// NOTE: this function must really be templated on the packet type (think about different packet types for the same scalar type)
+template<typename Packet>
+inline void pstore1(typename unpacket_traits<Packet>::type* to, const typename unpacket_traits<Packet>::type& a)
+{
+  pstore(to, pset1<Packet>(a));
+}
+
 /** \internal \returns a * b + c (coeff-wise) */
 template<typename Packet> inline Packet
 pmadd(const Packet&  a,
@@ -265,7 +286,7 @@ pmadd(const Packet&  a,
 { return padd(pmul(a, b),c); }
 
 /** \internal \returns a packet version of \a *from.
-  * \If LoadMode equals Aligned, \a from must be 16 bytes aligned */
+  * If LoadMode equals #Aligned, \a from must be 16 bytes aligned */
 template<typename Packet, int LoadMode>
 inline Packet ploadt(const typename unpacket_traits<Packet>::type* from)
 {
@@ -276,7 +297,7 @@ inline Packet ploadt(const typename unpacket_traits<Packet>::type* from)
 }
 
 /** \internal copy the packet \a from to \a *to.
-  * If StoreMode equals Aligned, \a to must be 16 bytes aligned */
+  * If StoreMode equals #Aligned, \a to must be 16 bytes aligned */
 template<typename Scalar, typename Packet, int LoadMode>
 inline void pstoret(Scalar* to, const Packet& from)
 {
@@ -291,7 +312,7 @@ template<int Offset,typename PacketType>
 struct palign_impl
 {
   // by default data are aligned, so there is nothing to be done :)
-  inline static void run(PacketType&, const PacketType&) {}
+  static inline void run(PacketType&, const PacketType&) {}
 };
 
 /** \internal update \a first using the concatenation of the \a Offset last elements

Eigen/src/Core/GlobalFunctions.h

@@ -56,6 +56,8 @@ namespace std
   EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(imag,scalar_imag_op)
   EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(sin,scalar_sin_op)
   EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(cos,scalar_cos_op)
+  EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(asin,scalar_asin_op)
+  EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(acos,scalar_acos_op)
   EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(tan,scalar_tan_op)
   EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(exp,scalar_exp_op)
   EIGEN_ARRAY_DECLARE_GLOBAL_STD_UNARY(log,scalar_log_op)
@@ -77,6 +79,8 @@ namespace Eigen
     EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(imag,scalar_imag_op)
     EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(sin,scalar_sin_op)
     EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(cos,scalar_cos_op)
+    EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(asin,scalar_asin_op)
+    EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(acos,scalar_acos_op)
     EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(tan,scalar_tan_op)
     EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(exp,scalar_exp_op)
     EIGEN_ARRAY_DECLARE_GLOBAL_EIGEN_UNARY(log,scalar_log_op)

Eigen/src/Core/IO.h

@@ -141,7 +141,8 @@ struct significant_decimals_default_impl
   typedef typename NumTraits<Scalar>::Real RealScalar;
   static inline int run()
   {
-    return cast<RealScalar,int>(std::ceil(-log(NumTraits<RealScalar>::epsilon())/log(RealScalar(10))));
+    using std::ceil;
+    return cast<RealScalar,int>(ceil(-log(NumTraits<RealScalar>::epsilon())/log(RealScalar(10))));
   }
 };
 
@@ -170,7 +171,7 @@ std::ostream & print_matrix(std::ostream & s, const Derived& _m, const IOFormat&
     return s;
   }
   
-  const typename Derived::Nested m = _m;
+  typename Derived::Nested m = _m;
   typedef typename Derived::Scalar Scalar;
   typedef typename Derived::Index Index;
 

Eigen/src/Core/Map.h

@@ -31,10 +31,10 @@
   *
   * \brief A matrix or vector expression mapping an existing array of data.
   *
-  * \param PlainObjectType the equivalent matrix type of the mapped data
-  * \param MapOptions specifies whether the pointer is \c Aligned, or \c Unaligned.
-  *                The default is \c Unaligned.
-  * \param StrideType optionnally specifies strides. By default, Map assumes the memory layout
+  * \tparam PlainObjectType the equivalent matrix type of the mapped data
+  * \tparam MapOptions specifies whether the pointer is \c #Aligned, or \c #Unaligned.
+  *                The default is \c #Unaligned.
+  * \tparam StrideType optionally specifies strides. By default, Map assumes the memory layout
   *                   of an ordinary, contiguous array. This can be overridden by specifying strides.
   *                   The type passed here must be a specialization of the Stride template, see examples below.
   *
@@ -72,9 +72,9 @@
   * Example: \include Map_placement_new.cpp
   * Output: \verbinclude Map_placement_new.out
   *
-  * This class is the return type of Matrix::Map() but can also be used directly.
+  * This class is the return type of PlainObjectBase::Map() but can also be used directly.
   *
-  * \sa Matrix::Map(), \ref TopicStorageOrders
+  * \sa PlainObjectBase::Map(), \ref TopicStorageOrders
   */
 
 namespace internal {
@@ -95,7 +95,7 @@ struct traits<Map<PlainObjectType, MapOptions, StrideType> >
     HasNoInnerStride = InnerStrideAtCompileTime == 1,
     HasNoOuterStride = StrideType::OuterStrideAtCompileTime == 0,
     HasNoStride = HasNoInnerStride && HasNoOuterStride,
-    IsAligned = int(int(MapOptions)&Aligned)==Aligned,
+    IsAligned = bool(EIGEN_ALIGN) && ((int(MapOptions)&Aligned)==Aligned),
     IsDynamicSize = PlainObjectType::SizeAtCompileTime==Dynamic,
     KeepsPacketAccess = bool(HasNoInnerStride)
                         && ( bool(IsDynamicSize)
@@ -192,14 +192,14 @@ template<typename _Scalar, int _Rows, int _Cols, int _Options, int _MaxRows, int
 inline Array<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols>
   ::Array(const Scalar *data)
 {
-  _set_noalias(Eigen::Map<const Array>(data));
+  this->_set_noalias(Eigen::Map<const Array>(data));
 }
 
 template<typename _Scalar, int _Rows, int _Cols, int _Options, int _MaxRows, int _MaxCols>
 inline Matrix<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols>
   ::Matrix(const Scalar *data)
 {
-  _set_noalias(Eigen::Map<const Matrix>(data));
+  this->_set_noalias(Eigen::Map<const Matrix>(data));
 }
 
 #endif // EIGEN_MAP_H

Eigen/src/Core/MapBase.h

@@ -85,6 +85,8 @@ template<typename Derived> class MapBase<Derived, ReadOnlyAccessors>
     using Base::rowStride;
     using Base::colStride;
 
+    // bug 217 - compile error on ICC 11.1
+    using Base::operator=;
 
     typedef typename Base::CoeffReturnType CoeffReturnType;
 
@@ -168,8 +170,8 @@ template<typename Derived> class MapBase<Derived, ReadOnlyAccessors>
       EIGEN_STATIC_ASSERT(EIGEN_IMPLIES(internal::traits<Derived>::Flags&PacketAccessBit,
                                         internal::inner_stride_at_compile_time<Derived>::ret==1),
                           PACKET_ACCESS_REQUIRES_TO_HAVE_INNER_STRIDE_FIXED_TO_1);
-      eigen_assert(EIGEN_IMPLIES(internal::traits<Derived>::Flags&AlignedBit, (size_t(m_data) % (sizeof(Scalar)*internal::packet_traits<Scalar>::size)) == 0)
-        && "data is not aligned");
+      eigen_assert(EIGEN_IMPLIES(internal::traits<Derived>::Flags&AlignedBit, (size_t(m_data) % 16) == 0)
+                   && "data is not aligned");
     }
 
     PointerType m_data;
@@ -236,7 +238,7 @@ template<typename Derived> class MapBase<Derived, WriteAccessors>
                 (this->m_data + index * innerStride(), x);
     }
 
-    inline MapBase(PointerType data) : Base(data) {}
+    explicit inline MapBase(PointerType data) : Base(data) {}
     inline MapBase(PointerType data, Index size) : Base(data, size) {}
     inline MapBase(PointerType data, Index rows, Index cols) : Base(data, rows, cols) {}
 

Eigen/src/Core/MathFunctions.h

@@ -87,7 +87,8 @@ struct real_impl<std::complex<RealScalar> >
 {
   static inline RealScalar run(const std::complex<RealScalar>& x)
   {
-    return std::real(x);
+    using std::real;
+    return real(x);
   }
 };
 
@@ -122,7 +123,8 @@ struct imag_impl<std::complex<RealScalar> >
 {
   static inline RealScalar run(const std::complex<RealScalar>& x)
   {
-    return std::imag(x);
+    using std::imag;
+    return imag(x);
   }
 };
 
@@ -244,7 +246,8 @@ struct conj_impl<std::complex<RealScalar> >
 {
   static inline std::complex<RealScalar> run(const std::complex<RealScalar>& x)
   {
-    return std::conj(x);
+    using std::conj;
+    return conj(x);
   }
 };
 
@@ -270,7 +273,8 @@ struct abs_impl
   typedef typename NumTraits<Scalar>::Real RealScalar;
   static inline RealScalar run(const Scalar& x)
   {
-    return std::abs(x);
+    using std::abs;
+    return abs(x);
   }
 };
 
@@ -305,7 +309,7 @@ struct abs2_impl<std::complex<RealScalar> >
 {
   static inline RealScalar run(const std::complex<RealScalar>& x)
   {
-    return std::norm(x);
+    return real(x)*real(x) + imag(x)*imag(x);
   }
 };
 
@@ -369,10 +373,12 @@ struct hypot_impl
   typedef typename NumTraits<Scalar>::Real RealScalar;
   static inline RealScalar run(const Scalar& x, const Scalar& y)
   {
+    using std::max;
+    using std::min;
     RealScalar _x = abs(x);
     RealScalar _y = abs(y);
-    RealScalar p = std::max(_x, _y);
-    RealScalar q = std::min(_x, _y);
+    RealScalar p = (max)(_x, _y);
+    RealScalar q = (min)(_x, _y);
     RealScalar qp = q/p;
     return p * sqrt(RealScalar(1) + qp*qp);
   }
@@ -420,7 +426,8 @@ struct sqrt_default_impl
 {
   static inline Scalar run(const Scalar& x)
   {
-    return std::sqrt(x);
+    using std::sqrt;
+    return sqrt(x);
   }
 };
 
@@ -454,194 +461,36 @@ inline EIGEN_MATHFUNC_RETVAL(sqrt, Scalar) sqrt(const Scalar& x)
 }
 
 /****************************************************************************
-* Implementation of exp                                                  *
+* Implementation of standard unary real functions (exp, log, sin, cos, ...  *
 ****************************************************************************/
 
-template<typename Scalar, bool IsInteger>
-struct exp_default_impl
-{
-  static inline Scalar run(const Scalar& x)
-  {
-    return std::exp(x);
+// This macro instanciate all the necessary template mechanism which is common to all unary real functions.
+#define EIGEN_MATHFUNC_STANDARD_REAL_UNARY(NAME) \
+  template<typename Scalar, bool IsInteger> struct NAME##_default_impl {            \
+    static inline Scalar run(const Scalar& x) { using std::NAME; return NAME(x); }  \
+  };                                                                                \
+  template<typename Scalar> struct NAME##_default_impl<Scalar, true> {              \
+    static inline Scalar run(const Scalar&) {                                       \
+      EIGEN_STATIC_ASSERT_NON_INTEGER(Scalar)                                       \
+      return Scalar(0);                                                             \
+    }                                                                               \
+  };                                                                                \
+  template<typename Scalar> struct NAME##_impl                                      \
+    : NAME##_default_impl<Scalar, NumTraits<Scalar>::IsInteger>                     \
+  {};                                                                               \
+  template<typename Scalar> struct NAME##_retval { typedef Scalar type; };          \
+  template<typename Scalar>                                                         \
+  inline EIGEN_MATHFUNC_RETVAL(NAME, Scalar) NAME(const Scalar& x) {                \
+    return EIGEN_MATHFUNC_IMPL(NAME, Scalar)::run(x);                               \
   }
-};
 
-template<typename Scalar>
-struct exp_default_impl<Scalar, true>
-{
-  static inline Scalar run(const Scalar&)
-  {
-    EIGEN_STATIC_ASSERT_NON_INTEGER(Scalar)
-    return Scalar(0);
-  }
-};
-
-template<typename Scalar>
-struct exp_impl : exp_default_impl<Scalar, NumTraits<Scalar>::IsInteger> {};
-
-template<typename Scalar>
-struct exp_retval
-{
-  typedef Scalar type;
-};
-
-template<typename Scalar>
-inline EIGEN_MATHFUNC_RETVAL(exp, Scalar) exp(const Scalar& x)
-{
-  return EIGEN_MATHFUNC_IMPL(exp, Scalar)::run(x);
-}
-
-/****************************************************************************
-* Implementation of cos                                                  *
-****************************************************************************/
-
-template<typename Scalar, bool IsInteger>
-struct cos_default_impl
-{
-  static inline Scalar run(const Scalar& x)
-  {
-    return std::cos(x);
-  }
-};
-
-template<typename Scalar>
-struct cos_default_impl<Scalar, true>
-{
-  static inline Scalar run(const Scalar&)
-  {
-    EIGEN_STATIC_ASSERT_NON_INTEGER(Scalar)
-    return Scalar(0);
-  }
-};
-
-template<typename Scalar>
-struct cos_impl : cos_default_impl<Scalar, NumTraits<Scalar>::IsInteger> {};
-
-template<typename Scalar>
-struct cos_retval
-{
-  typedef Scalar type;
-};
-
-template<typename Scalar>
-inline EIGEN_MATHFUNC_RETVAL(cos, Scalar) cos(const Scalar& x)
-{
-  return EIGEN_MATHFUNC_IMPL(cos, Scalar)::run(x);
-}
-
-/****************************************************************************
-* Implementation of sin                                                  *
-****************************************************************************/
-
-template<typename Scalar, bool IsInteger>
-struct sin_default_impl
-{
-  static inline Scalar run(const Scalar& x)
-  {
-    return std::sin(x);
-  }
-};
-
-template<typename Scalar>
-struct sin_default_impl<Scalar, true>
-{
-  static inline Scalar run(const Scalar&)
-  {
-    EIGEN_STATIC_ASSERT_NON_INTEGER(Scalar)
-    return Scalar(0);
-  }
-};
-
-template<typename Scalar>
-struct sin_impl : sin_default_impl<Scalar, NumTraits<Scalar>::IsInteger> {};
-
-template<typename Scalar>
-struct sin_retval
-{
-  typedef Scalar type;
-};
-
-template<typename Scalar>
-inline EIGEN_MATHFUNC_RETVAL(sin, Scalar) sin(const Scalar& x)
-{
-  return EIGEN_MATHFUNC_IMPL(sin, Scalar)::run(x);
-}
-
-/****************************************************************************
-* Implementation of tan                                                  *
-****************************************************************************/
-
-template<typename Scalar, bool IsInteger>
-struct tan_default_impl
-{
-  static inline Scalar run(const Scalar& x)
-  {
-    return std::tan(x);
-  }
-};
-
-template<typename Scalar>
-struct tan_default_impl<Scalar, true>
-{
-  static inline Scalar run(const Scalar&)
-  {
-    EIGEN_STATIC_ASSERT_NON_INTEGER(Scalar)
-    return Scalar(0);
-  }
-};
-
-template<typename Scalar>
-struct tan_impl : tan_default_impl<Scalar, NumTraits<Scalar>::IsInteger> {};
-
-template<typename Scalar>
-struct tan_retval
-{
-  typedef Scalar type;
-};
-
-template<typename Scalar>
-inline EIGEN_MATHFUNC_RETVAL(tan, Scalar) tan(const Scalar& x)
-{
-  return EIGEN_MATHFUNC_IMPL(tan, Scalar)::run(x);
-}
-
-/****************************************************************************
-* Implementation of log                                                  *
-****************************************************************************/
-
-template<typename Scalar, bool IsInteger>
-struct log_default_impl
-{
-  static inline Scalar run(const Scalar& x)
-  {
-    return std::log(x);
-  }
-};
-
-template<typename Scalar>
-struct log_default_impl<Scalar, true>
-{
-  static inline Scalar run(const Scalar&)
-  {
-    EIGEN_STATIC_ASSERT_NON_INTEGER(Scalar)
-    return Scalar(0);
-  }
-};
-
-template<typename Scalar>
-struct log_impl : log_default_impl<Scalar, NumTraits<Scalar>::IsInteger> {};
-
-template<typename Scalar>
-struct log_retval
-{
-  typedef Scalar type;
-};
-
-template<typename Scalar>
-inline EIGEN_MATHFUNC_RETVAL(log, Scalar) log(const Scalar& x)
-{
-  return EIGEN_MATHFUNC_IMPL(log, Scalar)::run(x);
-}
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(exp)
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(log)
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(sin)
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(cos)
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(tan)
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(asin)
+EIGEN_MATHFUNC_STANDARD_REAL_UNARY(acos)
 
 /****************************************************************************
 * Implementation of atan2                                                *
@@ -653,7 +502,8 @@ struct atan2_default_impl
   typedef Scalar retval;
   static inline Scalar run(const Scalar& x, const Scalar& y)
   {
-    return std::atan2(x, y);
+    using std::atan2;
+    return atan2(x, y);
   }
 };
 
@@ -692,7 +542,8 @@ struct pow_default_impl
   typedef Scalar retval;
   static inline Scalar run(const Scalar& x, const Scalar& y)
   {
-    return std::pow(x, y);
+    using std::pow;
+    return pow(x, y);
   }
 };
 
@@ -701,7 +552,7 @@ struct pow_default_impl<Scalar, true>
 {
   static inline Scalar run(Scalar x, Scalar y)
   {
-    Scalar res = 1;
+    Scalar res(1);
     eigen_assert(!NumTraits<Scalar>::IsSigned || y >= 0);
     if(y & 1) res *= x;
     y >>= 1;
@@ -884,7 +735,8 @@ struct scalar_fuzzy_default_impl<Scalar, false, false>
   }
   static inline bool isApprox(const Scalar& x, const Scalar& y, const RealScalar& prec)
   {
-    return abs(x - y) <= std::min(abs(x), abs(y)) * prec;
+    using std::min;
+    return abs(x - y) <= (min)(abs(x), abs(y)) * prec;
   }
   static inline bool isApproxOrLessThan(const Scalar& x, const Scalar& y, const RealScalar& prec)
   {
@@ -922,7 +774,8 @@ struct scalar_fuzzy_default_impl<Scalar, true, false>
   }
   static inline bool isApprox(const Scalar& x, const Scalar& y, const RealScalar& prec)
   {
-    return abs2(x - y) <= std::min(abs2(x), abs2(y)) * prec * prec;
+    using std::min;
+    return abs2(x - y) <= (min)(abs2(x), abs2(y)) * prec * prec;
   }
 };
 
@@ -984,6 +837,17 @@ template<> struct scalar_fuzzy_impl<bool>
   
 };
 
+/****************************************************************************
+* Special functions                                                          *
+****************************************************************************/
+
+// std::isfinite is non standard, so let's define our own version,
+// even though it is not very efficient.
+template<typename T> bool isfinite(const T& x)
+{
+  return x<NumTraits<T>::highest() && x>NumTraits<T>::lowest();
+}
+
 } // end namespace internal
 
 #endif // EIGEN_MATHFUNCTIONS_H

Eigen/src/Core/Matrix.h

@@ -43,8 +43,8 @@
   * \tparam _Cols Number of columns, or \b Dynamic
   *
   * The remaining template parameters are optional -- in most cases you don't have to worry about them.
-  * \tparam _Options \anchor matrix_tparam_options A combination of either \b RowMajor or \b ColMajor, and of either
-  *                 \b AutoAlign or \b DontAlign.
+  * \tparam _Options \anchor matrix_tparam_options A combination of either \b #RowMajor or \b #ColMajor, and of either
+  *                 \b #AutoAlign or \b #DontAlign.
   *                 The former controls \ref TopicStorageOrders "storage order", and defaults to column-major. The latter controls alignment, which is required
   *                 for vectorization. It defaults to aligning matrices except for fixed sizes that aren't a multiple of the packet size.
   * \tparam _MaxRows Maximum number of rows. Defaults to \a _Rows (\ref maxrows "note").
@@ -79,6 +79,9 @@
   * m(0, 3) = 3;
   * \endcode
   *
+  * This class can be extended with the help of the plugin mechanism described on the page
+  * \ref TopicCustomizingEigen by defining the preprocessor symbol \c EIGEN_MATRIX_PLUGIN.
+  *
   * <i><b>Some notes:</b></i>
   *
   * <dl>
@@ -150,10 +153,6 @@ class Matrix
 
     typedef typename Base::PlainObject PlainObject;
 
-    enum { NeedsToAlign = (!(Options&DontAlign))
-                          && SizeAtCompileTime!=Dynamic && ((static_cast<int>(sizeof(Scalar))*SizeAtCompileTime)%16)==0 };
-    EIGEN_MAKE_ALIGNED_OPERATOR_NEW_IF(NeedsToAlign)
-
     using Base::base;
     using Base::coeffRef;
 
@@ -412,25 +411,6 @@ EIGEN_MAKE_TYPEDEFS_ALL_SIZES(std::complex<double>, cd)
 
 #undef EIGEN_MAKE_TYPEDEFS_ALL_SIZES
 #undef EIGEN_MAKE_TYPEDEFS
-
-#undef EIGEN_MAKE_TYPEDEFS_LARGE
-
-#define EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, SizeSuffix) \
-using Eigen::Matrix##SizeSuffix##TypeSuffix; \
-using Eigen::Vector##SizeSuffix##TypeSuffix; \
-using Eigen::RowVector##SizeSuffix##TypeSuffix;
-
-#define EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(TypeSuffix) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, 2) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, 3) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, 4) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE_AND_SIZE(TypeSuffix, X) \
-
-#define EIGEN_USING_MATRIX_TYPEDEFS \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(i) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(f) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(d) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(cf) \
-EIGEN_USING_MATRIX_TYPEDEFS_FOR_TYPE(cd)
+#undef EIGEN_MAKE_FIXED_TYPEDEFS
 
 #endif // EIGEN_MATRIX_H

Eigen/src/Core/MatrixBase.h

@@ -38,7 +38,7 @@
   * Note that some methods are defined in other modules such as the \ref LU_Module LU module
   * for all functions related to matrix inversions.
   *
-  * \param Derived is the derived type, e.g. a matrix type, or an expression, etc.
+  * \tparam Derived is the derived type, e.g. a matrix type, or an expression, etc.
   *
   * When writing a function taking Eigen objects as argument, if you want your function
   * to take as argument any matrix, vector, or expression, just let it take a
@@ -53,6 +53,9 @@
     }
   * \endcode
   *
+  * This class can be extended with the help of the plugin mechanism described on the page
+  * \ref TopicCustomizingEigen by defining the preprocessor symbol \c EIGEN_MATRIXBASE_PLUGIN.
+  *
   * \sa \ref TopicClassHierarchy
   */
 template<typename Derived> class MatrixBase
@@ -108,7 +111,7 @@ template<typename Derived> class MatrixBase
 
     /** \returns the size of the main diagonal, which is min(rows(),cols()).
       * \sa rows(), cols(), SizeAtCompileTime. */
-    inline Index diagonalSize() const { return std::min(rows(),cols()); }
+    inline Index diagonalSize() const { return (std::min)(rows(),cols()); }
 
     /** \brief The plain matrix type corresponding to this expression.
       *
@@ -327,7 +330,7 @@ template<typename Derived> class MatrixBase
     /** \returns an \link ArrayBase Array \endlink expression of this matrix
       * \sa ArrayBase::matrix() */
     ArrayWrapper<Derived> array() { return derived(); }
-    const ArrayWrapper<Derived> array() const { return derived(); }
+    const ArrayWrapper<const Derived> array() const { return derived(); }
 
 /////////// LU module ///////////
 
@@ -462,6 +465,8 @@ template<typename Derived> class MatrixBase
     const MatrixFunctionReturnValue<Derived> sinh() const;
     const MatrixFunctionReturnValue<Derived> cos() const;
     const MatrixFunctionReturnValue<Derived> sin() const;
+    const MatrixSquareRootReturnValue<Derived> sqrt() const;
+    const MatrixLogarithmReturnValue<Derived> log() const;
 
 #ifdef EIGEN2_SUPPORT
     template<typename ProductDerived, typename Lhs, typename Rhs>
@@ -508,10 +513,10 @@ template<typename Derived> class MatrixBase
   protected:
     // mixing arrays and matrices is not legal
     template<typename OtherDerived> Derived& operator+=(const ArrayBase<OtherDerived>& )
-    {EIGEN_STATIC_ASSERT(sizeof(typename OtherDerived::Scalar)==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES);}
+    {EIGEN_STATIC_ASSERT(std::ptrdiff_t(sizeof(typename OtherDerived::Scalar))==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES); return *this;}
     // mixing arrays and matrices is not legal
     template<typename OtherDerived> Derived& operator-=(const ArrayBase<OtherDerived>& )
-    {EIGEN_STATIC_ASSERT(sizeof(typename OtherDerived::Scalar)==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES);}
+    {EIGEN_STATIC_ASSERT(std::ptrdiff_t(sizeof(typename OtherDerived::Scalar))==-1,YOU_CANNOT_MIX_ARRAYS_AND_MATRICES); return *this;}
 };
 
 #endif // EIGEN_MATRIXBASE_H

Eigen/src/Core/NumTraits.h

@@ -81,14 +81,14 @@ template<typename T> struct GenericNumTraits
                    >::type NonInteger;
   typedef T Nested;
 
-  inline static Real epsilon() { return std::numeric_limits<T>::epsilon(); }
-  inline static Real dummy_precision()
+  static inline Real epsilon() { return std::numeric_limits<T>::epsilon(); }
+  static inline Real dummy_precision()
   {
     // make sure to override this for floating-point types
     return Real(0);
   }
-  inline static T highest() { return std::numeric_limits<T>::max(); }
-  inline static T lowest()  { return IsInteger ? std::numeric_limits<T>::min() : (-std::numeric_limits<T>::max()); }
+  static inline T highest() { return (std::numeric_limits<T>::max)(); }
+  static inline T lowest()  { return IsInteger ? (std::numeric_limits<T>::min)() : (-(std::numeric_limits<T>::max)()); }
   
 #ifdef EIGEN2_SUPPORT
   enum {
@@ -104,12 +104,12 @@ template<typename T> struct NumTraits : GenericNumTraits<T>
 template<> struct NumTraits<float>
   : GenericNumTraits<float>
 {
-  inline static float dummy_precision() { return 1e-5f; }
+  static inline float dummy_precision() { return 1e-5f; }
 };
 
 template<> struct NumTraits<double> : GenericNumTraits<double>
 {
-  inline static double dummy_precision() { return 1e-12; }
+  static inline double dummy_precision() { return 1e-12; }
 };
 
 template<> struct NumTraits<long double>
@@ -130,8 +130,8 @@ template<typename _Real> struct NumTraits<std::complex<_Real> >
     MulCost = 4 * NumTraits<Real>::MulCost + 2 * NumTraits<Real>::AddCost
   };
 
-  inline static Real epsilon() { return NumTraits<Real>::epsilon(); }
-  inline static Real dummy_precision() { return NumTraits<Real>::dummy_precision(); }
+  static inline Real epsilon() { return NumTraits<Real>::epsilon(); }
+  static inline Real dummy_precision() { return NumTraits<Real>::dummy_precision(); }
 };
 
 template<typename Scalar, int Rows, int Cols, int Options, int MaxRows, int MaxCols>

Eigen/src/Core/PermutationMatrix.h

@@ -511,7 +511,7 @@ class PermutationWrapper : public PermutationBase<PermutationWrapper<_IndicesTyp
 
   protected:
 
-    const typename IndicesType::Nested m_indices;
+    typename IndicesType::Nested m_indices;
 };
 
 /** \returns the matrix with the permutation applied to the columns.
@@ -608,7 +608,7 @@ struct permut_matrix_product_retval
 
   protected:
     const PermutationType& m_permutation;
-    const typename MatrixType::Nested m_matrix;
+    typename MatrixType::Nested m_matrix;
 };
 
 /* Template partial specialization for transposed/inverse permutations */

Eigen/src/Core/PlainObjectBase.h

@@ -34,18 +34,56 @@
 
 namespace internal {
 
+template<typename Index>
+EIGEN_ALWAYS_INLINE void check_rows_cols_for_overflow(Index rows, Index cols)
+{
+  // http://hg.mozilla.org/mozilla-central/file/6c8a909977d3/xpcom/ds/CheckedInt.h#l242
+  // we assume Index is signed
+  Index max_index = (size_t(1) << (8 * sizeof(Index) - 1)) - 1; // assume Index is signed
+  bool error = (rows < 0  || cols < 0)  ? true
+             : (rows == 0 || cols == 0) ? false
+                                        : (rows > max_index / cols);
+  if (error)
+    throw_std_bad_alloc();
+}
+
 template <typename Derived, typename OtherDerived = Derived, bool IsVector = static_cast<bool>(Derived::IsVectorAtCompileTime)> struct conservative_resize_like_impl;
 
 template<typename MatrixTypeA, typename MatrixTypeB, bool SwapPointers> struct matrix_swap_impl;
 
 } // end namespace internal
 
-/**
+/** \class PlainObjectBase
   * \brief %Dense storage base class for matrices and arrays.
+  *
+  * This class can be extended with the help of the plugin mechanism described on the page
+  * \ref TopicCustomizingEigen by defining the preprocessor symbol \c EIGEN_PLAINOBJECTBASE_PLUGIN.
+  *
   * \sa \ref TopicClassHierarchy
   */
+#ifdef EIGEN_PARSED_BY_DOXYGEN
+namespace internal {
+
+// this is a warkaround to doxygen not being able to understand the inheritence logic
+// when it is hidden by the dense_xpr_base helper struct.
+template<typename Derived> struct dense_xpr_base_dispatcher_for_doxygen;// : public MatrixBase<Derived> {};
+/** This class is just a workaround for Doxygen and it does not not actually exist. */
+template<typename _Scalar, int _Rows, int _Cols, int _Options, int _MaxRows, int _MaxCols>
+struct dense_xpr_base_dispatcher_for_doxygen<Matrix<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols> >
+    : public MatrixBase<Matrix<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols> > {};
+/** This class is just a workaround for Doxygen and it does not not actually exist. */
+template<typename _Scalar, int _Rows, int _Cols, int _Options, int _MaxRows, int _MaxCols>
+struct dense_xpr_base_dispatcher_for_doxygen<Array<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols> >
+    : public ArrayBase<Array<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols> > {};
+
+} // namespace internal
+
+template<typename Derived>
+class PlainObjectBase : public internal::dense_xpr_base_dispatcher_for_doxygen<Derived>
+#else
 template<typename Derived>
 class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
+#endif
 {
   public:
     enum { Options = internal::traits<Derived>::Options };
@@ -80,14 +118,12 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
     template<typename StrideType> struct StridedConstMapType { typedef Eigen::Map<const Derived, Unaligned, StrideType> type; };
     template<typename StrideType> struct StridedAlignedMapType { typedef Eigen::Map<Derived, Aligned, StrideType> type; };
     template<typename StrideType> struct StridedConstAlignedMapType { typedef Eigen::Map<const Derived, Aligned, StrideType> type; };
-    
 
   protected:
     DenseStorage<Scalar, Base::MaxSizeAtCompileTime, Base::RowsAtCompileTime, Base::ColsAtCompileTime, Options> m_storage;
 
   public:
-    enum { NeedsToAlign = (!(Options&DontAlign))
-                          && SizeAtCompileTime!=Dynamic && ((static_cast<int>(sizeof(Scalar))*SizeAtCompileTime)%16)==0 };
+    enum { NeedsToAlign = SizeAtCompileTime != Dynamic && (internal::traits<Derived>::Flags & AlignedBit) != 0 };
     EIGEN_MAKE_ALIGNED_OPERATOR_NEW_IF(NeedsToAlign)
 
     Base& base() { return *static_cast<Base*>(this); }
@@ -196,11 +232,13 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
     EIGEN_STRONG_INLINE void resize(Index rows, Index cols)
     {
       #ifdef EIGEN_INITIALIZE_MATRICES_BY_ZERO
+        internal::check_rows_cols_for_overflow(rows, cols);
         Index size = rows*cols;
         bool size_changed = size != this->size();
         m_storage.resize(size, rows, cols);
         if(size_changed) EIGEN_INITIALIZE_BY_ZERO_IF_THAT_OPTION_IS_ENABLED
       #else
+        internal::check_rows_cols_for_overflow(rows, cols);
         m_storage.resize(rows*cols, rows, cols);
       #endif
     }
@@ -269,6 +307,7 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
     EIGEN_STRONG_INLINE void resizeLike(const EigenBase<OtherDerived>& _other)
     {
       const OtherDerived& other = _other.derived();
+      internal::check_rows_cols_for_overflow(other.rows(), other.cols());
       const Index othersize = other.rows()*other.cols();
       if(RowsAtCompileTime == 1)
       {
@@ -283,33 +322,47 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
       else resize(other.rows(), other.cols());
     }
 
-    /** Resizes \c *this to a \a rows x \a cols matrix while leaving old values of \c *this untouched.
+    /** Resizes the matrix to \a rows x \a cols while leaving old values untouched.
       *
-      * This method is intended for dynamic-size matrices. If you only want to change the number
-      * of rows and/or of columns, you can use conservativeResize(NoChange_t, Index),
+      * The method is intended for matrices of dynamic size. If you only want to change the number
+      * of rows and/or of columns, you can use conservativeResize(NoChange_t, Index) or
       * conservativeResize(Index, NoChange_t).
       *
-      * The top-left part of the resized matrix will be the same as the overlapping top-left corner
-      * of \c *this. In case values need to be appended to the matrix they will be uninitialized.
+      * Matrices are resized relative to the top-left element. In case values need to be 
+      * appended to the matrix they will be uninitialized.
       */
     EIGEN_STRONG_INLINE void conservativeResize(Index rows, Index cols)
     {
       internal::conservative_resize_like_impl<Derived>::run(*this, rows, cols);
     }
 
+    /** Resizes the matrix to \a rows x \a cols while leaving old values untouched.
+      *
+      * As opposed to conservativeResize(Index rows, Index cols), this version leaves
+      * the number of columns unchanged.
+      *
+      * In case the matrix is growing, new rows will be uninitialized.
+      */
     EIGEN_STRONG_INLINE void conservativeResize(Index rows, NoChange_t)
     {
       // Note: see the comment in conservativeResize(Index,Index)
       conservativeResize(rows, cols());
     }
 
+    /** Resizes the matrix to \a rows x \a cols while leaving old values untouched.
+      *
+      * As opposed to conservativeResize(Index rows, Index cols), this version leaves
+      * the number of rows unchanged.
+      *
+      * In case the matrix is growing, new columns will be uninitialized.
+      */
     EIGEN_STRONG_INLINE void conservativeResize(NoChange_t, Index cols)
     {
       // Note: see the comment in conservativeResize(Index,Index)
       conservativeResize(rows(), cols);
     }
 
-    /** Resizes \c *this to a vector of length \a size while retaining old values of *this.
+    /** Resizes the vector to \a size while retaining old values.
       *
       * \only_for_vectors. This method does not work for
       * partially dynamic matrices when the static dimension is anything other
@@ -322,6 +375,15 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
       internal::conservative_resize_like_impl<Derived>::run(*this, size);
     }
 
+    /** Resizes the matrix to \a rows x \a cols of \c other, while leaving old values untouched.
+      *
+      * The method is intended for matrices of dynamic size. If you only want to change the number
+      * of rows and/or of columns, you can use conservativeResize(NoChange_t, Index) or
+      * conservativeResize(Index, NoChange_t).
+      *
+      * Matrices are resized relative to the top-left element. In case values need to be 
+      * appended to the matrix they will copied from \c other.
+      */
     template<typename OtherDerived>
     EIGEN_STRONG_INLINE void conservativeResizeLike(const DenseBase<OtherDerived>& other)
     {
@@ -390,6 +452,7 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
       : m_storage(other.derived().rows() * other.derived().cols(), other.derived().rows(), other.derived().cols())
     {
       _check_template_params();
+      internal::check_rows_cols_for_overflow(other.derived().rows(), other.derived().cols());
       Base::operator=(other.derived());
     }
 
@@ -398,74 +461,71 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
       * while the AlignedMap() functions return aligned Map objects and thus should be called only with 16-byte-aligned
       * \a data pointers.
       *
-      * These methods do not allow to specify strides. If you need to specify strides, you have to
-      * use the Map class directly.
-      *
       * \see class Map
       */
     //@{
-    inline static ConstMapType Map(const Scalar* data)
+    static inline ConstMapType Map(const Scalar* data)
     { return ConstMapType(data); }
-    inline static MapType Map(Scalar* data)
+    static inline MapType Map(Scalar* data)
     { return MapType(data); }
-    inline static ConstMapType Map(const Scalar* data, Index size)
+    static inline ConstMapType Map(const Scalar* data, Index size)
     { return ConstMapType(data, size); }
-    inline static MapType Map(Scalar* data, Index size)
+    static inline MapType Map(Scalar* data, Index size)
     { return MapType(data, size); }
-    inline static ConstMapType Map(const Scalar* data, Index rows, Index cols)
+    static inline ConstMapType Map(const Scalar* data, Index rows, Index cols)
     { return ConstMapType(data, rows, cols); }
-    inline static MapType Map(Scalar* data, Index rows, Index cols)
+    static inline MapType Map(Scalar* data, Index rows, Index cols)
     { return MapType(data, rows, cols); }
 
-    inline static ConstAlignedMapType MapAligned(const Scalar* data)
+    static inline ConstAlignedMapType MapAligned(const Scalar* data)
     { return ConstAlignedMapType(data); }
-    inline static AlignedMapType MapAligned(Scalar* data)
+    static inline AlignedMapType MapAligned(Scalar* data)
     { return AlignedMapType(data); }
-    inline static ConstAlignedMapType MapAligned(const Scalar* data, Index size)
+    static inline ConstAlignedMapType MapAligned(const Scalar* data, Index size)
     { return ConstAlignedMapType(data, size); }
-    inline static AlignedMapType MapAligned(Scalar* data, Index size)
+    static inline AlignedMapType MapAligned(Scalar* data, Index size)
     { return AlignedMapType(data, size); }
-    inline static ConstAlignedMapType MapAligned(const Scalar* data, Index rows, Index cols)
+    static inline ConstAlignedMapType MapAligned(const Scalar* data, Index rows, Index cols)
     { return ConstAlignedMapType(data, rows, cols); }
-    inline static AlignedMapType MapAligned(Scalar* data, Index rows, Index cols)
+    static inline AlignedMapType MapAligned(Scalar* data, Index rows, Index cols)
     { return AlignedMapType(data, rows, cols); }
 
     template<int Outer, int Inner>
-    inline static typename StridedConstMapType<Stride<Outer, Inner> >::type Map(const Scalar* data, const Stride<Outer, Inner>& stride)
+    static inline typename StridedConstMapType<Stride<Outer, Inner> >::type Map(const Scalar* data, const Stride<Outer, Inner>& stride)
     { return typename StridedConstMapType<Stride<Outer, Inner> >::type(data, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedMapType<Stride<Outer, Inner> >::type Map(Scalar* data, const Stride<Outer, Inner>& stride)
+    static inline typename StridedMapType<Stride<Outer, Inner> >::type Map(Scalar* data, const Stride<Outer, Inner>& stride)
     { return typename StridedMapType<Stride<Outer, Inner> >::type(data, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedConstMapType<Stride<Outer, Inner> >::type Map(const Scalar* data, Index size, const Stride<Outer, Inner>& stride)
+    static inline typename StridedConstMapType<Stride<Outer, Inner> >::type Map(const Scalar* data, Index size, const Stride<Outer, Inner>& stride)
     { return typename StridedConstMapType<Stride<Outer, Inner> >::type(data, size, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedMapType<Stride<Outer, Inner> >::type Map(Scalar* data, Index size, const Stride<Outer, Inner>& stride)
+    static inline typename StridedMapType<Stride<Outer, Inner> >::type Map(Scalar* data, Index size, const Stride<Outer, Inner>& stride)
     { return typename StridedMapType<Stride<Outer, Inner> >::type(data, size, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedConstMapType<Stride<Outer, Inner> >::type Map(const Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
+    static inline typename StridedConstMapType<Stride<Outer, Inner> >::type Map(const Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
     { return typename StridedConstMapType<Stride<Outer, Inner> >::type(data, rows, cols, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedMapType<Stride<Outer, Inner> >::type Map(Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
+    static inline typename StridedMapType<Stride<Outer, Inner> >::type Map(Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
     { return typename StridedMapType<Stride<Outer, Inner> >::type(data, rows, cols, stride); }
 
     template<int Outer, int Inner>
-    inline static typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type MapAligned(const Scalar* data, const Stride<Outer, Inner>& stride)
+    static inline typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type MapAligned(const Scalar* data, const Stride<Outer, Inner>& stride)
     { return typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type(data, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedAlignedMapType<Stride<Outer, Inner> >::type MapAligned(Scalar* data, const Stride<Outer, Inner>& stride)
+    static inline typename StridedAlignedMapType<Stride<Outer, Inner> >::type MapAligned(Scalar* data, const Stride<Outer, Inner>& stride)
     { return typename StridedAlignedMapType<Stride<Outer, Inner> >::type(data, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type MapAligned(const Scalar* data, Index size, const Stride<Outer, Inner>& stride)
+    static inline typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type MapAligned(const Scalar* data, Index size, const Stride<Outer, Inner>& stride)
     { return typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type(data, size, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedAlignedMapType<Stride<Outer, Inner> >::type MapAligned(Scalar* data, Index size, const Stride<Outer, Inner>& stride)
+    static inline typename StridedAlignedMapType<Stride<Outer, Inner> >::type MapAligned(Scalar* data, Index size, const Stride<Outer, Inner>& stride)
     { return typename StridedAlignedMapType<Stride<Outer, Inner> >::type(data, size, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type MapAligned(const Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
+    static inline typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type MapAligned(const Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
     { return typename StridedConstAlignedMapType<Stride<Outer, Inner> >::type(data, rows, cols, stride); }
     template<int Outer, int Inner>
-    inline static typename StridedAlignedMapType<Stride<Outer, Inner> >::type MapAligned(Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
+    static inline typename StridedAlignedMapType<Stride<Outer, Inner> >::type MapAligned(Scalar* data, Index rows, Index cols, const Stride<Outer, Inner>& stride)
     { return typename StridedAlignedMapType<Stride<Outer, Inner> >::type(data, rows, cols, stride); }
     //@}
 
@@ -555,8 +615,12 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
     template<typename T0, typename T1>
     EIGEN_STRONG_INLINE void _init2(Index rows, Index cols, typename internal::enable_if<Base::SizeAtCompileTime!=2,T0>::type* = 0)
     {
+      EIGEN_STATIC_ASSERT(bool(NumTraits<T0>::IsInteger) &&
+                          bool(NumTraits<T1>::IsInteger),
+                          FLOATING_POINT_ARGUMENT_PASSED__INTEGER_WAS_EXPECTED)
       eigen_assert(rows >= 0 && (RowsAtCompileTime == Dynamic || RowsAtCompileTime == rows)
              && cols >= 0 && (ColsAtCompileTime == Dynamic || ColsAtCompileTime == cols));
+      internal::check_rows_cols_for_overflow(rows, cols);      
       m_storage.resize(rows*cols,rows,cols);
       EIGEN_INITIALIZE_BY_ZERO_IF_THAT_OPTION_IS_ENABLED
     }
@@ -583,7 +647,7 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
 
   public:
 #ifndef EIGEN_PARSED_BY_DOXYGEN
-    EIGEN_STRONG_INLINE static void _check_template_params()
+    static EIGEN_STRONG_INLINE void _check_template_params()
     {
       EIGEN_STATIC_ASSERT((EIGEN_IMPLIES(MaxRowsAtCompileTime==1 && MaxColsAtCompileTime!=1, (Options&RowMajor)==RowMajor)
                         && EIGEN_IMPLIES(MaxColsAtCompileTime==1 && MaxRowsAtCompileTime!=1, (Options&RowMajor)==0)
@@ -614,14 +678,15 @@ struct internal::conservative_resize_like_impl
     if ( ( Derived::IsRowMajor && _this.cols() == cols) || // row-major and we change only the number of rows
          (!Derived::IsRowMajor && _this.rows() == rows) )  // column-major and we change only the number of columns
     {
+      internal::check_rows_cols_for_overflow(rows, cols);
       _this.derived().m_storage.conservativeResize(rows*cols,rows,cols);
     }
     else
     {
       // The storage order does not allow us to use reallocation.
       typename Derived::PlainObject tmp(rows,cols);
-      const Index common_rows = std::min(rows, _this.rows());
-      const Index common_cols = std::min(cols, _this.cols());
+      const Index common_rows = (std::min)(rows, _this.rows());
+      const Index common_cols = (std::min)(cols, _this.cols());
       tmp.block(0,0,common_rows,common_cols) = _this.block(0,0,common_rows,common_cols);
       _this.derived().swap(tmp);
     }
@@ -654,8 +719,8 @@ struct internal::conservative_resize_like_impl
     {
       // The storage order does not allow us to use reallocation.
       typename Derived::PlainObject tmp(other);
-      const Index common_rows = std::min(tmp.rows(), _this.rows());
-      const Index common_cols = std::min(tmp.cols(), _this.cols());
+      const Index common_rows = (std::min)(tmp.rows(), _this.rows());
+      const Index common_cols = (std::min)(tmp.cols(), _this.cols());
       tmp.block(0,0,common_rows,common_cols) = _this.block(0,0,common_rows,common_cols);
       _this.derived().swap(tmp);
     }

Eigen/src/Core/Product.h

@@ -1,8 +1,7 @@
 // This file is part of Eigen, a lightweight C++ template library
 // for linear algebra.
 //
-// Copyright (C) 2006-2008 Benoit Jacob <jacob.benoit.1@gmail.com>
-// Copyright (C) 2008 Gael Guennebaud <gael.guennebaud@inria.fr>
+// Copyright (C) 2008-2011 Gael Guennebaud <gael.guennebaud@inria.fr>
 //
 // Eigen is free software; you can redistribute it and/or
 // modify it under the terms of the GNU Lesser General Public
@@ -26,603 +25,89 @@
 #ifndef EIGEN_PRODUCT_H
 #define EIGEN_PRODUCT_H
 
-/** \class GeneralProduct
+template<typename Lhs, typename Rhs> class Product;
+template<typename Lhs, typename Rhs, typename StorageKind> class ProductImpl;
+
+/** \class Product
   * \ingroup Core_Module
   *
-  * \brief Expression of the product of two general matrices or vectors
+  * \brief Expression of the product of two arbitrary matrices or vectors
   *
-  * \param LhsNested the type used to store the left-hand side
-  * \param RhsNested the type used to store the right-hand side
-  * \param ProductMode the type of the product
+  * \param Lhs the type of the left-hand side expression
+  * \param Rhs the type of the right-hand side expression
   *
-  * This class represents an expression of the product of two general matrices.
-  * We call a general matrix, a dense matrix with full storage. For instance,
-  * This excludes triangular, selfadjoint, and sparse matrices.
-  * It is the return type of the operator* between general matrices. Its template
-  * arguments are determined automatically by ProductReturnType. Therefore,
-  * GeneralProduct should never be used direclty. To determine the result type of a
-  * function which involves a matrix product, use ProductReturnType::Type.
+  * This class represents an expression of the product of two arbitrary matrices.
   *
-  * \sa ProductReturnType, MatrixBase::operator*(const MatrixBase<OtherDerived>&)
   */
-template<typename Lhs, typename Rhs, int ProductType = internal::product_type<Lhs,Rhs>::value>
-class GeneralProduct;
-
-enum {
-  Large = 2,
-  Small = 3
-};
 
 namespace internal {
-
-template<int Rows, int Cols, int Depth> struct product_type_selector;
-
-template<int Size, int MaxSize> struct product_size_category
+template<typename Lhs, typename Rhs>
+struct traits<Product<Lhs, Rhs> >
 {
-  enum { is_large = MaxSize == Dynamic ||
-                    Size >= EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD,
-         value = is_large  ? Large
-               : Size == 1 ? 1
-                           : Small
+  typedef MatrixXpr XprKind;
+  typedef typename remove_all<Lhs>::type LhsCleaned;
+  typedef typename remove_all<Rhs>::type RhsCleaned;
+  typedef typename scalar_product_traits<typename traits<LhsCleaned>::Scalar, typename traits<RhsCleaned>::Scalar>::ReturnType Scalar;
+  typedef typename promote_storage_type<typename traits<LhsCleaned>::StorageKind,
+                                        typename traits<RhsCleaned>::StorageKind>::ret StorageKind;
+  typedef typename promote_index_type<typename traits<LhsCleaned>::Index,
+                                      typename traits<RhsCleaned>::Index>::type Index;
+  enum {
+    RowsAtCompileTime = LhsCleaned::RowsAtCompileTime,
+    ColsAtCompileTime = RhsCleaned::ColsAtCompileTime,
+    MaxRowsAtCompileTime = LhsCleaned::MaxRowsAtCompileTime,
+    MaxColsAtCompileTime = RhsCleaned::MaxColsAtCompileTime,
+    Flags = (MaxRowsAtCompileTime==1 ? RowMajorBit : 0), // TODO should be no storage order
+    CoeffReadCost = 0 // TODO CoeffReadCost should not be part of the expression traits
   };
 };
-
-template<typename Lhs, typename Rhs> struct product_type
-{
-  typedef typename remove_all<Lhs>::type _Lhs;
-  typedef typename remove_all<Rhs>::type _Rhs;
-  enum {
-    MaxRows  = _Lhs::MaxRowsAtCompileTime,
-    Rows  = _Lhs::RowsAtCompileTime,
-    MaxCols  = _Rhs::MaxColsAtCompileTime,
-    Cols  = _Rhs::ColsAtCompileTime,
-    MaxDepth = EIGEN_SIZE_MIN_PREFER_FIXED(_Lhs::MaxColsAtCompileTime,
-                                           _Rhs::MaxRowsAtCompileTime),
-    Depth = EIGEN_SIZE_MIN_PREFER_FIXED(_Lhs::ColsAtCompileTime,
-                                        _Rhs::RowsAtCompileTime),
-    LargeThreshold = EIGEN_CACHEFRIENDLY_PRODUCT_THRESHOLD
-  };
-
-  // the splitting into different lines of code here, introducing the _select enums and the typedef below,
-  // is to work around an internal compiler error with gcc 4.1 and 4.2.
-private:
-  enum {
-    rows_select = product_size_category<Rows,MaxRows>::value,
-    cols_select = product_size_category<Cols,MaxCols>::value,
-    depth_select = product_size_category<Depth,MaxDepth>::value
-  };
-  typedef product_type_selector<rows_select, cols_select, depth_select> selector;
-
-public:
-  enum {
-    value = selector::ret
-  };
-#ifdef EIGEN_DEBUG_PRODUCT
-  static void debug()
-  {
-      EIGEN_DEBUG_VAR(Rows);
-      EIGEN_DEBUG_VAR(Cols);
-      EIGEN_DEBUG_VAR(Depth);
-      EIGEN_DEBUG_VAR(rows_select);
-      EIGEN_DEBUG_VAR(cols_select);
-      EIGEN_DEBUG_VAR(depth_select);
-      EIGEN_DEBUG_VAR(value);
-  }
-#endif
-};
-
-
-/* The following allows to select the kind of product at compile time
- * based on the three dimensions of the product.
- * This is a compile time mapping from {1,Small,Large}^3 -> {product types} */
-// FIXME I'm not sure the current mapping is the ideal one.
-template<int M, int N>  struct product_type_selector<M,N,1>              { enum { ret = OuterProduct }; };
-template<int Depth>     struct product_type_selector<1,    1,    Depth>  { enum { ret = InnerProduct }; };
-template<>              struct product_type_selector<1,    1,    1>      { enum { ret = InnerProduct }; };
-template<>              struct product_type_selector<Small,1,    Small>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<1,    Small,Small>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<Small,Small,Small>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<Small, Small, 1>    { enum { ret = LazyCoeffBasedProductMode }; };
-template<>              struct product_type_selector<Small, Large, 1>    { enum { ret = LazyCoeffBasedProductMode }; };
-template<>              struct product_type_selector<Large, Small, 1>    { enum { ret = LazyCoeffBasedProductMode }; };
-template<>              struct product_type_selector<1,    Large,Small>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<1,    Large,Large>  { enum { ret = GemvProduct }; };
-template<>              struct product_type_selector<1,    Small,Large>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<Large,1,    Small>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<Large,1,    Large>  { enum { ret = GemvProduct }; };
-template<>              struct product_type_selector<Small,1,    Large>  { enum { ret = CoeffBasedProductMode }; };
-template<>              struct product_type_selector<Small,Small,Large>  { enum { ret = GemmProduct }; };
-template<>              struct product_type_selector<Large,Small,Large>  { enum { ret = GemmProduct }; };
-template<>              struct product_type_selector<Small,Large,Large>  { enum { ret = GemmProduct }; };
-template<>              struct product_type_selector<Large,Large,Large>  { enum { ret = GemmProduct }; };
-template<>              struct product_type_selector<Large,Small,Small>  { enum { ret = GemmProduct }; };
-template<>              struct product_type_selector<Small,Large,Small>  { enum { ret = GemmProduct }; };
-template<>              struct product_type_selector<Large,Large,Small>  { enum { ret = GemmProduct }; };
-
 } // end namespace internal
 
-/** \class ProductReturnType
-  * \ingroup Core_Module
-  *
-  * \brief Helper class to get the correct and optimized returned type of operator*
-  *
-  * \param Lhs the type of the left-hand side
-  * \param Rhs the type of the right-hand side
-  * \param ProductMode the type of the product (determined automatically by internal::product_mode)
-  *
-  * This class defines the typename Type representing the optimized product expression
-  * between two matrix expressions. In practice, using ProductReturnType<Lhs,Rhs>::Type
-  * is the recommended way to define the result type of a function returning an expression
-  * which involve a matrix product. The class Product should never be
-  * used directly.
-  *
-  * \sa class Product, MatrixBase::operator*(const MatrixBase<OtherDerived>&)
-  */
-template<typename Lhs, typename Rhs, int ProductType>
-struct ProductReturnType
-{
-  // TODO use the nested type to reduce instanciations ????
-//   typedef typename internal::nested<Lhs,Rhs::ColsAtCompileTime>::type LhsNested;
-//   typedef typename internal::nested<Rhs,Lhs::RowsAtCompileTime>::type RhsNested;
-
-  typedef GeneralProduct<Lhs/*Nested*/, Rhs/*Nested*/, ProductType> Type;
-};
 
 template<typename Lhs, typename Rhs>
-struct ProductReturnType<Lhs,Rhs,CoeffBasedProductMode>
-{
-  typedef typename internal::nested<Lhs, Rhs::ColsAtCompileTime, typename internal::plain_matrix_type<Lhs>::type >::type LhsNested;
-  typedef typename internal::nested<Rhs, Lhs::RowsAtCompileTime, typename internal::plain_matrix_type<Rhs>::type >::type RhsNested;
-  typedef CoeffBasedProduct<LhsNested, RhsNested, EvalBeforeAssigningBit | EvalBeforeNestingBit> Type;
-};
-
-template<typename Lhs, typename Rhs>
-struct ProductReturnType<Lhs,Rhs,LazyCoeffBasedProductMode>
-{
-  typedef typename internal::nested<Lhs, Rhs::ColsAtCompileTime, typename internal::plain_matrix_type<Lhs>::type >::type LhsNested;
-  typedef typename internal::nested<Rhs, Lhs::RowsAtCompileTime, typename internal::plain_matrix_type<Rhs>::type >::type RhsNested;
-  typedef CoeffBasedProduct<LhsNested, RhsNested, NestByRefBit> Type;
-};
-
-// this is a workaround for sun CC
-template<typename Lhs, typename Rhs>
-struct LazyProductReturnType : public ProductReturnType<Lhs,Rhs,LazyCoeffBasedProductMode>
-{};
-
-/***********************************************************************
-*  Implementation of Inner Vector Vector Product
-***********************************************************************/
-
-// FIXME : maybe the "inner product" could return a Scalar
-// instead of a 1x1 matrix ??
-// Pro: more natural for the user
-// Cons: this could be a problem if in a meta unrolled algorithm a matrix-matrix
-// product ends up to a row-vector times col-vector product... To tackle this use
-// case, we could have a specialization for Block<MatrixType,1,1> with: operator=(Scalar x);
-
-namespace internal {
-
-template<typename Lhs, typename Rhs>
-struct traits<GeneralProduct<Lhs,Rhs,InnerProduct> >
- : traits<Matrix<typename scalar_product_traits<typename Lhs::Scalar, typename Rhs::Scalar>::ReturnType,1,1> >
-{};
-
-}
-
-template<typename Lhs, typename Rhs>
-class GeneralProduct<Lhs, Rhs, InnerProduct>
-  : internal::no_assignment_operator,
-    public Matrix<typename internal::scalar_product_traits<typename Lhs::Scalar, typename Rhs::Scalar>::ReturnType,1,1>
-{
-    typedef Matrix<typename internal::scalar_product_traits<typename Lhs::Scalar, typename Rhs::Scalar>::ReturnType,1,1> Base;
-  public:
-    GeneralProduct(const Lhs& lhs, const Rhs& rhs)
-    {
-      EIGEN_STATIC_ASSERT((internal::is_same<typename Lhs::RealScalar, typename Rhs::RealScalar>::value),
-        YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
-
-      Base::coeffRef(0,0) = (lhs.transpose().cwiseProduct(rhs)).sum();
-    }
-
-    /** Convertion to scalar */
-    operator const typename Base::Scalar() const {
-      return Base::coeff(0,0);
-    }
-};
-
-/***********************************************************************
-*  Implementation of Outer Vector Vector Product
-***********************************************************************/
-
-namespace internal {
-template<int StorageOrder> struct outer_product_selector;
-
-template<typename Lhs, typename Rhs>
-struct traits<GeneralProduct<Lhs,Rhs,OuterProduct> >
- : traits<ProductBase<GeneralProduct<Lhs,Rhs,OuterProduct>, Lhs, Rhs> >
-{};
-
-}
-
-template<typename Lhs, typename Rhs>
-class GeneralProduct<Lhs, Rhs, OuterProduct>
-  : public ProductBase<GeneralProduct<Lhs,Rhs,OuterProduct>, Lhs, Rhs>
+class Product : public ProductImpl<Lhs,Rhs,typename internal::promote_storage_type<typename internal::traits<Lhs>::StorageKind,
+                                                                            typename internal::traits<Rhs>::StorageKind>::ret>
 {
   public:
-    EIGEN_PRODUCT_PUBLIC_INTERFACE(GeneralProduct)
-
-    GeneralProduct(const Lhs& lhs, const Rhs& rhs) : Base(lhs,rhs)
-    {
-      EIGEN_STATIC_ASSERT((internal::is_same<typename Lhs::RealScalar, typename Rhs::RealScalar>::value),
-        YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
-    }
-
-    template<typename Dest> void scaleAndAddTo(Dest& dest, Scalar alpha) const
-    {
-      internal::outer_product_selector<(int(Dest::Flags)&RowMajorBit) ? RowMajor : ColMajor>::run(*this, dest, alpha);
-    }
-};
-
-namespace internal {
-
-template<> struct outer_product_selector<ColMajor> {
-  template<typename ProductType, typename Dest>
-  static EIGEN_DONT_INLINE void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha) {
-    typedef typename Dest::Index Index;
-    // FIXME make sure lhs is sequentially stored
-    // FIXME not very good if rhs is real and lhs complex while alpha is real too
-    const Index cols = dest.cols();
-    for (Index j=0; j<cols; ++j)
-      dest.col(j) += (alpha * prod.rhs().coeff(j)) * prod.lhs();
-  }
-};
-
-template<> struct outer_product_selector<RowMajor> {
-  template<typename ProductType, typename Dest>
-  static EIGEN_DONT_INLINE void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha) {
-    typedef typename Dest::Index Index;
-    // FIXME make sure rhs is sequentially stored
-    // FIXME not very good if lhs is real and rhs complex while alpha is real too
-    const Index rows = dest.rows();
-    for (Index i=0; i<rows; ++i)
-      dest.row(i) += (alpha * prod.lhs().coeff(i)) * prod.rhs();
-  }
-};
-
-} // end namespace internal
-
-/***********************************************************************
-*  Implementation of General Matrix Vector Product
-***********************************************************************/
-
-/*  According to the shape/flags of the matrix we have to distinghish 3 different cases:
- *   1 - the matrix is col-major, BLAS compatible and M is large => call fast BLAS-like colmajor routine
- *   2 - the matrix is row-major, BLAS compatible and N is large => call fast BLAS-like rowmajor routine
- *   3 - all other cases are handled using a simple loop along the outer-storage direction.
- *  Therefore we need a lower level meta selector.
- *  Furthermore, if the matrix is the rhs, then the product has to be transposed.
- */
-namespace internal {
-
-template<typename Lhs, typename Rhs>
-struct traits<GeneralProduct<Lhs,Rhs,GemvProduct> >
- : traits<ProductBase<GeneralProduct<Lhs,Rhs,GemvProduct>, Lhs, Rhs> >
-{};
-
-template<int Side, int StorageOrder, bool BlasCompatible>
-struct gemv_selector;
-
-} // end namespace internal
-
-template<typename Lhs, typename Rhs>
-class GeneralProduct<Lhs, Rhs, GemvProduct>
-  : public ProductBase<GeneralProduct<Lhs,Rhs,GemvProduct>, Lhs, Rhs>
-{
-  public:
-    EIGEN_PRODUCT_PUBLIC_INTERFACE(GeneralProduct)
-
-    typedef typename Lhs::Scalar LhsScalar;
-    typedef typename Rhs::Scalar RhsScalar;
-
-    GeneralProduct(const Lhs& lhs, const Rhs& rhs) : Base(lhs,rhs)
-    {
-//       EIGEN_STATIC_ASSERT((internal::is_same<typename Lhs::Scalar, typename Rhs::Scalar>::value),
-//         YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
-    }
-
-    enum { Side = Lhs::IsVectorAtCompileTime ? OnTheLeft : OnTheRight };
-    typedef typename internal::conditional<int(Side)==OnTheRight,_LhsNested,_RhsNested>::type MatrixType;
-
-    template<typename Dest> void scaleAndAddTo(Dest& dst, Scalar alpha) const
-    {
-      eigen_assert(m_lhs.rows() == dst.rows() && m_rhs.cols() == dst.cols());
-      internal::gemv_selector<Side,(int(MatrixType::Flags)&RowMajorBit) ? RowMajor : ColMajor,
-                       bool(internal::blas_traits<MatrixType>::HasUsableDirectAccess)>::run(*this, dst, alpha);
-    }
-};
-
-namespace internal {
-
-// The vector is on the left => transposition
-template<int StorageOrder, bool BlasCompatible>
-struct gemv_selector<OnTheLeft,StorageOrder,BlasCompatible>
-{
-  template<typename ProductType, typename Dest>
-  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
-  {
-    Transpose<Dest> destT(dest);
-    enum { OtherStorageOrder = StorageOrder == RowMajor ? ColMajor : RowMajor };
-    gemv_selector<OnTheRight,OtherStorageOrder,BlasCompatible>
-      ::run(GeneralProduct<Transpose<const typename ProductType::_RhsNested>,Transpose<const typename ProductType::_LhsNested>, GemvProduct>
-        (prod.rhs().transpose(), prod.lhs().transpose()), destT, alpha);
-  }
-};
-
-template<typename Scalar,int Size,int MaxSize,bool Cond> struct gemv_static_vector_if;
-
-template<typename Scalar,int Size,int MaxSize>
-struct gemv_static_vector_if<Scalar,Size,MaxSize,false>
-{
-  EIGEN_STRONG_INLINE  Scalar* data() { eigen_internal_assert(false && "should never be called"); return 0; }
-};
-
-template<typename Scalar,int Size>
-struct gemv_static_vector_if<Scalar,Size,Dynamic,true>
-{
-  EIGEN_STRONG_INLINE Scalar* data() { return 0; }
-};
-
-template<typename Scalar,int Size,int MaxSize>
-struct gemv_static_vector_if<Scalar,Size,MaxSize,true>
-{
-  internal::plain_array<Scalar,EIGEN_SIZE_MIN_PREFER_FIXED(Size,MaxSize),0> m_data;
-  EIGEN_STRONG_INLINE Scalar* data() { return m_data.array; }
-};
-
-template<> struct gemv_selector<OnTheRight,ColMajor,true>
-{
-  template<typename ProductType, typename Dest>
-  static inline void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
-  {
-    typedef typename ProductType::Index Index;
-    typedef typename ProductType::LhsScalar   LhsScalar;
-    typedef typename ProductType::RhsScalar   RhsScalar;
-    typedef typename ProductType::Scalar      ResScalar;
-    typedef typename ProductType::RealScalar  RealScalar;
-    typedef typename ProductType::ActualLhsType ActualLhsType;
-    typedef typename ProductType::ActualRhsType ActualRhsType;
-    typedef typename ProductType::LhsBlasTraits LhsBlasTraits;
-    typedef typename ProductType::RhsBlasTraits RhsBlasTraits;
-    typedef Map<Matrix<ResScalar,Dynamic,1>, Aligned> MappedDest;
-
-    const ActualLhsType actualLhs = LhsBlasTraits::extract(prod.lhs());
-    const ActualRhsType actualRhs = RhsBlasTraits::extract(prod.rhs());
-
-    ResScalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(prod.lhs())
-                                  * RhsBlasTraits::extractScalarFactor(prod.rhs());
-
-    enum {
-      // FIXME find a way to allow an inner stride on the result if packet_traits<Scalar>::size==1
-      // on, the other hand it is good for the cache to pack the vector anyways...
-      EvalToDestAtCompileTime = Dest::InnerStrideAtCompileTime==1,
-      ComplexByReal = (NumTraits<LhsScalar>::IsComplex) && (!NumTraits<RhsScalar>::IsComplex),
-      MightCannotUseDest = (Dest::InnerStrideAtCompileTime!=1) || ComplexByReal
-    };
-
-    gemv_static_vector_if<ResScalar,Dest::SizeAtCompileTime,Dest::MaxSizeAtCompileTime,MightCannotUseDest> static_dest;
-
-    bool alphaIsCompatible = (!ComplexByReal) || (imag(actualAlpha)==RealScalar(0));
-    bool evalToDest = EvalToDestAtCompileTime && alphaIsCompatible;
     
-    RhsScalar compatibleAlpha = get_factor<ResScalar,RhsScalar>::run(actualAlpha);
+    typedef typename ProductImpl<
+        Lhs, Rhs,
+        typename internal::promote_storage_type<typename Lhs::StorageKind,
+                                                typename Rhs::StorageKind>::ret>::Base Base;
+    EIGEN_GENERIC_PUBLIC_INTERFACE(Product)
 
-    ResScalar* actualDestPtr;
-    bool freeDestPtr = false;
-    if (evalToDest)
+    typedef typename Lhs::Nested LhsNested;
+    typedef typename Rhs::Nested RhsNested;
+    typedef typename internal::remove_all<LhsNested>::type LhsNestedCleaned;
+    typedef typename internal::remove_all<RhsNested>::type RhsNestedCleaned;
+
+    Product(const Lhs& lhs, const Rhs& rhs) : m_lhs(lhs), m_rhs(rhs)
     {
-      actualDestPtr = &dest.coeffRef(0);
-    }
-    else
-    {
-      #ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
-      int size = dest.size();
-      EIGEN_DENSE_STORAGE_CTOR_PLUGIN
-      #endif
-      if((actualDestPtr = static_dest.data())==0)
-      {
-        freeDestPtr = true;
-        actualDestPtr = ei_aligned_stack_new(ResScalar,dest.size());
-      }
-      if(!alphaIsCompatible)
-      {
-        MappedDest(actualDestPtr, dest.size()).setZero();
-        compatibleAlpha = RhsScalar(1);
-      }
-      else
-        MappedDest(actualDestPtr, dest.size()) = dest;
+      eigen_assert(lhs.cols() == rhs.rows()
+        && "invalid matrix product"
+        && "if you wanted a coeff-wise or a dot product use the respective explicit functions");
     }
 
-    general_matrix_vector_product
-      <Index,LhsScalar,ColMajor,LhsBlasTraits::NeedToConjugate,RhsScalar,RhsBlasTraits::NeedToConjugate>::run(
-        actualLhs.rows(), actualLhs.cols(),
-        &actualLhs.coeffRef(0,0), actualLhs.outerStride(),
-        actualRhs.data(), actualRhs.innerStride(),
-        actualDestPtr, 1,
-        compatibleAlpha);
+    inline Index rows() const { return m_lhs.rows(); }
+    inline Index cols() const { return m_rhs.cols(); }
 
-    if (!evalToDest)
-    {
-      if(!alphaIsCompatible)
-        dest += actualAlpha * MappedDest(actualDestPtr, dest.size());
-      else
-        dest = MappedDest(actualDestPtr, dest.size());
-      if(freeDestPtr) ei_aligned_stack_delete(ResScalar, actualDestPtr, dest.size());
-    }
-  }
+    const LhsNestedCleaned& lhs() const { return m_lhs; }
+    const RhsNestedCleaned& rhs() const { return m_rhs; }
+
+  protected:
+
+    const LhsNested m_lhs;
+    const RhsNested m_rhs;
 };
 
-template<> struct gemv_selector<OnTheRight,RowMajor,true>
+template<typename Lhs, typename Rhs>
+class ProductImpl<Lhs,Rhs,Dense> : public internal::dense_xpr_base<Product<Lhs,Rhs> >::type
 {
-  template<typename ProductType, typename Dest>
-  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
-  {
-    typedef typename ProductType::LhsScalar LhsScalar;
-    typedef typename ProductType::RhsScalar RhsScalar;
-    typedef typename ProductType::Scalar    ResScalar;
-    typedef typename ProductType::Index Index;
-    typedef typename ProductType::ActualLhsType ActualLhsType;
-    typedef typename ProductType::ActualRhsType ActualRhsType;
-    typedef typename ProductType::_ActualRhsType _ActualRhsType;
-    typedef typename ProductType::LhsBlasTraits LhsBlasTraits;
-    typedef typename ProductType::RhsBlasTraits RhsBlasTraits;
+    typedef Product<Lhs, Rhs> Derived;
+  public:
 
-    typename add_const<ActualLhsType>::type actualLhs = LhsBlasTraits::extract(prod.lhs());
-    typename add_const<ActualRhsType>::type actualRhs = RhsBlasTraits::extract(prod.rhs());
-
-    ResScalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(prod.lhs())
-                                  * RhsBlasTraits::extractScalarFactor(prod.rhs());
-
-    enum {
-      // FIXME find a way to allow an inner stride on the result if packet_traits<Scalar>::size==1
-      // on, the other hand it is good for the cache to pack the vector anyways...
-      DirectlyUseRhs = _ActualRhsType::InnerStrideAtCompileTime==1
-    };
-
-    gemv_static_vector_if<RhsScalar,_ActualRhsType::SizeAtCompileTime,_ActualRhsType::MaxSizeAtCompileTime,!DirectlyUseRhs> static_rhs;
-
-    RhsScalar* actualRhsPtr;
-    bool freeRhsPtr = false;
-    if (DirectlyUseRhs)
-    {
-      actualRhsPtr = const_cast<RhsScalar*>(&actualRhs.coeffRef(0));
-    }
-    else
-    {
-      #ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
-      int size = actualRhs.size();
-      EIGEN_DENSE_STORAGE_CTOR_PLUGIN
-      #endif
-      if((actualRhsPtr = static_rhs.data())==0)
-      {
-        freeRhsPtr = true;
-        actualRhsPtr = ei_aligned_stack_new(RhsScalar, actualRhs.size());
-      }
-      Map<typename _ActualRhsType::PlainObject>(actualRhsPtr, actualRhs.size()) = actualRhs;
-    }
-
-    general_matrix_vector_product
-      <Index,LhsScalar,RowMajor,LhsBlasTraits::NeedToConjugate,RhsScalar,RhsBlasTraits::NeedToConjugate>::run(
-        actualLhs.rows(), actualLhs.cols(),
-        &actualLhs.coeffRef(0,0), actualLhs.outerStride(),
-        actualRhsPtr, 1,
-        &dest.coeffRef(0,0), dest.innerStride(),
-        actualAlpha);
-
-    if((!DirectlyUseRhs) && freeRhsPtr) ei_aligned_stack_delete(RhsScalar, actualRhsPtr, prod.rhs().size());
-  }
+    typedef typename internal::dense_xpr_base<Product<Lhs, Rhs> >::type Base;
+    EIGEN_DENSE_PUBLIC_INTERFACE(Derived)
 };
 
-template<> struct gemv_selector<OnTheRight,ColMajor,false>
-{
-  template<typename ProductType, typename Dest>
-  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
-  {
-    typedef typename Dest::Index Index;
-    // TODO makes sure dest is sequentially stored in memory, otherwise use a temp
-    const Index size = prod.rhs().rows();
-    for(Index k=0; k<size; ++k)
-      dest += (alpha*prod.rhs().coeff(k)) * prod.lhs().col(k);
-  }
-};
-
-template<> struct gemv_selector<OnTheRight,RowMajor,false>
-{
-  template<typename ProductType, typename Dest>
-  static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
-  {
-    typedef typename Dest::Index Index;
-    // TODO makes sure rhs is sequentially stored in memory, otherwise use a temp
-    const Index rows = prod.rows();
-    for(Index i=0; i<rows; ++i)
-      dest.coeffRef(i) += alpha * (prod.lhs().row(i).cwiseProduct(prod.rhs().transpose())).sum();
-  }
-};
-
-} // end namespace internal
-
-/***************************************************************************
-* Implementation of matrix base methods
-***************************************************************************/
-
-/** \returns the matrix product of \c *this and \a other.
-  *
-  * \note If instead of the matrix product you want the coefficient-wise product, see Cwise::operator*().
-  *
-  * \sa lazyProduct(), operator*=(const MatrixBase&), Cwise::operator*()
-  */
-template<typename Derived>
-template<typename OtherDerived>
-inline const typename ProductReturnType<Derived,OtherDerived>::Type
-MatrixBase<Derived>::operator*(const MatrixBase<OtherDerived> &other) const
-{
-  // A note regarding the function declaration: In MSVC, this function will sometimes
-  // not be inlined since DenseStorage is an unwindable object for dynamic
-  // matrices and product types are holding a member to store the result.
-  // Thus it does not help tagging this function with EIGEN_STRONG_INLINE.
-  enum {
-    ProductIsValid =  Derived::ColsAtCompileTime==Dynamic
-                   || OtherDerived::RowsAtCompileTime==Dynamic
-                   || int(Derived::ColsAtCompileTime)==int(OtherDerived::RowsAtCompileTime),
-    AreVectors = Derived::IsVectorAtCompileTime && OtherDerived::IsVectorAtCompileTime,
-    SameSizes = EIGEN_PREDICATE_SAME_MATRIX_SIZE(Derived,OtherDerived)
-  };
-  // note to the lost user:
-  //    * for a dot product use: v1.dot(v2)
-  //    * for a coeff-wise product use: v1.cwiseProduct(v2)
-  EIGEN_STATIC_ASSERT(ProductIsValid || !(AreVectors && SameSizes),
-    INVALID_VECTOR_VECTOR_PRODUCT__IF_YOU_WANTED_A_DOT_OR_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTIONS)
-  EIGEN_STATIC_ASSERT(ProductIsValid || !(SameSizes && !AreVectors),
-    INVALID_MATRIX_PRODUCT__IF_YOU_WANTED_A_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTION)
-  EIGEN_STATIC_ASSERT(ProductIsValid || SameSizes, INVALID_MATRIX_PRODUCT)
-#ifdef EIGEN_DEBUG_PRODUCT
-  internal::product_type<Derived,OtherDerived>::debug();
-#endif
-  return typename ProductReturnType<Derived,OtherDerived>::Type(derived(), other.derived());
-}
-
-/** \returns an expression of the matrix product of \c *this and \a other without implicit evaluation.
-  *
-  * The returned product will behave like any other expressions: the coefficients of the product will be
-  * computed once at a time as requested. This might be useful in some extremely rare cases when only
-  * a small and no coherent fraction of the result's coefficients have to be computed.
-  *
-  * \warning This version of the matrix product can be much much slower. So use it only if you know
-  * what you are doing and that you measured a true speed improvement.
-  *
-  * \sa operator*(const MatrixBase&)
-  */
-template<typename Derived>
-template<typename OtherDerived>
-const typename LazyProductReturnType<Derived,OtherDerived>::Type
-MatrixBase<Derived>::lazyProduct(const MatrixBase<OtherDerived> &other) const
-{
-  enum {
-    ProductIsValid =  Derived::ColsAtCompileTime==Dynamic
-                   || OtherDerived::RowsAtCompileTime==Dynamic
-                   || int(Derived::ColsAtCompileTime)==int(OtherDerived::RowsAtCompileTime),
-    AreVectors = Derived::IsVectorAtCompileTime && OtherDerived::IsVectorAtCompileTime,
-    SameSizes = EIGEN_PREDICATE_SAME_MATRIX_SIZE(Derived,OtherDerived)
-  };
-  // note to the lost user:
-  //    * for a dot product use: v1.dot(v2)
-  //    * for a coeff-wise product use: v1.cwiseProduct(v2)
-  EIGEN_STATIC_ASSERT(ProductIsValid || !(AreVectors && SameSizes),
-    INVALID_VECTOR_VECTOR_PRODUCT__IF_YOU_WANTED_A_DOT_OR_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTIONS)
-  EIGEN_STATIC_ASSERT(ProductIsValid || !(SameSizes && !AreVectors),
-    INVALID_MATRIX_PRODUCT__IF_YOU_WANTED_A_COEFF_WISE_PRODUCT_YOU_MUST_USE_THE_EXPLICIT_FUNCTION)
-  EIGEN_STATIC_ASSERT(ProductIsValid || SameSizes, INVALID_MATRIX_PRODUCT)
-
-  return typename LazyProductReturnType<Derived,OtherDerived>::Type(derived(), other.derived());
-}
-
 #endif // EIGEN_PRODUCT_H

Eigen/src/Core/ProductBase.h

@@ -115,10 +115,10 @@ class ProductBase : public MatrixBase<Derived>
     inline void evalTo(Dest& dst) const { dst.setZero(); scaleAndAddTo(dst,Scalar(1)); }
 
     template<typename Dest>
-    inline void addTo(Dest& dst) const { scaleAndAddTo(dst,1); }
+    inline void addTo(Dest& dst) const { scaleAndAddTo(dst,Scalar(1)); }
 
     template<typename Dest>
-    inline void subTo(Dest& dst) const { scaleAndAddTo(dst,-1); }
+    inline void subTo(Dest& dst) const { scaleAndAddTo(dst,Scalar(-1)); }
 
     template<typename Dest>
     inline void scaleAndAddTo(Dest& dst,Scalar alpha) const { derived().scaleAndAddTo(dst,alpha); }
@@ -179,8 +179,8 @@ class ProductBase : public MatrixBase<Derived>
 
   protected:
 
-    const LhsNested m_lhs;
-    const RhsNested m_rhs;
+    LhsNested m_lhs;
+    RhsNested m_rhs;
 
     mutable PlainObject m_result;
 };
@@ -256,16 +256,16 @@ class ScaledProduct
     : Base(prod.lhs(),prod.rhs()), m_prod(prod), m_alpha(x) {}
 
     template<typename Dest>
-    inline void evalTo(Dest& dst) const { dst.setZero(); scaleAndAddTo(dst,m_alpha); }
+    inline void evalTo(Dest& dst) const { dst.setZero(); scaleAndAddTo(dst, Scalar(1)); }
 
     template<typename Dest>
-    inline void addTo(Dest& dst) const { scaleAndAddTo(dst,m_alpha); }
+    inline void addTo(Dest& dst) const { scaleAndAddTo(dst, Scalar(1)); }
 
     template<typename Dest>
-    inline void subTo(Dest& dst) const { scaleAndAddTo(dst,-m_alpha); }
+    inline void subTo(Dest& dst) const { scaleAndAddTo(dst, Scalar(-1)); }
 
     template<typename Dest>
-    inline void scaleAndAddTo(Dest& dst,Scalar alpha) const { m_prod.derived().scaleAndAddTo(dst,alpha); }
+    inline void scaleAndAddTo(Dest& dst,Scalar alpha) const { m_prod.derived().scaleAndAddTo(dst,alpha * m_alpha); }
 
     const Scalar& alpha() const { return m_alpha; }
     

Eigen/src/Core/Redux.h

@@ -95,7 +95,7 @@ struct redux_novec_unroller
 
   typedef typename Derived::Scalar Scalar;
 
-  EIGEN_STRONG_INLINE static Scalar run(const Derived &mat, const Func& func)
+  static EIGEN_STRONG_INLINE Scalar run(const Derived &mat, const Func& func)
   {
     return func(redux_novec_unroller<Func, Derived, Start, HalfLength>::run(mat,func),
                 redux_novec_unroller<Func, Derived, Start+HalfLength, Length-HalfLength>::run(mat,func));
@@ -112,7 +112,7 @@ struct redux_novec_unroller<Func, Derived, Start, 1>
 
   typedef typename Derived::Scalar Scalar;
 
-  EIGEN_STRONG_INLINE static Scalar run(const Derived &mat, const Func&)
+  static EIGEN_STRONG_INLINE Scalar run(const Derived &mat, const Func&)
   {
     return mat.coeffByOuterInner(outer, inner);
   }
@@ -125,7 +125,7 @@ template<typename Func, typename Derived, int Start>
 struct redux_novec_unroller<Func, Derived, Start, 0>
 {
   typedef typename Derived::Scalar Scalar;
-  EIGEN_STRONG_INLINE static Scalar run(const Derived&, const Func&) { return Scalar(); }
+  static EIGEN_STRONG_INLINE Scalar run(const Derived&, const Func&) { return Scalar(); }
 };
 
 /*** vectorization ***/
@@ -141,7 +141,7 @@ struct redux_vec_unroller
   typedef typename Derived::Scalar Scalar;
   typedef typename packet_traits<Scalar>::type PacketScalar;
 
-  EIGEN_STRONG_INLINE static PacketScalar run(const Derived &mat, const Func& func)
+  static EIGEN_STRONG_INLINE PacketScalar run(const Derived &mat, const Func& func)
   {
     return func.packetOp(
             redux_vec_unroller<Func, Derived, Start, HalfLength>::run(mat,func),
@@ -162,7 +162,7 @@ struct redux_vec_unroller<Func, Derived, Start, 1>
   typedef typename Derived::Scalar Scalar;
   typedef typename packet_traits<Scalar>::type PacketScalar;
 
-  EIGEN_STRONG_INLINE static PacketScalar run(const Derived &mat, const Func&)
+  static EIGEN_STRONG_INLINE PacketScalar run(const Derived &mat, const Func&)
   {
     return mat.template packetByOuterInner<alignment>(outer, inner);
   }
@@ -214,20 +214,33 @@ struct redux_impl<Func, Derived, LinearVectorizedTraversal, NoUnrolling>
     const Index size = mat.size();
     eigen_assert(size && "you are using an empty matrix");
     const Index packetSize = packet_traits<Scalar>::size;
-    const Index alignedStart = first_aligned(mat);
+    const Index alignedStart = internal::first_aligned(mat);
     enum {
-      alignment = (Derived::Flags & DirectAccessBit) || (Derived::Flags & AlignedBit)
+      alignment = bool(Derived::Flags & DirectAccessBit) || bool(Derived::Flags & AlignedBit)
                 ? Aligned : Unaligned
     };
-    const Index alignedSize = ((size-alignedStart)/packetSize)*packetSize;
-    const Index alignedEnd = alignedStart + alignedSize;
+    const Index alignedSize2 = ((size-alignedStart)/(2*packetSize))*(2*packetSize);
+    const Index alignedSize = ((size-alignedStart)/(packetSize))*(packetSize);
+    const Index alignedEnd2 = alignedStart + alignedSize2;
+    const Index alignedEnd  = alignedStart + alignedSize;
     Scalar res;
     if(alignedSize)
     {
-      PacketScalar packet_res = mat.template packet<alignment>(alignedStart);
-      for(Index index = alignedStart + packetSize; index < alignedEnd; index += packetSize)
-        packet_res = func.packetOp(packet_res, mat.template packet<alignment>(index));
-      res = func.predux(packet_res);
+      PacketScalar packet_res0 = mat.template packet<alignment>(alignedStart);
+      if(alignedSize>packetSize) // we have at least two packets to partly unroll the loop
+      {
+        PacketScalar packet_res1 = mat.template packet<alignment>(alignedStart+packetSize);
+        for(Index index = alignedStart + 2*packetSize; index < alignedEnd2; index += 2*packetSize)
+        {
+          packet_res0 = func.packetOp(packet_res0, mat.template packet<alignment>(index));
+          packet_res1 = func.packetOp(packet_res1, mat.template packet<alignment>(index+packetSize));
+        }
+
+        packet_res0 = func.packetOp(packet_res0,packet_res1);
+        if(alignedEnd>alignedEnd2)
+          packet_res0 = func.packetOp(packet_res0, mat.template packet<alignment>(alignedEnd2));
+      }
+      res = func.predux(packet_res0);
 
       for(Index index = 0; index < alignedStart; ++index)
         res = func(res,mat.coeff(index));
@@ -296,7 +309,7 @@ struct redux_impl<Func, Derived, LinearVectorizedTraversal, CompleteUnrolling>
     Size = Derived::SizeAtCompileTime,
     VectorizedSize = (Size / PacketSize) * PacketSize
   };
-  EIGEN_STRONG_INLINE static Scalar run(const Derived& mat, const Func& func)
+  static EIGEN_STRONG_INLINE Scalar run(const Derived& mat, const Func& func)
   {
     eigen_assert(mat.rows()>0 && mat.cols()>0 && "you are using an empty matrix");
     Scalar res = func.predux(redux_vec_unroller<Func, Derived, 0, Size / PacketSize>::run(mat,func));

Eigen/src/Core/Replicate.h

@@ -122,9 +122,13 @@ template<typename MatrixType,int RowFactor,int ColFactor> class Replicate
       return m_matrix.template packet<LoadMode>(actual_row, actual_col);
     }
 
+    const typename internal::remove_all<typename MatrixType::Nested>::type& nestedExpression() const 
+    { 
+      return m_matrix; 
+    }
 
   protected:
-    const typename MatrixType::Nested m_matrix;
+    typename MatrixType::Nested m_matrix;
     const internal::variable_if_dynamic<Index, RowFactor> m_rowFactor;
     const internal::variable_if_dynamic<Index, ColFactor> m_colFactor;
 };

Eigen/src/Core/ReturnByValue.h

@@ -71,9 +71,9 @@ template<typename Derived> class ReturnByValue
 
     template<typename Dest>
     inline void evalTo(Dest& dst) const
-    { static_cast<const Derived* const>(this)->evalTo(dst); }
-    inline Index rows() const { return static_cast<const Derived* const>(this)->rows(); }
-    inline Index cols() const { return static_cast<const Derived* const>(this)->cols(); }
+    { static_cast<const Derived*>(this)->evalTo(dst); }
+    inline Index rows() const { return static_cast<const Derived*>(this)->rows(); }
+    inline Index cols() const { return static_cast<const Derived*>(this)->cols(); }
 
 #ifndef EIGEN_PARSED_BY_DOXYGEN
 #define Unusable YOU_ARE_TRYING_TO_ACCESS_A_SINGLE_COEFFICIENT_IN_A_SPECIAL_EXPRESSION_WHERE_THAT_IS_NOT_ALLOWED_BECAUSE_THAT_WOULD_BE_INEFFICIENT

Eigen/src/Core/Reverse.h

@@ -183,8 +183,14 @@ template<typename MatrixType, int Direction> class Reverse
       m_matrix.const_cast_derived().template writePacket<LoadMode>(m_matrix.size() - index - PacketSize, internal::preverse(x));
     }
 
+    const typename internal::remove_all<typename MatrixType::Nested>::type& 
+    nestedExpression() const 
+    {
+      return m_matrix;
+    }
+
   protected:
-    const typename MatrixType::Nested m_matrix;
+    typename MatrixType::Nested m_matrix;
 };
 
 /** \returns an expression of the reverse of *this.

Eigen/src/Core/Select.h

@@ -101,10 +101,25 @@ class Select : internal::no_assignment_operator,
         return m_else.coeff(i);
     }
 
+    const ConditionMatrixType& conditionMatrix() const
+    {
+      return m_condition;
+    }
+
+    const ThenMatrixType& thenMatrix() const
+    {
+      return m_then;
+    }
+
+    const ElseMatrixType& elseMatrix() const
+    {
+      return m_else;
+    }
+
   protected:
-    const typename ConditionMatrixType::Nested m_condition;
-    const typename ThenMatrixType::Nested m_then;
-    const typename ElseMatrixType::Nested m_else;
+    typename ConditionMatrixType::Nested m_condition;
+    typename ThenMatrixType::Nested m_then;
+    typename ElseMatrixType::Nested m_else;
 };
 
 

Eigen/src/Core/SelfAdjointView.h

@@ -32,13 +32,13 @@
   * \brief Expression of a selfadjoint matrix from a triangular part of a dense matrix
   *
   * \param MatrixType the type of the dense matrix storing the coefficients
-  * \param TriangularPart can be either \c Lower or \c Upper
+  * \param TriangularPart can be either \c #Lower or \c #Upper
   *
   * This class is an expression of a sefladjoint matrix from a triangular part of a matrix
   * with given dense storage of the coefficients. It is the return type of MatrixBase::selfadjointView()
   * and most of the time this is the only way that it is used.
   *
-  * \sa class TriangularBase, MatrixBase::selfAdjointView()
+  * \sa class TriangularBase, MatrixBase::selfadjointView()
   */
 
 namespace internal {
@@ -82,7 +82,7 @@ template<typename MatrixType, unsigned int UpLo> class SelfAdjointView
     };
     typedef typename MatrixType::PlainObject PlainObject;
 
-    inline SelfAdjointView(const MatrixType& matrix) : m_matrix(matrix)
+    inline SelfAdjointView(MatrixType& matrix) : m_matrix(matrix)
     {}
 
     inline Index rows() const { return m_matrix.rows(); }
@@ -199,7 +199,7 @@ template<typename MatrixType, unsigned int UpLo> class SelfAdjointView
     #endif
 
   protected:
-    const MatrixTypeNested m_matrix;
+    MatrixTypeNested m_matrix;
 };
 
 
@@ -222,7 +222,7 @@ struct triangular_assignment_selector<Derived1, Derived2, (SelfAdjoint|Upper), U
     row = (UnrollCount-1) % Derived1::RowsAtCompileTime
   };
 
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     triangular_assignment_selector<Derived1, Derived2, (SelfAdjoint|Upper), UnrollCount-1, ClearOpposite>::run(dst, src);
 
@@ -236,7 +236,7 @@ struct triangular_assignment_selector<Derived1, Derived2, (SelfAdjoint|Upper), U
 template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, SelfAdjoint|Upper, 0, ClearOpposite>
 {
-  inline static void run(Derived1 &, const Derived2 &) {}
+  static inline void run(Derived1 &, const Derived2 &) {}
 };
 
 template<typename Derived1, typename Derived2, int UnrollCount, bool ClearOpposite>
@@ -247,7 +247,7 @@ struct triangular_assignment_selector<Derived1, Derived2, (SelfAdjoint|Lower), U
     row = (UnrollCount-1) % Derived1::RowsAtCompileTime
   };
 
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     triangular_assignment_selector<Derived1, Derived2, (SelfAdjoint|Lower), UnrollCount-1, ClearOpposite>::run(dst, src);
 
@@ -261,14 +261,14 @@ struct triangular_assignment_selector<Derived1, Derived2, (SelfAdjoint|Lower), U
 template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, SelfAdjoint|Lower, 0, ClearOpposite>
 {
-  inline static void run(Derived1 &, const Derived2 &) {}
+  static inline void run(Derived1 &, const Derived2 &) {}
 };
 
 template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, SelfAdjoint|Upper, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
@@ -285,7 +285,7 @@ struct triangular_assignment_selector<Derived1, Derived2, SelfAdjoint|Upper, Dyn
 template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, SelfAdjoint|Lower, Dynamic, ClearOpposite>
 {
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
   typedef typename Derived1::Index Index;
     for(Index i = 0; i < dst.rows(); ++i)

Eigen/src/Core/SelfCwiseBinaryOp.h

@@ -163,6 +163,16 @@ template<typename BinaryOp, typename Lhs, typename Rhs> class SelfCwiseBinaryOp
       return Base::operator=(rhs);
     }
 
+    Lhs& expression() const 
+    { 
+      return m_matrix;
+    }
+
+    const BinaryOp& functor() const 
+    { 
+      return m_functor;
+    }
+
   protected:
     Lhs& m_matrix;
     const BinaryOp& m_functor;

Eigen/src/Core/SolveTriangular.h

@@ -74,26 +74,19 @@ struct triangular_solver_selector<Lhs,Rhs,Side,Mode,NoUnrolling,1>
     // FIXME find a way to allow an inner stride if packet_traits<Scalar>::size==1
 
     bool useRhsDirectly = Rhs::InnerStrideAtCompileTime==1 || rhs.innerStride()==1;
-    RhsScalar* actualRhs;
-    if(useRhsDirectly)
-    {
-      actualRhs = &rhs.coeffRef(0);
-    }
-    else
-    {
-      actualRhs = ei_aligned_stack_new(RhsScalar,rhs.size());
+
+    ei_declare_aligned_stack_constructed_variable(RhsScalar,actualRhs,rhs.size(),
+                                                  (useRhsDirectly ? rhs.data() : 0));
+                                                  
+    if(!useRhsDirectly)
       MappedRhs(actualRhs,rhs.size()) = rhs;
-    }
 
     triangular_solve_vector<LhsScalar, RhsScalar, typename Lhs::Index, Side, Mode, LhsProductTraits::NeedToConjugate,
                             (int(Lhs::Flags) & RowMajorBit) ? RowMajor : ColMajor>
       ::run(actualLhs.cols(), actualLhs.data(), actualLhs.outerStride(), actualRhs);
 
     if(!useRhsDirectly)
-    {
       rhs = MappedRhs(actualRhs, rhs.size());
-      ei_aligned_stack_delete(RhsScalar, actualRhs, rhs.size());
-    }
   }
 };
 
@@ -107,7 +100,7 @@ struct triangular_solver_selector<Lhs,Rhs,Side,Mode,NoUnrolling,Dynamic>
   typedef typename LhsProductTraits::DirectLinearAccessType ActualLhsType;
   static void run(const Lhs& lhs, Rhs& rhs)
   {
-    const ActualLhsType actualLhs = LhsProductTraits::extract(lhs);
+    typename internal::add_const_on_value_type<ActualLhsType>::type actualLhs = LhsProductTraits::extract(lhs);
     triangular_solve_matrix<Scalar,Index,Side,Mode,LhsProductTraits::NeedToConjugate,(int(Lhs::Flags) & RowMajorBit) ? RowMajor : ColMajor,
                                (Rhs::Flags&RowMajorBit) ? RowMajor : ColMajor>
       ::run(lhs.rows(), Side==OnTheLeft? rhs.cols() : rhs.rows(), &actualLhs.coeffRef(0,0), actualLhs.outerStride(), &rhs.coeffRef(0,0), rhs.outerStride());
@@ -184,10 +177,8 @@ template<int Side, typename OtherDerived>
 void TriangularView<MatrixType,Mode>::solveInPlace(const MatrixBase<OtherDerived>& _other) const
 {
   OtherDerived& other = _other.const_cast_derived();
-  eigen_assert(cols() == rows());
-  eigen_assert( (Side==OnTheLeft && cols() == other.rows()) || (Side==OnTheRight && cols() == other.cols()) );
-  eigen_assert(!(Mode & ZeroDiag));
-  eigen_assert(Mode & (Upper|Lower));
+  eigen_assert( cols() == rows() && ((Side==OnTheLeft && cols() == other.rows()) || (Side==OnTheRight && cols() == other.cols())) );
+  eigen_assert((!(Mode & ZeroDiag)) && bool(Mode & (Upper|Lower)));
 
   enum { copy = internal::traits<OtherDerived>::Flags & RowMajorBit  && OtherDerived::IsVectorAtCompileTime };
   typedef typename internal::conditional<copy,
@@ -262,7 +253,7 @@ template<int Side, typename TriangularType, typename Rhs> struct triangular_solv
 
   protected:
     const TriangularType& m_triangularMatrix;
-    const typename Rhs::Nested m_rhs;
+    typename Rhs::Nested m_rhs;
 };
 
 } // namespace internal

Eigen/src/Core/StableNorm.h

@@ -56,10 +56,11 @@ template<typename Derived>
 inline typename NumTraits<typename internal::traits<Derived>::Scalar>::Real
 MatrixBase<Derived>::stableNorm() const
 {
+  using std::min;
   const Index blockSize = 4096;
-  RealScalar scale = 0;
-  RealScalar invScale = 1;
-  RealScalar ssq = 0; // sum of square
+  RealScalar scale(0);
+  RealScalar invScale(1);
+  RealScalar ssq(0); // sum of square
   enum {
     Alignment = (int(Flags)&DirectAccessBit) || (int(Flags)&AlignedBit) ? 1 : 0
   };
@@ -68,7 +69,7 @@ MatrixBase<Derived>::stableNorm() const
   if (bi>0)
     internal::stable_norm_kernel(this->head(bi), ssq, scale, invScale);
   for (; bi<n; bi+=blockSize)
-    internal::stable_norm_kernel(this->segment(bi,std::min(blockSize, n - bi)).template forceAlignedAccessIf<Alignment>(), ssq, scale, invScale);
+    internal::stable_norm_kernel(this->segment(bi,(min)(blockSize, n - bi)).template forceAlignedAccessIf<Alignment>(), ssq, scale, invScale);
   return scale * internal::sqrt(ssq);
 }
 
@@ -85,6 +86,9 @@ template<typename Derived>
 inline typename NumTraits<typename internal::traits<Derived>::Scalar>::Real
 MatrixBase<Derived>::blueNorm() const
 {
+  using std::pow;
+  using std::min;
+  using std::max;
   static Index nmax = -1;
   static RealScalar b1, b2, s1m, s2m, overfl, rbig, relerr;
   if(nmax <= 0)
@@ -99,25 +103,25 @@ MatrixBase<Derived>::blueNorm() const
     // For portability, the PORT subprograms "ilmaeh" and "rlmach"
     // are used. For any specific computer, each of the assignment
     // statements can be replaced
-    nbig  = std::numeric_limits<Index>::max();            // largest integer
+    nbig  = (std::numeric_limits<Index>::max)();            // largest integer
     ibeta = std::numeric_limits<RealScalar>::radix;         // base for floating-point numbers
     it    = std::numeric_limits<RealScalar>::digits;        // number of base-beta digits in mantissa
     iemin = std::numeric_limits<RealScalar>::min_exponent;  // minimum exponent
     iemax = std::numeric_limits<RealScalar>::max_exponent;  // maximum exponent
-    rbig  = std::numeric_limits<RealScalar>::max();         // largest floating-point number
+    rbig  = (std::numeric_limits<RealScalar>::max)();         // largest floating-point number
 
     iexp  = -((1-iemin)/2);
-    b1    = RealScalar(std::pow(RealScalar(ibeta),RealScalar(iexp)));  // lower boundary of midrange
+    b1    = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));  // lower boundary of midrange
     iexp  = (iemax + 1 - it)/2;
-    b2    = RealScalar(std::pow(RealScalar(ibeta),RealScalar(iexp)));   // upper boundary of midrange
+    b2    = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));   // upper boundary of midrange
 
     iexp  = (2-iemin)/2;
-    s1m   = RealScalar(std::pow(RealScalar(ibeta),RealScalar(iexp)));   // scaling factor for lower range
+    s1m   = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));   // scaling factor for lower range
     iexp  = - ((iemax+it)/2);
-    s2m   = RealScalar(std::pow(RealScalar(ibeta),RealScalar(iexp)));   // scaling factor for upper range
+    s2m   = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));   // scaling factor for upper range
 
     overfl  = rbig*s2m;             // overflow boundary for abig
-    eps     = RealScalar(std::pow(double(ibeta), 1-it));
+    eps     = RealScalar(pow(double(ibeta), 1-it));
     relerr  = internal::sqrt(eps);         // tolerance for neglecting asml
     abig    = RealScalar(1.0/eps - 1.0);
     if (RealScalar(nbig)>abig)  nmax = int(abig);  // largest safe n
@@ -163,8 +167,8 @@ MatrixBase<Derived>::blueNorm() const
   }
   else
     return internal::sqrt(amed);
-  asml = std::min(abig, amed);
-  abig = std::max(abig, amed);
+  asml = (min)(abig, amed);
+  abig = (max)(abig, amed);
   if(asml <= abig*relerr)
     return abig;
   else

Eigen/src/Core/Swap.h

@@ -52,6 +52,15 @@ template<typename ExpressionType> class SwapWrapper
     inline Index cols() const { return m_expression.cols(); }
     inline Index outerStride() const { return m_expression.outerStride(); }
     inline Index innerStride() const { return m_expression.innerStride(); }
+    
+    typedef typename internal::conditional<
+                       internal::is_lvalue<ExpressionType>::value,
+                       Scalar,
+                       const Scalar
+                     >::type ScalarWithConstIfNotLvalue;
+                     
+    inline ScalarWithConstIfNotLvalue* data() { return m_expression.data(); }
+    inline const Scalar* data() const { return m_expression.data(); }
 
     inline Scalar& coeffRef(Index row, Index col)
     {
@@ -119,6 +128,8 @@ template<typename ExpressionType> class SwapWrapper
       _other.template writePacket<LoadMode>(index, tmp);
     }
 
+    ExpressionType& expression() const { return m_expression; }
+
   protected:
     ExpressionType& m_expression;
 };

Eigen/src/Core/Transpose.h

@@ -91,7 +91,7 @@ template<typename MatrixType> class Transpose
     nestedExpression() { return m_matrix.const_cast_derived(); }
 
   protected:
-    const typename MatrixType::Nested m_matrix;
+    typename MatrixType::Nested m_matrix;
 };
 
 namespace internal {
@@ -152,12 +152,12 @@ template<typename MatrixType> class TransposeImpl<MatrixType,Dense>
       return derived().nestedExpression().coeffRef(index);
     }
 
-    inline const CoeffReturnType coeff(Index row, Index col) const
+    inline CoeffReturnType coeff(Index row, Index col) const
     {
       return derived().nestedExpression().coeff(col, row);
     }
 
-    inline const CoeffReturnType coeff(Index index) const
+    inline CoeffReturnType coeff(Index index) const
     {
       return derived().nestedExpression().coeff(index);
     }
@@ -350,15 +350,14 @@ struct blas_traits<SelfCwiseBinaryOp<BinOp,NestedXpr,Rhs> >
 template<bool DestIsTransposed, typename OtherDerived>
 struct check_transpose_aliasing_compile_time_selector
 {
-  enum { ret = blas_traits<OtherDerived>::IsTransposed != DestIsTransposed
-  };
+  enum { ret = bool(blas_traits<OtherDerived>::IsTransposed) != DestIsTransposed };
 };
 
 template<bool DestIsTransposed, typename BinOp, typename DerivedA, typename DerivedB>
 struct check_transpose_aliasing_compile_time_selector<DestIsTransposed,CwiseBinaryOp<BinOp,DerivedA,DerivedB> >
 {
-  enum { ret =    blas_traits<DerivedA>::IsTransposed != DestIsTransposed
-               || blas_traits<DerivedB>::IsTransposed != DestIsTransposed
+  enum { ret =    bool(blas_traits<DerivedA>::IsTransposed) != DestIsTransposed
+               || bool(blas_traits<DerivedB>::IsTransposed) != DestIsTransposed
   };
 };
 
@@ -367,7 +366,7 @@ struct check_transpose_aliasing_run_time_selector
 {
   static bool run(const Scalar* dest, const OtherDerived& src)
   {
-    return (blas_traits<OtherDerived>::IsTransposed != DestIsTransposed) && (dest!=0 && dest==(Scalar*)extract_data(src));
+    return (bool(blas_traits<OtherDerived>::IsTransposed) != DestIsTransposed) && (dest!=0 && dest==(Scalar*)extract_data(src));
   }
 };
 

Eigen/src/Core/Transpositions.h

@@ -404,7 +404,7 @@ struct transposition_matrix_product_retval
 
   protected:
     const TranspositionType& m_transpositions;
-    const typename MatrixType::Nested m_matrix;
+    typename MatrixType::Nested m_matrix;
 };
 
 } // end namespace internal

Eigen/src/Core/TriangularMatrix.h

@@ -111,6 +111,7 @@ template<typename Derived> class TriangularBase : public EigenBase<Derived>
       EIGEN_ONLY_USED_FOR_DEBUG(col);
       eigen_assert(col>=0 && col<cols() && row>=0 && row<rows());
       const int mode = int(Mode) & ~SelfAdjoint;
+      EIGEN_ONLY_USED_FOR_DEBUG(mode);
       eigen_assert((mode==Upper && col>=row)
                 || (mode==Lower && col<=row)
                 || ((mode==StrictlyUpper || mode==UnitUpper) && col>row)
@@ -134,13 +135,13 @@ template<typename Derived> class TriangularBase : public EigenBase<Derived>
   * \brief Base class for triangular part in a matrix
   *
   * \param MatrixType the type of the object in which we are taking the triangular part
-  * \param Mode the kind of triangular matrix expression to construct. Can be Upper,
-  *             Lower, UpperSelfadjoint, or LowerSelfadjoint. This is in fact a bit field;
-  *             it must have either Upper or Lower, and additionnaly it may have either
-  *             UnitDiag or Selfadjoint.
+  * \param Mode the kind of triangular matrix expression to construct. Can be #Upper,
+  *             #Lower, #UnitUpper, #UnitLower, #StrictlyUpper, or #StrictlyLower.
+  *             This is in fact a bit field; it must have either #Upper or #Lower, 
+  *             and additionnaly it may have #UnitDiag or #ZeroDiag or neither.
   *
   * This class represents a triangular part of a matrix, not necessarily square. Strictly speaking, for rectangular
-  * matrices one should speak ok "trapezoid" parts. This class is the return type
+  * matrices one should speak of "trapezoid" parts. This class is the return type
   * of MatrixBase::triangularView() and most of the time this is the only way it is used.
   *
   * \sa MatrixBase::triangularView()
@@ -272,11 +273,8 @@ template<typename _MatrixType, unsigned int _Mode> class TriangularView
     inline const TriangularView<MatrixConjugateReturnType,Mode> conjugate() const
     { return m_matrix.conjugate(); }
 
-    /** \sa MatrixBase::adjoint() */
-    inline TriangularView<typename MatrixType::AdjointReturnType,TransposeMode> adjoint()
-    { return m_matrix.adjoint(); }
     /** \sa MatrixBase::adjoint() const */
-    inline const TriangularView<typename MatrixType::AdjointReturnType,TransposeMode> adjoint() const
+    inline const TriangularView<const typename MatrixType::AdjointReturnType,TransposeMode> adjoint() const
     { return m_matrix.adjoint(); }
 
     /** \sa MatrixBase::transpose() */
@@ -287,11 +285,13 @@ template<typename _MatrixType, unsigned int _Mode> class TriangularView
     }
     /** \sa MatrixBase::transpose() const */
     inline const TriangularView<Transpose<MatrixType>,TransposeMode> transpose() const
-    { return m_matrix.transpose(); }
+    {
+      return m_matrix.transpose();
+    }
 
     /** Efficient triangular matrix times vector/matrix product */
     template<typename OtherDerived>
-    TriangularProduct<Mode,true,MatrixType,false,OtherDerived,OtherDerived::IsVectorAtCompileTime>
+    TriangularProduct<Mode,true,MatrixType,false,OtherDerived, OtherDerived::IsVectorAtCompileTime>
     operator*(const MatrixBase<OtherDerived>& rhs) const
     {
       return TriangularProduct
@@ -374,7 +374,8 @@ template<typename _MatrixType, unsigned int _Mode> class TriangularView
     template<typename OtherDerived>
     void swap(MatrixBase<OtherDerived> const & other)
     {
-      TriangularView<SwapWrapper<MatrixType>,Mode>(const_cast<MatrixType&>(m_matrix)).lazyAssign(other.derived());
+      SwapWrapper<MatrixType> swaper(const_cast<MatrixType&>(m_matrix));
+      TriangularView<SwapWrapper<MatrixType>,Mode>(swaper).lazyAssign(other.derived());
     }
 
     Scalar determinant() const
@@ -432,7 +433,7 @@ template<typename _MatrixType, unsigned int _Mode> class TriangularView
     template<typename ProductDerived, typename Lhs, typename Rhs>
     EIGEN_STRONG_INLINE TriangularView& assignProduct(const ProductBase<ProductDerived, Lhs,Rhs>& prod, const Scalar& alpha);
 
-    const MatrixTypeNested m_matrix;
+    MatrixTypeNested m_matrix;
 };
 
 /***************************************************************************
@@ -448,8 +449,10 @@ struct triangular_assignment_selector
     col = (UnrollCount-1) / Derived1::RowsAtCompileTime,
     row = (UnrollCount-1) % Derived1::RowsAtCompileTime
   };
+  
+  typedef typename Derived1::Scalar Scalar;
 
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     triangular_assignment_selector<Derived1, Derived2, Mode, UnrollCount-1, ClearOpposite>::run(dst, src);
 
@@ -466,9 +469,9 @@ struct triangular_assignment_selector
     else if(ClearOpposite)
     {
       if (Mode&UnitDiag && row==col)
-        dst.coeffRef(row, col) = 1;
+        dst.coeffRef(row, col) = Scalar(1);
       else
-        dst.coeffRef(row, col) = 0;
+        dst.coeffRef(row, col) = Scalar(0);
     }
   }
 };
@@ -477,23 +480,24 @@ struct triangular_assignment_selector
 template<typename Derived1, typename Derived2, unsigned int Mode, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, Mode, 0, ClearOpposite>
 {
-  inline static void run(Derived1 &, const Derived2 &) {}
+  static inline void run(Derived1 &, const Derived2 &) {}
 };
 
 template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, Upper, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  typedef typename Derived1::Scalar Scalar;
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
-      Index maxi = std::min(j, dst.rows()-1);
+      Index maxi = (std::min)(j, dst.rows()-1);
       for(Index i = 0; i <= maxi; ++i)
         dst.copyCoeff(i, j, src);
       if (ClearOpposite)
         for(Index i = maxi+1; i < dst.rows(); ++i)
-          dst.coeffRef(i, j) = 0;
+          dst.coeffRef(i, j) = Scalar(0);
     }
   }
 };
@@ -502,16 +506,16 @@ template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, Lower, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
       for(Index i = j; i < dst.rows(); ++i)
         dst.copyCoeff(i, j, src);
-      Index maxi = std::min(j, dst.rows());
+      Index maxi = (std::min)(j, dst.rows());
       if (ClearOpposite)
         for(Index i = 0; i < maxi; ++i)
-          dst.coeffRef(i, j) = 0;
+          dst.coeffRef(i, j) = static_cast<typename Derived1::Scalar>(0);
     }
   }
 };
@@ -520,11 +524,11 @@ template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, StrictlyUpper, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
-      Index maxi = std::min(j, dst.rows());
+      Index maxi = (std::min)(j, dst.rows());
       for(Index i = 0; i < maxi; ++i)
         dst.copyCoeff(i, j, src);
       if (ClearOpposite)
@@ -538,16 +542,16 @@ template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, StrictlyLower, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
       for(Index i = j+1; i < dst.rows(); ++i)
         dst.copyCoeff(i, j, src);
-      Index maxi = std::min(j, dst.rows()-1);
+      Index maxi = (std::min)(j, dst.rows()-1);
       if (ClearOpposite)
         for(Index i = 0; i <= maxi; ++i)
-          dst.coeffRef(i, j) = 0;
+          dst.coeffRef(i, j) = static_cast<typename Derived1::Scalar>(0);
     }
   }
 };
@@ -556,11 +560,11 @@ template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, UnitUpper, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
-      Index maxi = std::min(j, dst.rows());
+      Index maxi = (std::min)(j, dst.rows());
       for(Index i = 0; i < maxi; ++i)
         dst.copyCoeff(i, j, src);
       if (ClearOpposite)
@@ -576,11 +580,11 @@ template<typename Derived1, typename Derived2, bool ClearOpposite>
 struct triangular_assignment_selector<Derived1, Derived2, UnitLower, Dynamic, ClearOpposite>
 {
   typedef typename Derived1::Index Index;
-  inline static void run(Derived1 &dst, const Derived2 &src)
+  static inline void run(Derived1 &dst, const Derived2 &src)
   {
     for(Index j = 0; j < dst.cols(); ++j)
     {
-      Index maxi = std::min(j, dst.rows());
+      Index maxi = (std::min)(j, dst.rows());
       for(Index i = maxi+1; i < dst.rows(); ++i)
         dst.copyCoeff(i, j, src);
       if (ClearOpposite)
@@ -756,8 +760,8 @@ typename internal::eigen2_part_return_type<Derived, Mode>::type MatrixBase<Deriv
 /**
   * \returns an expression of a triangular view extracted from the current matrix
   *
-  * The parameter \a Mode can have the following values: \c Upper, \c StrictlyUpper, \c UnitUpper,
-  * \c Lower, \c StrictlyLower, \c UnitLower.
+  * The parameter \a Mode can have the following values: \c #Upper, \c #StrictlyUpper, \c #UnitUpper,
+  * \c #Lower, \c #StrictlyLower, \c #UnitLower.
   *
   * Example: \include MatrixBase_extract.cpp
   * Output: \verbinclude MatrixBase_extract.out
@@ -792,7 +796,7 @@ bool MatrixBase<Derived>::isUpperTriangular(RealScalar prec) const
   RealScalar maxAbsOnUpperPart = static_cast<RealScalar>(-1);
   for(Index j = 0; j < cols(); ++j)
   {
-    Index maxi = std::min(j, rows()-1);
+    Index maxi = (std::min)(j, rows()-1);
     for(Index i = 0; i <= maxi; ++i)
     {
       RealScalar absValue = internal::abs(coeff(i,j));
@@ -824,7 +828,7 @@ bool MatrixBase<Derived>::isLowerTriangular(RealScalar prec) const
   RealScalar threshold = maxAbsOnLowerPart * prec;
   for(Index j = 1; j < cols(); ++j)
   {
-    Index maxi = std::min(j, rows()-1);
+    Index maxi = (std::min)(j, rows()-1);
     for(Index i = 0; i < maxi; ++i)
       if(internal::abs(coeff(i, j)) > threshold) return false;
   }

Eigen/src/Core/VectorwiseOp.h

@@ -31,9 +31,9 @@
   *
   * \brief Generic expression of a partially reduxed matrix
   *
-  * \param MatrixType the type of the matrix we are applying the redux operation
-  * \param MemberOp type of the member functor
-  * \param Direction indicates the direction of the redux (Vertical or Horizontal)
+  * \tparam MatrixType the type of the matrix we are applying the redux operation
+  * \tparam MemberOp type of the member functor
+  * \tparam Direction indicates the direction of the redux (#Vertical or #Horizontal)
   *
   * This class represents an expression of a partial redux operator of a matrix.
   * It is the return type of some VectorwiseOp functions,
@@ -110,7 +110,7 @@ class PartialReduxExpr : internal::no_assignment_operator,
     }
 
   protected:
-    const MatrixTypeNested m_matrix;
+    MatrixTypeNested m_matrix;
     const MemberOp m_functor;
 };
 
@@ -164,7 +164,7 @@ struct member_redux {
   * \brief Pseudo expression providing partial reduction operations
   *
   * \param ExpressionType the type of the object on which to do partial reductions
-  * \param Direction indicates the direction of the redux (Vertical or Horizontal)
+  * \param Direction indicates the direction of the redux (#Vertical or #Horizontal)
   *
   * This class represents a pseudo expression with partial reduction features.
   * It is the return type of DenseBase::colwise() and DenseBase::rowwise()
@@ -237,7 +237,10 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
     typename ExtendedType<OtherDerived>::Type
     extendedTo(const DenseBase<OtherDerived>& other) const
     {
-      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived);
+      EIGEN_STATIC_ASSERT(EIGEN_IMPLIES(Direction==Vertical, OtherDerived::MaxColsAtCompileTime==1),
+                          YOU_PASSED_A_ROW_VECTOR_BUT_A_COLUMN_VECTOR_WAS_EXPECTED)
+      EIGEN_STATIC_ASSERT(EIGEN_IMPLIES(Direction==Horizontal, OtherDerived::MaxRowsAtCompileTime==1),
+                          YOU_PASSED_A_COLUMN_VECTOR_BUT_A_ROW_VECTOR_WAS_EXPECTED)
       return typename ExtendedType<OtherDerived>::Type
                       (other.derived(),
                        Direction==Vertical   ? 1 : m_matrix.rows(),
@@ -418,10 +421,9 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
     ExpressionType& operator=(const DenseBase<OtherDerived>& other)
     {
       EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
       //eigen_assert((m_matrix.isNull()) == (other.isNull())); FIXME
-      for(Index j=0; j<subVectors(); ++j)
-        subVector(j) = other;
-      return const_cast<ExpressionType&>(m_matrix);
+      return const_cast<ExpressionType&>(m_matrix = extendedTo(other.derived()));
     }
 
     /** Adds the vector \a other to each subvector of \c *this */
@@ -429,9 +431,8 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
     ExpressionType& operator+=(const DenseBase<OtherDerived>& other)
     {
       EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
-      for(Index j=0; j<subVectors(); ++j)
-        subVector(j) += other.derived();
-      return const_cast<ExpressionType&>(m_matrix);
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
+      return const_cast<ExpressionType&>(m_matrix += extendedTo(other.derived()));
     }
 
     /** Substracts the vector \a other to each subvector of \c *this */
@@ -439,8 +440,29 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
     ExpressionType& operator-=(const DenseBase<OtherDerived>& other)
     {
       EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
-      for(Index j=0; j<subVectors(); ++j)
-        subVector(j) -= other.derived();
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
+      return const_cast<ExpressionType&>(m_matrix -= extendedTo(other.derived()));
+    }
+
+    /** Multiples each subvector of \c *this by the vector \a other */
+    template<typename OtherDerived>
+    ExpressionType& operator*=(const DenseBase<OtherDerived>& other)
+    {
+      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_ARRAYXPR(ExpressionType)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
+      m_matrix *= extendedTo(other.derived());
+      return const_cast<ExpressionType&>(m_matrix);
+    }
+
+    /** Divides each subvector of \c *this by the vector \a other */
+    template<typename OtherDerived>
+    ExpressionType& operator/=(const DenseBase<OtherDerived>& other)
+    {
+      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_ARRAYXPR(ExpressionType)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
+      m_matrix /= extendedTo(other.derived());
       return const_cast<ExpressionType&>(m_matrix);
     }
 
@@ -451,7 +473,8 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
                   const typename ExtendedType<OtherDerived>::Type>
     operator+(const DenseBase<OtherDerived>& other) const
     {
-      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived);
+      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
       return m_matrix + extendedTo(other.derived());
     }
 
@@ -462,10 +485,39 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
                   const typename ExtendedType<OtherDerived>::Type>
     operator-(const DenseBase<OtherDerived>& other) const
     {
-      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived);
+      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
       return m_matrix - extendedTo(other.derived());
     }
 
+    /** Returns the expression where each subvector is the product of the vector \a other
+      * by the corresponding subvector of \c *this */
+    template<typename OtherDerived> EIGEN_STRONG_INLINE
+    CwiseBinaryOp<internal::scalar_product_op<Scalar>,
+                  const ExpressionTypeNestedCleaned,
+                  const typename ExtendedType<OtherDerived>::Type>
+    operator*(const DenseBase<OtherDerived>& other) const
+    {
+      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_ARRAYXPR(ExpressionType)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
+      return m_matrix * extendedTo(other.derived());
+    }
+
+    /** Returns the expression where each subvector is the quotient of the corresponding
+      * subvector of \c *this by the vector \a other */
+    template<typename OtherDerived>
+    CwiseBinaryOp<internal::scalar_quotient_op<Scalar>,
+                  const ExpressionTypeNestedCleaned,
+                  const typename ExtendedType<OtherDerived>::Type>
+    operator/(const DenseBase<OtherDerived>& other) const
+    {
+      EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
+      EIGEN_STATIC_ASSERT_ARRAYXPR(ExpressionType)
+      EIGEN_STATIC_ASSERT_SAME_XPR_KIND(ExpressionType, OtherDerived)
+      return m_matrix / extendedTo(other.derived());
+    }
+
 /////////// Geometry module ///////////
 
     #if EIGEN2_SUPPORT_STAGE > STAGE20_RESOLVE_API_CONFLICTS
@@ -509,7 +561,7 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
   * Example: \include MatrixBase_colwise.cpp
   * Output: \verbinclude MatrixBase_colwise.out
   *
-  * \sa rowwise(), class VectorwiseOp
+  * \sa rowwise(), class VectorwiseOp, \ref TutorialReductionsVisitorsBroadcasting
   */
 template<typename Derived>
 inline const typename DenseBase<Derived>::ConstColwiseReturnType
@@ -520,7 +572,7 @@ DenseBase<Derived>::colwise() const
 
 /** \returns a writable VectorwiseOp wrapper of *this providing additional partial reduction operations
   *
-  * \sa rowwise(), class VectorwiseOp
+  * \sa rowwise(), class VectorwiseOp, \ref TutorialReductionsVisitorsBroadcasting
   */
 template<typename Derived>
 inline typename DenseBase<Derived>::ColwiseReturnType
@@ -534,7 +586,7 @@ DenseBase<Derived>::colwise()
   * Example: \include MatrixBase_rowwise.cpp
   * Output: \verbinclude MatrixBase_rowwise.out
   *
-  * \sa colwise(), class VectorwiseOp
+  * \sa colwise(), class VectorwiseOp, \ref TutorialReductionsVisitorsBroadcasting
   */
 template<typename Derived>
 inline const typename DenseBase<Derived>::ConstRowwiseReturnType
@@ -545,7 +597,7 @@ DenseBase<Derived>::rowwise() const
 
 /** \returns a writable VectorwiseOp wrapper of *this providing additional partial reduction operations
   *
-  * \sa colwise(), class VectorwiseOp
+  * \sa colwise(), class VectorwiseOp, \ref TutorialReductionsVisitorsBroadcasting
   */
 template<typename Derived>
 inline typename DenseBase<Derived>::RowwiseReturnType

Eigen/src/Core/Visitor.h

@@ -35,7 +35,7 @@ struct visitor_impl
     row = (UnrollCount-1) % Derived::RowsAtCompileTime
   };
 
-  inline static void run(const Derived &mat, Visitor& visitor)
+  static inline void run(const Derived &mat, Visitor& visitor)
   {
     visitor_impl<Visitor, Derived, UnrollCount-1>::run(mat, visitor);
     visitor(mat.coeff(row, col), row, col);
@@ -45,7 +45,7 @@ struct visitor_impl
 template<typename Visitor, typename Derived>
 struct visitor_impl<Visitor, Derived, 1>
 {
-  inline static void run(const Derived &mat, Visitor& visitor)
+  static inline void run(const Derived &mat, Visitor& visitor)
   {
     return visitor.init(mat.coeff(0, 0), 0, 0);
   }
@@ -55,7 +55,7 @@ template<typename Visitor, typename Derived>
 struct visitor_impl<Visitor, Derived, Dynamic>
 {
   typedef typename Derived::Index Index;
-  inline static void run(const Derived& mat, Visitor& visitor)
+  static inline void run(const Derived& mat, Visitor& visitor)
   {
     visitor.init(mat.coeff(0,0), 0, 0);
     for(Index i = 1; i < mat.rows(); ++i)

Eigen/src/Core/arch/AltiVec/Complex.h

@@ -48,6 +48,7 @@ template<> struct packet_traits<std::complex<float> >  : default_packet_traits
   typedef Packet2cf type;
   enum {
     Vectorizable = 1,
+    AlignedOnScalar = 1,
     size = 2,
 
     HasAdd    = 1,
@@ -69,19 +70,17 @@ template<> EIGEN_STRONG_INLINE Packet2cf pset1<Packet2cf>(const std::complex<flo
 {
   Packet2cf res;
   /* On AltiVec we cannot load 64-bit registers, so wa have to take care of alignment */
-  if ((ptrdiff_t)&from % 16 == 0) {
-    res.v = pload((const float *)&from);
-    res.v = vec_perm(res.v, res.v, p16uc_PSET_HI);
-  } else {
-    res.v = ploadu((const float *)&from);
-    res.v = vec_perm(res.v, res.v, p16uc_PSET_LO);
-  }
+  if((ptrdiff_t(&from) % 16) == 0)
+    res.v = pload<Packet4f>((const float *)&from);
+  else
+    res.v = ploadu<Packet4f>((const float *)&from);
+  res.v = vec_perm(res.v, res.v, p16uc_PSET_HI);
   return res;
 }
 
 template<> EIGEN_STRONG_INLINE Packet2cf padd<Packet2cf>(const Packet2cf& a, const Packet2cf& b) { return Packet2cf(vec_add(a.v,b.v)); }
 template<> EIGEN_STRONG_INLINE Packet2cf psub<Packet2cf>(const Packet2cf& a, const Packet2cf& b) { return Packet2cf(vec_sub(a.v,b.v)); }
-template<> EIGEN_STRONG_INLINE Packet2cf pnegate(const Packet2cf& a) { return Packet2cf(psub<Packet4f>(p4f_ZERO, a.v)); }
+template<> EIGEN_STRONG_INLINE Packet2cf pnegate(const Packet2cf& a) { return Packet2cf(pnegate(a.v)); }
 template<> EIGEN_STRONG_INLINE Packet2cf pconj(const Packet2cf& a) { return Packet2cf((Packet4f)vec_xor((Packet4ui)a.v, p4ui_CONJ_XOR)); }
 
 template<> EIGEN_STRONG_INLINE Packet2cf pmul<Packet2cf>(const Packet2cf& a, const Packet2cf& b)
@@ -108,8 +107,13 @@ template<> EIGEN_STRONG_INLINE Packet2cf por    <Packet2cf>(const Packet2cf& a,
 template<> EIGEN_STRONG_INLINE Packet2cf pxor   <Packet2cf>(const Packet2cf& a, const Packet2cf& b) { return Packet2cf(vec_xor(a.v,b.v)); }
 template<> EIGEN_STRONG_INLINE Packet2cf pandnot<Packet2cf>(const Packet2cf& a, const Packet2cf& b) { return Packet2cf(vec_and(a.v, vec_nor(b.v,b.v))); }
 
-template<> EIGEN_STRONG_INLINE Packet2cf pload <std::complex<float> >(const std::complex<float>* from) { EIGEN_DEBUG_ALIGNED_LOAD return Packet2cf(pload((const float*)from)); }
-template<> EIGEN_STRONG_INLINE Packet2cf ploadu<std::complex<float> >(const std::complex<float>* from) { EIGEN_DEBUG_UNALIGNED_LOAD return Packet2cf(ploadu((const float*)from)); }
+template<> EIGEN_STRONG_INLINE Packet2cf pload <Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_ALIGNED_LOAD return Packet2cf(pload<Packet4f>((const float*)from)); }
+template<> EIGEN_STRONG_INLINE Packet2cf ploadu<Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_UNALIGNED_LOAD return Packet2cf(ploadu<Packet4f>((const float*)from)); }
+
+template<> EIGEN_STRONG_INLINE Packet2cf ploaddup<Packet2cf>(const std::complex<float>*     from)
+{
+  return pset1<Packet2cf>(*from);
+}
 
 template<> EIGEN_STRONG_INLINE void pstore <std::complex<float> >(std::complex<float> *   to, const Packet2cf& from) { EIGEN_DEBUG_ALIGNED_STORE pstore((float*)to, from.v); }
 template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<float> >(std::complex<float> *   to, const Packet2cf& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((float*)to, from.v); }
@@ -136,7 +140,7 @@ template<> EIGEN_STRONG_INLINE std::complex<float> predux<Packet2cf>(const Packe
   Packet4f b;
   b = (Packet4f) vec_sld(a.v, a.v, 8);
   b = padd(a.v, b);
-  return pfirst(Packet2cf(sum));
+  return pfirst(Packet2cf(b));
 }
 
 template<> EIGEN_STRONG_INLINE Packet2cf preduxp<Packet2cf>(const Packet2cf* vecs)
@@ -164,7 +168,7 @@ template<> EIGEN_STRONG_INLINE std::complex<float> predux_mul<Packet2cf>(const P
 template<int Offset>
 struct palign_impl<Offset,Packet2cf>
 {
-  EIGEN_STRONG_INLINE static void run(Packet2cf& first, const Packet2cf& second)
+  static EIGEN_STRONG_INLINE void run(Packet2cf& first, const Packet2cf& second)
   {
     if (Offset==1)
     {
@@ -180,7 +184,7 @@ template<> struct conj_helper<Packet2cf, Packet2cf, false,true>
 
   EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
   {
-    return pmul(a, pconj(b));
+    return internal::pmul(a, pconj(b));
   }
 };
 
@@ -191,7 +195,7 @@ template<> struct conj_helper<Packet2cf, Packet2cf, true,false>
 
   EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
   {
-    return pmul(pconj(a), b);
+    return internal::pmul(pconj(a), b);
   }
 };
 
@@ -202,7 +206,7 @@ template<> struct conj_helper<Packet2cf, Packet2cf, true,true>
 
   EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
   {
-    return pconj(pmul(a, b));
+    return pconj(internal::pmul(a, b));
   }
 };
 
@@ -214,6 +218,11 @@ template<> EIGEN_STRONG_INLINE Packet2cf pdiv<Packet2cf>(const Packet2cf& a, con
   return Packet2cf(pdiv(res.v, vec_add(s,vec_perm(s, s, p16uc_COMPLEX_REV))));
 }
 
+template<> EIGEN_STRONG_INLINE Packet2cf pcplxflip<Packet2cf>(const Packet2cf& x)
+{
+  return Packet2cf(vec_perm(x.v, x.v, p16uc_COMPLEX_REV));
+}
+
 } // end namespace internal
 
 #endif // EIGEN_COMPLEX_ALTIVEC_H

Eigen/src/Core/arch/AltiVec/PacketMath.h

@@ -73,6 +73,7 @@ static Packet4f p4f_COUNTDOWN = { 3.0, 2.0, 1.0, 0.0 };
 static Packet4i p4i_COUNTDOWN = { 3, 2, 1, 0 };
 static Packet16uc p16uc_REVERSE = {12,13,14,15, 8,9,10,11, 4,5,6,7, 0,1,2,3};
 static Packet16uc p16uc_FORWARD = vec_lvsl(0, (float*)0);
+static Packet16uc p16uc_DUPLICATE = {0,1,2,3, 0,1,2,3, 4,5,6,7, 4,5,6,7};
 
 static _EIGEN_DECLARE_CONST_FAST_Packet4f(ZERO, 0);
 static _EIGEN_DECLARE_CONST_FAST_Packet4i(ZERO, 0);
@@ -292,6 +293,21 @@ template<> EIGEN_STRONG_INLINE Packet4i ploadu<Packet4i>(const int* from)
   return (Packet4i) vec_perm(MSQ, LSQ, mask);    // align the data
 }
 
+template<> EIGEN_STRONG_INLINE Packet4f ploaddup<Packet4f>(const float*   from)
+{
+  Packet4f p;
+  if((ptrdiff_t(&from) % 16) == 0)  p = pload<Packet4f>(from);
+  else                              p = ploadu<Packet4f>(from);
+  return vec_perm(p, p, p16uc_DUPLICATE);
+}
+template<> EIGEN_STRONG_INLINE Packet4i ploaddup<Packet4i>(const int*     from)
+{
+  Packet4i p;
+  if((ptrdiff_t(&from) % 16) == 0)  p = pload<Packet4i>(from);
+  else                              p = ploadu<Packet4i>(from);
+  return vec_perm(p, p, p16uc_DUPLICATE);
+}
+
 template<> EIGEN_STRONG_INLINE void pstore<float>(float*   to, const Packet4f& from) { EIGEN_DEBUG_ALIGNED_STORE vec_st(from, 0, to); }
 template<> EIGEN_STRONG_INLINE void pstore<int>(int*       to, const Packet4i& from) { EIGEN_DEBUG_ALIGNED_STORE vec_st(from, 0, to); }
 
@@ -471,7 +487,7 @@ template<> EIGEN_STRONG_INLINE int predux_max<Packet4i>(const Packet4i& a)
 template<int Offset>
 struct palign_impl<Offset,Packet4f>
 {
-  EIGEN_STRONG_INLINE static void run(Packet4f& first, const Packet4f& second)
+  static EIGEN_STRONG_INLINE void run(Packet4f& first, const Packet4f& second)
   {
     if (Offset!=0)
       first = vec_sld(first, second, Offset*4);
@@ -481,7 +497,7 @@ struct palign_impl<Offset,Packet4f>
 template<int Offset>
 struct palign_impl<Offset,Packet4i>
 {
-  EIGEN_STRONG_INLINE static void run(Packet4i& first, const Packet4i& second)
+  static EIGEN_STRONG_INLINE void run(Packet4i& first, const Packet4i& second)
   {
     if (Offset!=0)
       first = vec_sld(first, second, Offset*4);

Eigen/src/Core/arch/NEON/Complex.h

@@ -27,8 +27,8 @@
 
 namespace internal {
 
-static uint32x4_t p4ui_CONJ_XOR = { 0x00000000, 0x80000000, 0x00000000, 0x80000000 };
-static uint32x2_t p2ui_CONJ_XOR = { 0x00000000, 0x80000000 };
+static uint32x4_t p4ui_CONJ_XOR = EIGEN_INIT_NEON_PACKET4(0x00000000, 0x80000000, 0x00000000, 0x80000000);
+static uint32x2_t p2ui_CONJ_XOR = EIGEN_INIT_NEON_PACKET2(0x00000000, 0x80000000);
 
 //---------- float ----------
 struct Packet2cf
@@ -43,6 +43,7 @@ template<> struct packet_traits<std::complex<float> >  : default_packet_traits
   typedef Packet2cf type;
   enum {
     Vectorizable = 1,
+    AlignedOnScalar = 1,
     size = 2,
 
     HasAdd    = 1,
@@ -120,6 +121,8 @@ template<> EIGEN_STRONG_INLINE Packet2cf pandnot<Packet2cf>(const Packet2cf& a,
 template<> EIGEN_STRONG_INLINE Packet2cf pload<Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_ALIGNED_LOAD return Packet2cf(pload<Packet4f>((const float*)from)); }
 template<> EIGEN_STRONG_INLINE Packet2cf ploadu<Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_UNALIGNED_LOAD return Packet2cf(ploadu<Packet4f>((const float*)from)); }
 
+template<> EIGEN_STRONG_INLINE Packet2cf ploaddup<Packet2cf>(const std::complex<float>* from) { return pset1<Packet2cf>(*from); }
+
 template<> EIGEN_STRONG_INLINE void pstore <std::complex<float> >(std::complex<float> *   to, const Packet2cf& from) { EIGEN_DEBUG_ALIGNED_STORE pstore((float*)to, from.v); }
 template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<float> >(std::complex<float> *   to, const Packet2cf& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((float*)to, from.v); }
 
@@ -144,7 +147,7 @@ template<> EIGEN_STRONG_INLINE Packet2cf preverse(const Packet2cf& a)
   return Packet2cf(a_r128);
 }
 
-EIGEN_STRONG_INLINE Packet2cf pcplxflip(const Packet2cf& a)
+template<> EIGEN_STRONG_INLINE Packet2cf pcplxflip<Packet2cf>(const Packet2cf& a)
 {
   return Packet2cf(vrev64q_f32(a.v));
 }
@@ -220,7 +223,7 @@ template<> struct conj_helper<Packet2cf, Packet2cf, false,true>
 
   EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
   {
-    return pmul(a, pconj(b));
+    return internal::pmul(a, pconj(b));
   }
 };
 
@@ -231,7 +234,7 @@ template<> struct conj_helper<Packet2cf, Packet2cf, true,false>
 
   EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
   {
-    return pmul(pconj(a), b);
+    return internal::pmul(pconj(a), b);
   }
 };
 
@@ -242,7 +245,7 @@ template<> struct conj_helper<Packet2cf, Packet2cf, true,true>
 
   EIGEN_STRONG_INLINE Packet2cf pmul(const Packet2cf& a, const Packet2cf& b) const
   {
-    return pconj(pmul(a, b));
+    return pconj(internal::pmul(a, b));
   }
 };
 

Eigen/src/Core/arch/NEON/PacketMath.h

@@ -41,7 +41,7 @@ namespace internal {
 
 typedef float32x4_t Packet4f;
 typedef int32x4_t   Packet4i;
-typedef uint32x4_t   Packet4ui;
+typedef uint32x4_t  Packet4ui;
 
 #define _EIGEN_DECLARE_CONST_Packet4f(NAME,X) \
   const Packet4f p4f_##NAME = pset1<Packet4f>(X)
@@ -52,6 +52,16 @@ typedef uint32x4_t   Packet4ui;
 #define _EIGEN_DECLARE_CONST_Packet4i(NAME,X) \
   const Packet4i p4i_##NAME = pset1<Packet4i>(X)
 
+#if defined(__llvm__) && !defined(__clang__)
+  //Special treatment for Apple's llvm-gcc, its NEON packet types are unions
+  #define EIGEN_INIT_NEON_PACKET2(X, Y)       {{X, Y}}
+  #define EIGEN_INIT_NEON_PACKET4(X, Y, Z, W) {{X, Y, Z, W}}
+#else
+  //Default initializer for packets
+  #define EIGEN_INIT_NEON_PACKET2(X, Y)       {X, Y}
+  #define EIGEN_INIT_NEON_PACKET4(X, Y, Z, W) {X, Y, Z, W}
+#endif
+    
 #ifndef __pld
 #define __pld(x) asm volatile ( "   pld [%[addr]]\n" :: [addr] "r" (x) : "cc" );
 #endif
@@ -84,8 +94,8 @@ template<> struct packet_traits<int>    : default_packet_traits
   };
 };
 
-#if (defined __GNUC__) && (!(EIGEN_GNUC_AT_LEAST(4,4)))
-// workaround gcc 4.2 and 4.3 compilatin issue
+#if EIGEN_GNUC_AT_MOST(4,4) && !defined(__llvm__)
+// workaround gcc 4.2, 4.3 and 4.4 compilatin issue
 EIGEN_STRONG_INLINE float32x4_t vld1q_f32(const float* x) { return ::vld1q_f32((const float32_t*)x); }
 EIGEN_STRONG_INLINE float32x2_t vld1_f32 (const float* x) { return ::vld1_f32 ((const float32_t*)x); }
 EIGEN_STRONG_INLINE void        vst1q_f32(float* to, float32x4_t from) { ::vst1q_f32((float32_t*)to,from); }
@@ -100,12 +110,12 @@ template<> EIGEN_STRONG_INLINE Packet4i pset1<Packet4i>(const int&    from)   {
 
 template<> EIGEN_STRONG_INLINE Packet4f plset<float>(const float& a)
 {
-  Packet4f countdown = { 3, 2, 1, 0 };
+  Packet4f countdown = EIGEN_INIT_NEON_PACKET4(0, 1, 2, 3);
   return vaddq_f32(pset1<Packet4f>(a), countdown);
 }
 template<> EIGEN_STRONG_INLINE Packet4i plset<int>(const int& a)
 {
-  Packet4i countdown = { 3, 2, 1, 0 };
+  Packet4i countdown = EIGEN_INIT_NEON_PACKET4(0, 1, 2, 3);
   return vaddq_s32(pset1<Packet4i>(a), countdown);
 }
 
@@ -191,14 +201,14 @@ template<> EIGEN_STRONG_INLINE Packet4f ploaddup<Packet4f>(const float*   from)
 {
   float32x2_t lo, hi;
   lo = vdup_n_f32(*from);
-  hi = vdup_n_f32(*from);
+  hi = vdup_n_f32(*(from+1));
   return vcombine_f32(lo, hi);
 }
 template<> EIGEN_STRONG_INLINE Packet4i ploaddup<Packet4i>(const int*     from)
 {
   int32x2_t lo, hi;
   lo = vdup_n_s32(*from);
-  hi = vdup_n_s32(*from);
+  hi = vdup_n_s32(*(from+1));
   return vcombine_s32(lo, hi);
 }
 
@@ -395,25 +405,29 @@ template<> EIGEN_STRONG_INLINE int predux_max<Packet4i>(const Packet4i& a)
   return s[0];
 }
 
-template<int Offset>
-struct palign_impl<Offset,Packet4f>
-{
-  EIGEN_STRONG_INLINE static void run(Packet4f& first, const Packet4f& second)
-  {
-    if (Offset!=0)
-      first = vextq_f32(first, second, Offset);
-  }
-};
+// this PALIGN_NEON business is to work around a bug in LLVM Clang 3.0 causing incorrect compilation errors,
+// see bug 347 and this LLVM bug: http://llvm.org/bugs/show_bug.cgi?id=11074
+#define PALIGN_NEON(Offset,Type,Command) \
+template<>\
+struct palign_impl<Offset,Type>\
+{\
+    EIGEN_STRONG_INLINE static void run(Type& first, const Type& second)\
+    {\
+        if (Offset!=0)\
+            first = Command(first, second, Offset);\
+    }\
+};\
 
-template<int Offset>
-struct palign_impl<Offset,Packet4i>
-{
-  EIGEN_STRONG_INLINE static void run(Packet4i& first, const Packet4i& second)
-  {
-    if (Offset!=0)
-      first = vextq_s32(first, second, Offset);
-  }
-};
+PALIGN_NEON(0,Packet4f,vextq_f32)
+PALIGN_NEON(1,Packet4f,vextq_f32)
+PALIGN_NEON(2,Packet4f,vextq_f32)
+PALIGN_NEON(3,Packet4f,vextq_f32)
+PALIGN_NEON(0,Packet4i,vextq_s32)
+PALIGN_NEON(1,Packet4i,vextq_s32)
+PALIGN_NEON(2,Packet4i,vextq_s32)
+PALIGN_NEON(3,Packet4i,vextq_s32)
+    
+#undef PALIGN_NEON
 
 } // end namespace internal
 

Eigen/src/Core/arch/SSE/Complex.h

@@ -97,23 +97,30 @@ template<> EIGEN_STRONG_INLINE Packet2cf pandnot<Packet2cf>(const Packet2cf& a,
 template<> EIGEN_STRONG_INLINE Packet2cf pload <Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_ALIGNED_LOAD return Packet2cf(pload<Packet4f>(&real_ref(*from))); }
 template<> EIGEN_STRONG_INLINE Packet2cf ploadu<Packet2cf>(const std::complex<float>* from) { EIGEN_DEBUG_UNALIGNED_LOAD return Packet2cf(ploadu<Packet4f>(&real_ref(*from))); }
 
+template<> EIGEN_STRONG_INLINE Packet2cf pset1<Packet2cf>(const std::complex<float>&  from)
+{
+  Packet2cf res;
+  #if EIGEN_GNUC_AT_MOST(4,2)
+  // workaround annoying "may be used uninitialized in this function" warning with gcc 4.2
+  res.v = _mm_loadl_pi(_mm_set1_ps(0.0f), reinterpret_cast<const __m64*>(&from));
+  #else
+  res.v = _mm_loadl_pi(res.v, (const __m64*)&from);
+  #endif
+  return Packet2cf(_mm_movelh_ps(res.v,res.v));
+}
+
+template<> EIGEN_STRONG_INLINE Packet2cf ploaddup<Packet2cf>(const std::complex<float>* from) { return pset1<Packet2cf>(*from); }
+
 template<> EIGEN_STRONG_INLINE void pstore <std::complex<float> >(std::complex<float> *   to, const Packet2cf& from) { EIGEN_DEBUG_ALIGNED_STORE pstore(&real_ref(*to), from.v); }
 template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<float> >(std::complex<float> *   to, const Packet2cf& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu(&real_ref(*to), from.v); }
 
 template<> EIGEN_STRONG_INLINE void prefetch<std::complex<float> >(const std::complex<float> *   addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); }
 
-template<> EIGEN_STRONG_INLINE Packet2cf pset1<Packet2cf>(const std::complex<float>&  from)
-{
-  Packet2cf res;
-  res.v = _mm_loadl_pi(res.v, (const __m64*)&from);
-  return Packet2cf(_mm_movelh_ps(res.v,res.v));
-}
-
 template<> EIGEN_STRONG_INLINE std::complex<float>  pfirst<Packet2cf>(const Packet2cf& a)
 {
-  #if (defined __GNUC__) && (__GNUC__==4) && (__GNUC_MINOR__==2) && (__GNUC_PATCHLEVEL__<=3)
-  // workaround gcc 4.2.1 ICE (mac's gcc version) - I'm not sure how the 4.2.2 and 4.2.3 deal with it, but 4.2.4 works well.
-  // this is not performance wise ideal, but who cares...
+  #if EIGEN_GNUC_AT_MOST(4,3)
+  // Workaround gcc 4.2 ICE - this is not performance wise ideal, but who cares...
+  // This workaround also fix invalid code generation with gcc 4.3
   EIGEN_ALIGN16 std::complex<float> res[2];
   _mm_store_ps((float*)res, a.v);
   return res[0];
@@ -144,7 +151,7 @@ template<> EIGEN_STRONG_INLINE std::complex<float> predux_mul<Packet2cf>(const P
 template<int Offset>
 struct palign_impl<Offset,Packet2cf>
 {
-  EIGEN_STRONG_INLINE static void run(Packet2cf& first, const Packet2cf& second)
+  static EIGEN_STRONG_INLINE void run(Packet2cf& first, const Packet2cf& second)
   {
     if (Offset==1)
     {
@@ -308,6 +315,8 @@ template<> EIGEN_STRONG_INLINE Packet1cd ploadu<Packet1cd>(const std::complex<do
 template<> EIGEN_STRONG_INLINE Packet1cd pset1<Packet1cd>(const std::complex<double>&  from)
 { /* here we really have to use unaligned loads :( */ return ploadu<Packet1cd>(&from); }
 
+template<> EIGEN_STRONG_INLINE Packet1cd ploaddup<Packet1cd>(const std::complex<double>* from) { return pset1<Packet1cd>(*from); }
+
 // FIXME force unaligned store, this is a temporary fix
 template<> EIGEN_STRONG_INLINE void pstore <std::complex<double> >(std::complex<double> *   to, const Packet1cd& from) { EIGEN_DEBUG_ALIGNED_STORE pstore((double*)to, from.v); }
 template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<double> >(std::complex<double> *   to, const Packet1cd& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((double*)to, from.v); }
@@ -341,7 +350,7 @@ template<> EIGEN_STRONG_INLINE std::complex<double> predux_mul<Packet1cd>(const
 template<int Offset>
 struct palign_impl<Offset,Packet1cd>
 {
-  EIGEN_STRONG_INLINE static void run(Packet1cd& /*first*/, const Packet1cd& /*second*/)
+  static EIGEN_STRONG_INLINE void run(Packet1cd& /*first*/, const Packet1cd& /*second*/)
   {
     // FIXME is it sure we never have to align a Packet1cd?
     // Even though a std::complex<double> has 16 bytes, it is not necessarily aligned on a 16 bytes boundary...

Eigen/src/Core/arch/SSE/PacketMath.h

@@ -110,23 +110,18 @@ template<> struct unpacket_traits<Packet4f> { typedef float  type; enum {size=4}
 template<> struct unpacket_traits<Packet2d> { typedef double type; enum {size=2}; };
 template<> struct unpacket_traits<Packet4i> { typedef int    type; enum {size=4}; };
 
-#ifdef __GNUC__
-// Sometimes GCC implements _mm_set1_p* using multiple moves,
-// that is inefficient :( (e.g., see gemm_pack_rhs)
-template<> EIGEN_STRONG_INLINE Packet4f pset1<Packet4f>(const float&  from) {
-  Packet4f res = _mm_set_ss(from);
-  return vec4f_swizzle1(res,0,0,0,0);
-}
-template<> EIGEN_STRONG_INLINE Packet2d pset1<Packet2d>(const double&  from) {
-  // NOTE the SSE3 intrinsic _mm_loaddup_pd is never faster but sometimes much slower
-  Packet2d res = _mm_set_sd(from);
-  return vec2d_swizzle1(res, 0, 0);
-}
+#if defined(_MSC_VER) && (_MSC_VER==1500)
+// Workaround MSVC 9 internal compiler error.
+// TODO: It has been detected with win64 builds (amd64), so let's check whether it also happens in 32bits+SSE mode
+// TODO: let's check whether there does not exist a better fix, like adding a pset0() function. (it crashed on pset1(0)).
+template<> EIGEN_STRONG_INLINE Packet4f pset1<Packet4f>(const float&  from) { return _mm_set_ps(from,from,from,from); }
+template<> EIGEN_STRONG_INLINE Packet2d pset1<Packet2d>(const double& from) { return _mm_set_pd(from,from); }
+template<> EIGEN_STRONG_INLINE Packet4i pset1<Packet4i>(const int&    from) { return _mm_set_epi32(from,from,from,from); }
 #else
 template<> EIGEN_STRONG_INLINE Packet4f pset1<Packet4f>(const float&  from) { return _mm_set1_ps(from); }
 template<> EIGEN_STRONG_INLINE Packet2d pset1<Packet2d>(const double& from) { return _mm_set1_pd(from); }
-#endif
 template<> EIGEN_STRONG_INLINE Packet4i pset1<Packet4i>(const int&    from) { return _mm_set1_epi32(from); }
+#endif
 
 template<> EIGEN_STRONG_INLINE Packet4f plset<float>(const float& a) { return _mm_add_ps(pset1<Packet4f>(a), _mm_set_ps(3,2,1,0)); }
 template<> EIGEN_STRONG_INLINE Packet2d plset<double>(const double& a) { return _mm_add_pd(pset1<Packet2d>(a),_mm_set_pd(1,0)); }
@@ -245,35 +240,58 @@ template<> EIGEN_STRONG_INLINE Packet4i pload<Packet4i>(const int*     from) { E
 // a correct instruction dependency.
 // TODO: do the same for MSVC (ICC is compatible)
 // NOTE: with the code below, MSVC's compiler crashes!
+
+#if defined(__GNUC__) && defined(__i386__)
+  // bug 195: gcc/i386 emits weird x87 fldl/fstpl instructions for _mm_load_sd
+  #define EIGEN_AVOID_CUSTOM_UNALIGNED_LOADS 1
+#elif defined(__clang__)
+  // bug 201: Segfaults in __mm_loadh_pd with clang 2.8
+  #define EIGEN_AVOID_CUSTOM_UNALIGNED_LOADS 1
+#else
+  #define EIGEN_AVOID_CUSTOM_UNALIGNED_LOADS 0
+#endif
+
 template<> EIGEN_STRONG_INLINE Packet4f ploadu<Packet4f>(const float* from)
 {
   EIGEN_DEBUG_UNALIGNED_LOAD
+#if EIGEN_AVOID_CUSTOM_UNALIGNED_LOADS
+  return _mm_loadu_ps(from);
+#else
   __m128d res;
   res =  _mm_load_sd((const double*)(from)) ;
   res =  _mm_loadh_pd(res, (const double*)(from+2)) ;
   return _mm_castpd_ps(res);
+#endif
 }
 template<> EIGEN_STRONG_INLINE Packet2d ploadu<Packet2d>(const double* from)
 {
   EIGEN_DEBUG_UNALIGNED_LOAD
+#if EIGEN_AVOID_CUSTOM_UNALIGNED_LOADS
+  return _mm_loadu_pd(from);
+#else
   __m128d res;
   res = _mm_load_sd(from) ;
   res = _mm_loadh_pd(res,from+1);
   return res;
+#endif
 }
 template<> EIGEN_STRONG_INLINE Packet4i ploadu<Packet4i>(const int* from)
 {
   EIGEN_DEBUG_UNALIGNED_LOAD
+#if EIGEN_AVOID_CUSTOM_UNALIGNED_LOADS
+  return _mm_loadu_si128(reinterpret_cast<const Packet4i*>(from));
+#else
   __m128d res;
   res =  _mm_load_sd((const double*)(from)) ;
   res =  _mm_loadh_pd(res, (const double*)(from+2)) ;
   return _mm_castpd_si128(res);
+#endif
 }
 #endif
 
 template<> EIGEN_STRONG_INLINE Packet4f ploaddup<Packet4f>(const float*   from)
 {
-  return vec4f_swizzle1(_mm_castpd_ps(_mm_load_sd((const double*)from)), 0, 0, 1, 1);
+  return vec4f_swizzle1(_mm_castpd_ps(_mm_load_sd(reinterpret_cast<const double*>(from))), 0, 0, 1, 1);
 }
 template<> EIGEN_STRONG_INLINE Packet2d ploaddup<Packet2d>(const double*  from)
 { return pset1<Packet2d>(from[0]); }
@@ -293,8 +311,21 @@ template<> EIGEN_STRONG_INLINE void pstoreu<double>(double* to, const Packet2d&
   _mm_storel_pd((to), from);
   _mm_storeh_pd((to+1), from);
 }
-template<> EIGEN_STRONG_INLINE void pstoreu<float>(float*  to, const Packet4f& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((double*)to, _mm_castps_pd(from)); }
-template<> EIGEN_STRONG_INLINE void pstoreu<int>(int*      to, const Packet4i& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((double*)to, _mm_castsi128_pd(from)); }
+template<> EIGEN_STRONG_INLINE void pstoreu<float>(float*  to, const Packet4f& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu(reinterpret_cast<double*>(to), _mm_castps_pd(from)); }
+template<> EIGEN_STRONG_INLINE void pstoreu<int>(int*      to, const Packet4i& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu(reinterpret_cast<double*>(to), _mm_castsi128_pd(from)); }
+
+// some compilers might be tempted to perform multiple moves instead of using a vector path.
+template<> EIGEN_STRONG_INLINE void pstore1<Packet4f>(float* to, const float& a)
+{
+  Packet4f pa = _mm_set_ss(a);
+  pstore(to, vec4f_swizzle1(pa,0,0,0,0));
+}
+// some compilers might be tempted to perform multiple moves instead of using a vector path.
+template<> EIGEN_STRONG_INLINE void pstore1<Packet2d>(double* to, const double& a)
+{
+  Packet2d pa = _mm_set_sd(a);
+  pstore(to, vec2d_swizzle1(pa,0,0));
+}
 
 template<> EIGEN_STRONG_INLINE void prefetch<float>(const float*   addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); }
 template<> EIGEN_STRONG_INLINE void prefetch<double>(const double* addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); }
@@ -519,7 +550,7 @@ template<> EIGEN_STRONG_INLINE int predux_max<Packet4i>(const Packet4i& a)
 template<int Offset>
 struct palign_impl<Offset,Packet4f>
 {
-  EIGEN_STRONG_INLINE static void run(Packet4f& first, const Packet4f& second)
+  static EIGEN_STRONG_INLINE void run(Packet4f& first, const Packet4f& second)
   {
     if (Offset!=0)
       first = _mm_castsi128_ps(_mm_alignr_epi8(_mm_castps_si128(second), _mm_castps_si128(first), Offset*4));
@@ -529,7 +560,7 @@ struct palign_impl<Offset,Packet4f>
 template<int Offset>
 struct palign_impl<Offset,Packet4i>
 {
-  EIGEN_STRONG_INLINE static void run(Packet4i& first, const Packet4i& second)
+  static EIGEN_STRONG_INLINE void run(Packet4i& first, const Packet4i& second)
   {
     if (Offset!=0)
       first = _mm_alignr_epi8(second,first, Offset*4);
@@ -539,7 +570,7 @@ struct palign_impl<Offset,Packet4i>
 template<int Offset>
 struct palign_impl<Offset,Packet2d>
 {
-  EIGEN_STRONG_INLINE static void run(Packet2d& first, const Packet2d& second)
+  static EIGEN_STRONG_INLINE void run(Packet2d& first, const Packet2d& second)
   {
     if (Offset==1)
       first = _mm_castsi128_pd(_mm_alignr_epi8(_mm_castpd_si128(second), _mm_castpd_si128(first), 8));
@@ -550,7 +581,7 @@ struct palign_impl<Offset,Packet2d>
 template<int Offset>
 struct palign_impl<Offset,Packet4f>
 {
-  EIGEN_STRONG_INLINE static void run(Packet4f& first, const Packet4f& second)
+  static EIGEN_STRONG_INLINE void run(Packet4f& first, const Packet4f& second)
   {
     if (Offset==1)
     {
@@ -573,7 +604,7 @@ struct palign_impl<Offset,Packet4f>
 template<int Offset>
 struct palign_impl<Offset,Packet4i>
 {
-  EIGEN_STRONG_INLINE static void run(Packet4i& first, const Packet4i& second)
+  static EIGEN_STRONG_INLINE void run(Packet4i& first, const Packet4i& second)
   {
     if (Offset==1)
     {
@@ -596,7 +627,7 @@ struct palign_impl<Offset,Packet4i>
 template<int Offset>
 struct palign_impl<Offset,Packet2d>
 {
-  EIGEN_STRONG_INLINE static void run(Packet2d& first, const Packet2d& second)
+  static EIGEN_STRONG_INLINE void run(Packet2d& first, const Packet2d& second)
   {
     if (Offset==1)
     {

Eigen/src/Core/products/CoeffBasedProduct.h

@@ -224,8 +224,8 @@ class CoeffBasedProduct
     { return reinterpret_cast<const LazyCoeffBasedProductType&>(*this).diagonal(index); }
 
   protected:
-    const LhsNested m_lhs;
-    const RhsNested m_rhs;
+    typename internal::add_const_on_value_type<LhsNested>::type m_lhs;
+    typename internal::add_const_on_value_type<RhsNested>::type m_rhs;
 
     mutable PlainObject m_result;
 };
@@ -252,7 +252,7 @@ template<int UnrollingIndex, typename Lhs, typename Rhs, typename RetScalar>
 struct product_coeff_impl<DefaultTraversal, UnrollingIndex, Lhs, Rhs, RetScalar>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar &res)
   {
     product_coeff_impl<DefaultTraversal, UnrollingIndex-1, Lhs, Rhs, RetScalar>::run(row, col, lhs, rhs, res);
     res += lhs.coeff(row, UnrollingIndex) * rhs.coeff(UnrollingIndex, col);
@@ -263,7 +263,7 @@ template<typename Lhs, typename Rhs, typename RetScalar>
 struct product_coeff_impl<DefaultTraversal, 0, Lhs, Rhs, RetScalar>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar &res)
   {
     res = lhs.coeff(row, 0) * rhs.coeff(0, col);
   }
@@ -273,7 +273,7 @@ template<typename Lhs, typename Rhs, typename RetScalar>
 struct product_coeff_impl<DefaultTraversal, Dynamic, Lhs, Rhs, RetScalar>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar& res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar& res)
   {
     eigen_assert(lhs.cols()>0 && "you are using a non initialized matrix");
     res = lhs.coeff(row, 0) * rhs.coeff(0, col);
@@ -291,7 +291,7 @@ struct product_coeff_vectorized_unroller
 {
   typedef typename Lhs::Index Index;
   enum { PacketSize = packet_traits<typename Lhs::Scalar>::size };
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::PacketScalar &pres)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::PacketScalar &pres)
   {
     product_coeff_vectorized_unroller<UnrollingIndex-PacketSize, Lhs, Rhs, Packet>::run(row, col, lhs, rhs, pres);
     pres = padd(pres, pmul( lhs.template packet<Aligned>(row, UnrollingIndex) , rhs.template packet<Aligned>(UnrollingIndex, col) ));
@@ -302,7 +302,7 @@ template<typename Lhs, typename Rhs, typename Packet>
 struct product_coeff_vectorized_unroller<0, Lhs, Rhs, Packet>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::PacketScalar &pres)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::PacketScalar &pres)
   {
     pres = pmul(lhs.template packet<Aligned>(row, 0) , rhs.template packet<Aligned>(0, col));
   }
@@ -314,7 +314,7 @@ struct product_coeff_impl<InnerVectorizedTraversal, UnrollingIndex, Lhs, Rhs, Re
   typedef typename Lhs::PacketScalar Packet;
   typedef typename Lhs::Index Index;
   enum { PacketSize = packet_traits<typename Lhs::Scalar>::size };
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, RetScalar &res)
   {
     Packet pres;
     product_coeff_vectorized_unroller<UnrollingIndex+1-PacketSize, Lhs, Rhs, Packet>::run(row, col, lhs, rhs, pres);
@@ -327,7 +327,7 @@ template<typename Lhs, typename Rhs, int LhsRows = Lhs::RowsAtCompileTime, int R
 struct product_coeff_vectorized_dyn_selector
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
   {
     res = lhs.row(row).transpose().cwiseProduct(rhs.col(col)).sum();
   }
@@ -339,7 +339,7 @@ template<typename Lhs, typename Rhs, int RhsCols>
 struct product_coeff_vectorized_dyn_selector<Lhs,Rhs,1,RhsCols>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index /*row*/, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
+  static EIGEN_STRONG_INLINE void run(Index /*row*/, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
   {
     res = lhs.transpose().cwiseProduct(rhs.col(col)).sum();
   }
@@ -349,7 +349,7 @@ template<typename Lhs, typename Rhs, int LhsRows>
 struct product_coeff_vectorized_dyn_selector<Lhs,Rhs,LhsRows,1>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index /*col*/, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index /*col*/, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
   {
     res = lhs.row(row).transpose().cwiseProduct(rhs).sum();
   }
@@ -359,7 +359,7 @@ template<typename Lhs, typename Rhs>
 struct product_coeff_vectorized_dyn_selector<Lhs,Rhs,1,1>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index /*row*/, Index /*col*/, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
+  static EIGEN_STRONG_INLINE void run(Index /*row*/, Index /*col*/, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
   {
     res = lhs.transpose().cwiseProduct(rhs).sum();
   }
@@ -369,7 +369,7 @@ template<typename Lhs, typename Rhs, typename RetScalar>
 struct product_coeff_impl<InnerVectorizedTraversal, Dynamic, Lhs, Rhs, RetScalar>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, typename Lhs::Scalar &res)
   {
     product_coeff_vectorized_dyn_selector<Lhs,Rhs>::run(row, col, lhs, rhs, res);
   }
@@ -383,7 +383,7 @@ template<int UnrollingIndex, typename Lhs, typename Rhs, typename Packet, int Lo
 struct product_packet_impl<RowMajor, UnrollingIndex, Lhs, Rhs, Packet, LoadMode>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
   {
     product_packet_impl<RowMajor, UnrollingIndex-1, Lhs, Rhs, Packet, LoadMode>::run(row, col, lhs, rhs, res);
     res =  pmadd(pset1<Packet>(lhs.coeff(row, UnrollingIndex)), rhs.template packet<LoadMode>(UnrollingIndex, col), res);
@@ -394,7 +394,7 @@ template<int UnrollingIndex, typename Lhs, typename Rhs, typename Packet, int Lo
 struct product_packet_impl<ColMajor, UnrollingIndex, Lhs, Rhs, Packet, LoadMode>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
   {
     product_packet_impl<ColMajor, UnrollingIndex-1, Lhs, Rhs, Packet, LoadMode>::run(row, col, lhs, rhs, res);
     res =  pmadd(lhs.template packet<LoadMode>(row, UnrollingIndex), pset1<Packet>(rhs.coeff(UnrollingIndex, col)), res);
@@ -405,7 +405,7 @@ template<typename Lhs, typename Rhs, typename Packet, int LoadMode>
 struct product_packet_impl<RowMajor, 0, Lhs, Rhs, Packet, LoadMode>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
   {
     res = pmul(pset1<Packet>(lhs.coeff(row, 0)),rhs.template packet<LoadMode>(0, col));
   }
@@ -415,7 +415,7 @@ template<typename Lhs, typename Rhs, typename Packet, int LoadMode>
 struct product_packet_impl<ColMajor, 0, Lhs, Rhs, Packet, LoadMode>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet &res)
   {
     res = pmul(lhs.template packet<LoadMode>(row, 0), pset1<Packet>(rhs.coeff(0, col)));
   }
@@ -425,7 +425,7 @@ template<typename Lhs, typename Rhs, typename Packet, int LoadMode>
 struct product_packet_impl<RowMajor, Dynamic, Lhs, Rhs, Packet, LoadMode>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet& res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet& res)
   {
     eigen_assert(lhs.cols()>0 && "you are using a non initialized matrix");
     res = pmul(pset1<Packet>(lhs.coeff(row, 0)),rhs.template packet<LoadMode>(0, col));
@@ -438,7 +438,7 @@ template<typename Lhs, typename Rhs, typename Packet, int LoadMode>
 struct product_packet_impl<ColMajor, Dynamic, Lhs, Rhs, Packet, LoadMode>
 {
   typedef typename Lhs::Index Index;
-  EIGEN_STRONG_INLINE static void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet& res)
+  static EIGEN_STRONG_INLINE void run(Index row, Index col, const Lhs& lhs, const Rhs& rhs, Packet& res)
   {
     eigen_assert(lhs.cols()>0 && "you are using a non initialized matrix");
     res = pmul(lhs.template packet<LoadMode>(row, 0), pset1<Packet>(rhs.coeff(0, col)));

Eigen/src/Core/products/GeneralBlockPanelKernel.h

@@ -30,19 +30,18 @@ namespace internal {
 template<typename _LhsScalar, typename _RhsScalar, bool _ConjLhs=false, bool _ConjRhs=false>
 class gebp_traits;
 
+
+/** \internal \returns b if a<=0, and returns a otherwise. */
+inline std::ptrdiff_t manage_caching_sizes_helper(std::ptrdiff_t a, std::ptrdiff_t b)
+{
+  return a<=0 ? b : a;
+}
+
 /** \internal */
 inline void manage_caching_sizes(Action action, std::ptrdiff_t* l1=0, std::ptrdiff_t* l2=0)
 {
-  static std::ptrdiff_t m_l1CacheSize = 0;
-  static std::ptrdiff_t m_l2CacheSize = 0;
-  if(m_l1CacheSize==0)
-  {
-    m_l1CacheSize = queryL1CacheSize();
-    m_l2CacheSize = queryTopLevelCacheSize();
-
-    if(m_l1CacheSize<=0) m_l1CacheSize = 8 * 1024;
-    if(m_l2CacheSize<=0) m_l2CacheSize = 1 * 1024 * 1024;
-  }
+  static std::ptrdiff_t m_l1CacheSize = manage_caching_sizes_helper(queryL1CacheSize(),8 * 1024);
+  static std::ptrdiff_t m_l2CacheSize = manage_caching_sizes_helper(queryTopLevelCacheSize(),1*1024*1024);
 
   if(action==SetAction)
   {
@@ -81,6 +80,7 @@ inline void manage_caching_sizes(Action action, std::ptrdiff_t* l1=0, std::ptrdi
 template<typename LhsScalar, typename RhsScalar, int KcFactor>
 void computeProductBlockingSizes(std::ptrdiff_t& k, std::ptrdiff_t& m, std::ptrdiff_t& n)
 {
+  EIGEN_UNUSED_VARIABLE(n);
   // Explanations:
   // Let's recall the product algorithms form kc x nc horizontal panels B' on the rhs and
   // mc x kc blocks A' on the lhs. A' has to fit into L2 cache. Moreover, B' is processed
@@ -102,7 +102,6 @@ void computeProductBlockingSizes(std::ptrdiff_t& k, std::ptrdiff_t& m, std::ptrd
   k = std::min<std::ptrdiff_t>(k, l1/kdiv);
   std::ptrdiff_t _m = k>0 ? l2/(4 * sizeof(LhsScalar) * k) : 0;
   if(_m<m) m = _m & mr_mask;
-  n = n;
 }
 
 template<typename LhsScalar, typename RhsScalar>
@@ -118,14 +117,14 @@ inline void computeProductBlockingSizes(std::ptrdiff_t& k, std::ptrdiff_t& m, st
   // FIXME (a bit overkill maybe ?)
 
   template<typename CJ, typename A, typename B, typename C, typename T> struct gebp_madd_selector {
-    EIGEN_STRONG_INLINE EIGEN_ALWAYS_INLINE_ATTRIB static void run(const CJ& cj, A& a, B& b, C& c, T& /*t*/)
+    EIGEN_ALWAYS_INLINE static void run(const CJ& cj, A& a, B& b, C& c, T& /*t*/)
     {
       c = cj.pmadd(a,b,c);
     }
   };
 
   template<typename CJ, typename T> struct gebp_madd_selector<CJ,T,T,T,T> {
-    EIGEN_STRONG_INLINE EIGEN_ALWAYS_INLINE_ATTRIB static void run(const CJ& cj, T& a, T& b, T& c, T& t)
+    EIGEN_ALWAYS_INLINE static void run(const CJ& cj, T& a, T& b, T& c, T& t)
     {
       t = b; t = cj.pmul(a,t); c = padd(c,t);
     }
@@ -199,7 +198,7 @@ public:
   EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const RhsScalar* rhs, RhsScalar* b)
   {
     for(DenseIndex k=0; k<n; k++)
-      pstore(&b[k*RhsPacketSize], pset1<RhsPacket>(rhs[k]));
+      pstore1<RhsPacket>(&b[k*RhsPacketSize], rhs[k]);
   }
 
   EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, RhsPacket& dest) const
@@ -270,7 +269,7 @@ public:
   EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const RhsScalar* rhs, RhsScalar* b)
   {
     for(DenseIndex k=0; k<n; k++)
-      pstore(&b[k*RhsPacketSize], pset1<RhsPacket>(rhs[k]));
+      pstore1<RhsPacket>(&b[k*RhsPacketSize], rhs[k]);
   }
 
   EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, RhsPacket& dest) const
@@ -363,8 +362,8 @@ public:
     {
       if(Vectorizable)
       {
-        pstore((RealScalar*)&b[k*ResPacketSize*2+0], pset1<RealPacket>(real(rhs[k])));
-        pstore((RealScalar*)&b[k*ResPacketSize*2+ResPacketSize], pset1<RealPacket>(imag(rhs[k])));
+        pstore1<RealPacket>((RealScalar*)&b[k*ResPacketSize*2+0],             real(rhs[k]));
+        pstore1<RealPacket>((RealScalar*)&b[k*ResPacketSize*2+ResPacketSize], imag(rhs[k]));
       }
       else
         b[k] = rhs[k];
@@ -475,7 +474,7 @@ public:
   EIGEN_STRONG_INLINE void unpackRhs(DenseIndex n, const RhsScalar* rhs, RhsScalar* b)
   {
     for(DenseIndex k=0; k<n; k++)
-      pstore(&b[k*RhsPacketSize], pset1<RhsPacket>(rhs[k]));
+      pstore1<RhsPacket>(&b[k*RhsPacketSize], rhs[k]);
   }
 
   EIGEN_STRONG_INLINE void loadRhs(const RhsScalar* b, RhsPacket& dest) const
@@ -536,7 +535,7 @@ struct gebp_kernel
     ResPacketSize = Traits::ResPacketSize
   };
 
-  EIGEN_FLATTEN_ATTRIB
+  EIGEN_DONT_INLINE EIGEN_FLATTEN_ATTRIB
   void operator()(ResScalar* res, Index resStride, const LhsScalar* blockA, const RhsScalar* blockB, Index rows, Index depth, Index cols, ResScalar alpha,
                   Index strideA=-1, Index strideB=-1, Index offsetA=0, Index offsetB=0, RhsScalar* unpackedB = 0)
   {
@@ -598,64 +597,64 @@ struct gebp_kernel
           if(nr==2)
           {
             LhsPacket A0, A1;
-            RhsPacket B0;
+            RhsPacket B_0;
             RhsPacket T0;
             
 EIGEN_ASM_COMMENT("mybegin2");
             traits.loadLhs(&blA[0*LhsProgress], A0);
             traits.loadLhs(&blA[1*LhsProgress], A1);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[1*RhsProgress], B0);
-            traits.madd(A0,B0,C1,T0);
-            traits.madd(A1,B0,C5,B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[1*RhsProgress], B_0);
+            traits.madd(A0,B_0,C1,T0);
+            traits.madd(A1,B_0,C5,B_0);
 
             traits.loadLhs(&blA[2*LhsProgress], A0);
             traits.loadLhs(&blA[3*LhsProgress], A1);
-            traits.loadRhs(&blB[2*RhsProgress], B0);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[3*RhsProgress], B0);
-            traits.madd(A0,B0,C1,T0);
-            traits.madd(A1,B0,C5,B0);
+            traits.loadRhs(&blB[2*RhsProgress], B_0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[3*RhsProgress], B_0);
+            traits.madd(A0,B_0,C1,T0);
+            traits.madd(A1,B_0,C5,B_0);
 
             traits.loadLhs(&blA[4*LhsProgress], A0);
             traits.loadLhs(&blA[5*LhsProgress], A1);
-            traits.loadRhs(&blB[4*RhsProgress], B0);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[5*RhsProgress], B0);
-            traits.madd(A0,B0,C1,T0);
-            traits.madd(A1,B0,C5,B0);
+            traits.loadRhs(&blB[4*RhsProgress], B_0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[5*RhsProgress], B_0);
+            traits.madd(A0,B_0,C1,T0);
+            traits.madd(A1,B_0,C5,B_0);
 
             traits.loadLhs(&blA[6*LhsProgress], A0);
             traits.loadLhs(&blA[7*LhsProgress], A1);
-            traits.loadRhs(&blB[6*RhsProgress], B0);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[7*RhsProgress], B0);
-            traits.madd(A0,B0,C1,T0);
-            traits.madd(A1,B0,C5,B0);
+            traits.loadRhs(&blB[6*RhsProgress], B_0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[7*RhsProgress], B_0);
+            traits.madd(A0,B_0,C1,T0);
+            traits.madd(A1,B_0,C5,B_0);
 EIGEN_ASM_COMMENT("myend");
           }
           else
           {
 EIGEN_ASM_COMMENT("mybegin4");
             LhsPacket A0, A1;
-            RhsPacket B0, B1, B2, B3;
+            RhsPacket B_0, B1, B2, B3;
             RhsPacket T0;
             
             traits.loadLhs(&blA[0*LhsProgress], A0);
             traits.loadLhs(&blA[1*LhsProgress], A1);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
             traits.loadRhs(&blB[1*RhsProgress], B1);
 
-            traits.madd(A0,B0,C0,T0);
+            traits.madd(A0,B_0,C0,T0);
             traits.loadRhs(&blB[2*RhsProgress], B2);
-            traits.madd(A1,B0,C4,B0);
+            traits.madd(A1,B_0,C4,B_0);
             traits.loadRhs(&blB[3*RhsProgress], B3);
-            traits.loadRhs(&blB[4*RhsProgress], B0);
+            traits.loadRhs(&blB[4*RhsProgress], B_0);
             traits.madd(A0,B1,C1,T0);
             traits.madd(A1,B1,C5,B1);
             traits.loadRhs(&blB[5*RhsProgress], B1);
@@ -667,9 +666,9 @@ EIGEN_ASM_COMMENT("mybegin4");
             traits.madd(A1,B3,C7,B3);
             traits.loadLhs(&blA[3*LhsProgress], A1);
             traits.loadRhs(&blB[7*RhsProgress], B3);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[8*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[8*RhsProgress], B_0);
             traits.madd(A0,B1,C1,T0);
             traits.madd(A1,B1,C5,B1);
             traits.loadRhs(&blB[9*RhsProgress], B1);
@@ -682,9 +681,9 @@ EIGEN_ASM_COMMENT("mybegin4");
             traits.loadLhs(&blA[5*LhsProgress], A1);
             traits.loadRhs(&blB[11*RhsProgress], B3);
 
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[12*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[12*RhsProgress], B_0);
             traits.madd(A0,B1,C1,T0);
             traits.madd(A1,B1,C5,B1);
             traits.loadRhs(&blB[13*RhsProgress], B1);
@@ -696,8 +695,8 @@ EIGEN_ASM_COMMENT("mybegin4");
             traits.madd(A1,B3,C7,B3);
             traits.loadLhs(&blA[7*LhsProgress], A1);
             traits.loadRhs(&blB[15*RhsProgress], B3);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
             traits.madd(A0,B1,C1,T0);
             traits.madd(A1,B1,C5,B1);
             traits.madd(A0,B2,C2,T0);
@@ -715,32 +714,32 @@ EIGEN_ASM_COMMENT("mybegin4");
           if(nr==2)
           {
             LhsPacket A0, A1;
-            RhsPacket B0;
+            RhsPacket B_0;
             RhsPacket T0;
 
             traits.loadLhs(&blA[0*LhsProgress], A0);
             traits.loadLhs(&blA[1*LhsProgress], A1);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
-            traits.madd(A0,B0,C0,T0);
-            traits.madd(A1,B0,C4,B0);
-            traits.loadRhs(&blB[1*RhsProgress], B0);
-            traits.madd(A0,B0,C1,T0);
-            traits.madd(A1,B0,C5,B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
+            traits.madd(A0,B_0,C0,T0);
+            traits.madd(A1,B_0,C4,B_0);
+            traits.loadRhs(&blB[1*RhsProgress], B_0);
+            traits.madd(A0,B_0,C1,T0);
+            traits.madd(A1,B_0,C5,B_0);
           }
           else
           {
             LhsPacket A0, A1;
-            RhsPacket B0, B1, B2, B3;
+            RhsPacket B_0, B1, B2, B3;
             RhsPacket T0;
 
             traits.loadLhs(&blA[0*LhsProgress], A0);
             traits.loadLhs(&blA[1*LhsProgress], A1);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
             traits.loadRhs(&blB[1*RhsProgress], B1);
 
-            traits.madd(A0,B0,C0,T0);
+            traits.madd(A0,B_0,C0,T0);
             traits.loadRhs(&blB[2*RhsProgress], B2);
-            traits.madd(A1,B0,C4,B0);
+            traits.madd(A1,B_0,C4,B_0);
             traits.loadRhs(&blB[3*RhsProgress], B3);
             traits.madd(A0,B1,C1,T0);
             traits.madd(A1,B1,C5,B1);
@@ -827,42 +826,42 @@ EIGEN_ASM_COMMENT("mybegin4");
           if(nr==2)
           {
             LhsPacket A0;
-            RhsPacket B0, B1;
+            RhsPacket B_0, B1;
 
             traits.loadLhs(&blA[0*LhsProgress], A0);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
             traits.loadRhs(&blB[1*RhsProgress], B1);
-            traits.madd(A0,B0,C0,B0);
-            traits.loadRhs(&blB[2*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,B_0);
+            traits.loadRhs(&blB[2*RhsProgress], B_0);
             traits.madd(A0,B1,C1,B1);
             traits.loadLhs(&blA[1*LhsProgress], A0);
             traits.loadRhs(&blB[3*RhsProgress], B1);
-            traits.madd(A0,B0,C0,B0);
-            traits.loadRhs(&blB[4*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,B_0);
+            traits.loadRhs(&blB[4*RhsProgress], B_0);
             traits.madd(A0,B1,C1,B1);
             traits.loadLhs(&blA[2*LhsProgress], A0);
             traits.loadRhs(&blB[5*RhsProgress], B1);
-            traits.madd(A0,B0,C0,B0);
-            traits.loadRhs(&blB[6*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,B_0);
+            traits.loadRhs(&blB[6*RhsProgress], B_0);
             traits.madd(A0,B1,C1,B1);
             traits.loadLhs(&blA[3*LhsProgress], A0);
             traits.loadRhs(&blB[7*RhsProgress], B1);
-            traits.madd(A0,B0,C0,B0);
+            traits.madd(A0,B_0,C0,B_0);
             traits.madd(A0,B1,C1,B1);
           }
           else
           {
             LhsPacket A0;
-            RhsPacket B0, B1, B2, B3;
+            RhsPacket B_0, B1, B2, B3;
 
             traits.loadLhs(&blA[0*LhsProgress], A0);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
             traits.loadRhs(&blB[1*RhsProgress], B1);
 
-            traits.madd(A0,B0,C0,B0);
+            traits.madd(A0,B_0,C0,B_0);
             traits.loadRhs(&blB[2*RhsProgress], B2);
             traits.loadRhs(&blB[3*RhsProgress], B3);
-            traits.loadRhs(&blB[4*RhsProgress], B0);
+            traits.loadRhs(&blB[4*RhsProgress], B_0);
             traits.madd(A0,B1,C1,B1);
             traits.loadRhs(&blB[5*RhsProgress], B1);
             traits.madd(A0,B2,C2,B2);
@@ -870,8 +869,8 @@ EIGEN_ASM_COMMENT("mybegin4");
             traits.madd(A0,B3,C3,B3);
             traits.loadLhs(&blA[1*LhsProgress], A0);
             traits.loadRhs(&blB[7*RhsProgress], B3);
-            traits.madd(A0,B0,C0,B0);
-            traits.loadRhs(&blB[8*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,B_0);
+            traits.loadRhs(&blB[8*RhsProgress], B_0);
             traits.madd(A0,B1,C1,B1);
             traits.loadRhs(&blB[9*RhsProgress], B1);
             traits.madd(A0,B2,C2,B2);
@@ -880,8 +879,8 @@ EIGEN_ASM_COMMENT("mybegin4");
             traits.loadLhs(&blA[2*LhsProgress], A0);
             traits.loadRhs(&blB[11*RhsProgress], B3);
 
-            traits.madd(A0,B0,C0,B0);
-            traits.loadRhs(&blB[12*RhsProgress], B0);
+            traits.madd(A0,B_0,C0,B_0);
+            traits.loadRhs(&blB[12*RhsProgress], B_0);
             traits.madd(A0,B1,C1,B1);
             traits.loadRhs(&blB[13*RhsProgress], B1);
             traits.madd(A0,B2,C2,B2);
@@ -890,7 +889,7 @@ EIGEN_ASM_COMMENT("mybegin4");
 
             traits.loadLhs(&blA[3*LhsProgress], A0);
             traits.loadRhs(&blB[15*RhsProgress], B3);
-            traits.madd(A0,B0,C0,B0);
+            traits.madd(A0,B_0,C0,B_0);
             traits.madd(A0,B1,C1,B1);
             traits.madd(A0,B2,C2,B2);
             traits.madd(A0,B3,C3,B3);
@@ -905,26 +904,26 @@ EIGEN_ASM_COMMENT("mybegin4");
           if(nr==2)
           {
             LhsPacket A0;
-            RhsPacket B0, B1;
+            RhsPacket B_0, B1;
 
             traits.loadLhs(&blA[0*LhsProgress], A0);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
             traits.loadRhs(&blB[1*RhsProgress], B1);
-            traits.madd(A0,B0,C0,B0);
+            traits.madd(A0,B_0,C0,B_0);
             traits.madd(A0,B1,C1,B1);
           }
           else
           {
             LhsPacket A0;
-            RhsPacket B0, B1, B2, B3;
+            RhsPacket B_0, B1, B2, B3;
 
             traits.loadLhs(&blA[0*LhsProgress], A0);
-            traits.loadRhs(&blB[0*RhsProgress], B0);
+            traits.loadRhs(&blB[0*RhsProgress], B_0);
             traits.loadRhs(&blB[1*RhsProgress], B1);
             traits.loadRhs(&blB[2*RhsProgress], B2);
             traits.loadRhs(&blB[3*RhsProgress], B3);
 
-            traits.madd(A0,B0,C0,B0);
+            traits.madd(A0,B_0,C0,B_0);
             traits.madd(A0,B1,C1,B1);
             traits.madd(A0,B2,C2,B2);
             traits.madd(A0,B3,C3,B3);
@@ -971,26 +970,26 @@ EIGEN_ASM_COMMENT("mybegin4");
           if(nr==2)
           {
             LhsScalar A0;
-            RhsScalar B0, B1;
+            RhsScalar B_0, B1;
 
             A0 = blA[k];
-            B0 = blB[0];
+            B_0 = blB[0];
             B1 = blB[1];
-            MADD(cj,A0,B0,C0,B0);
+            MADD(cj,A0,B_0,C0,B_0);
             MADD(cj,A0,B1,C1,B1);
           }
           else
           {
             LhsScalar A0;
-            RhsScalar B0, B1, B2, B3;
+            RhsScalar B_0, B1, B2, B3;
 
             A0 = blA[k];
-            B0 = blB[0];
+            B_0 = blB[0];
             B1 = blB[1];
             B2 = blB[2];
             B3 = blB[3];
 
-            MADD(cj,A0,B0,C0,B0);
+            MADD(cj,A0,B_0,C0,B_0);
             MADD(cj,A0,B1,C1,B1);
             MADD(cj,A0,B2,C2,B2);
             MADD(cj,A0,B3,C3,B3);
@@ -1009,12 +1008,7 @@ EIGEN_ASM_COMMENT("mybegin4");
     for(Index j2=packet_cols; j2<cols; j2++)
     {
       // unpack B
-      {
-        traits.unpackRhs(depth, &blockB[j2*strideB+offsetB], unpackedB);
-//         const RhsScalar* blB = &blockB[j2*strideB+offsetB];
-//         for(Index k=0; k<depth; k++)
-//           pstore(&unpackedB[k*RhsPacketSize], pset1<RhsPacket>(blB[k]));
-      }
+      traits.unpackRhs(depth, &blockB[j2*strideB+offsetB], unpackedB);
 
       for(Index i=0; i<peeled_mc; i+=mr)
       {
@@ -1032,14 +1026,14 @@ EIGEN_ASM_COMMENT("mybegin4");
         for(Index k=0; k<depth; k++)
         {
           LhsPacket A0, A1;
-          RhsPacket B0;
+          RhsPacket B_0;
           RhsPacket T0;
 
           traits.loadLhs(&blA[0*LhsProgress], A0);
           traits.loadLhs(&blA[1*LhsProgress], A1);
-          traits.loadRhs(&blB[0*RhsProgress], B0);
-          traits.madd(A0,B0,C0,T0);
-          traits.madd(A1,B0,C4,B0);
+          traits.loadRhs(&blB[0*RhsProgress], B_0);
+          traits.madd(A0,B_0,C0,T0);
+          traits.madd(A1,B_0,C4,B_0);
 
           blB += RhsProgress;
           blA += 2*LhsProgress;
@@ -1071,10 +1065,10 @@ EIGEN_ASM_COMMENT("mybegin4");
         for(Index k=0; k<depth; k++)
         {
           LhsPacket A0;
-          RhsPacket B0;
+          RhsPacket B_0;
           traits.loadLhs(blA, A0);
-          traits.loadRhs(blB, B0);
-          traits.madd(A0, B0, C0, B0);
+          traits.loadRhs(blB, B_0);
+          traits.madd(A0, B_0, C0, B_0);
           blB += RhsProgress;
           blA += LhsProgress;
         }
@@ -1096,8 +1090,8 @@ EIGEN_ASM_COMMENT("mybegin4");
         for(Index k=0; k<depth; k++)
         {
           LhsScalar A0 = blA[k];
-          RhsScalar B0 = blB[k];
-          MADD(cj, A0, B0, C0, B0);
+          RhsScalar B_0 = blB[k];
+          MADD(cj, A0, B_0, C0, B_0);
         }
         res[(j2+0)*resStride + i] += alpha*C0;
       }
@@ -1108,7 +1102,7 @@ EIGEN_ASM_COMMENT("mybegin4");
 #undef CJMADD
 
 // pack a block of the lhs
-// The travesal is as follow (mr==4):
+// The traversal is as follow (mr==4):
 //   0  4  8 12 ...
 //   1  5  9 13 ...
 //   2  6 10 14 ...
@@ -1124,11 +1118,15 @@ EIGEN_ASM_COMMENT("mybegin4");
 template<typename Scalar, typename Index, int Pack1, int Pack2, int StorageOrder, bool Conjugate, bool PanelMode>
 struct gemm_pack_lhs
 {
-  void operator()(Scalar* blockA, const Scalar* EIGEN_RESTRICT _lhs, Index lhsStride, Index depth, Index rows,
+  EIGEN_DONT_INLINE void operator()(Scalar* blockA, const Scalar* EIGEN_RESTRICT _lhs, Index lhsStride, Index depth, Index rows,
                   Index stride=0, Index offset=0)
   {
-//     enum { PacketSize = packet_traits<Scalar>::size };
+    typedef typename packet_traits<Scalar>::type Packet;
+    enum { PacketSize = packet_traits<Scalar>::size };
+
+    EIGEN_ASM_COMMENT("EIGEN PRODUCT PACK LHS");
     eigen_assert(((!PanelMode) && stride==0 && offset==0) || (PanelMode && stride>=depth && offset<=stride));
+    eigen_assert( (StorageOrder==RowMajor) || ((Pack1%PacketSize)==0 && Pack1<=4*PacketSize) );
     conj_if<NumTraits<Scalar>::IsComplex && Conjugate> cj;
     const_blas_data_mapper<Scalar, Index, StorageOrder> lhs(_lhs,lhsStride);
     Index count = 0;
@@ -1136,9 +1134,44 @@ struct gemm_pack_lhs
     for(Index i=0; i<peeled_mc; i+=Pack1)
     {
       if(PanelMode) count += Pack1 * offset;
-      for(Index k=0; k<depth; k++)
-        for(Index w=0; w<Pack1; w++)
-          blockA[count++] = cj(lhs(i+w, k));
+
+      if(StorageOrder==ColMajor)
+      {
+        for(Index k=0; k<depth; k++)
+        {
+          Packet A, B, C, D;
+          if(Pack1>=1*PacketSize) A = ploadu<Packet>(&lhs(i+0*PacketSize, k));
+          if(Pack1>=2*PacketSize) B = ploadu<Packet>(&lhs(i+1*PacketSize, k));
+          if(Pack1>=3*PacketSize) C = ploadu<Packet>(&lhs(i+2*PacketSize, k));
+          if(Pack1>=4*PacketSize) D = ploadu<Packet>(&lhs(i+3*PacketSize, k));
+          if(Pack1>=1*PacketSize) { pstore(blockA+count, cj.pconj(A)); count+=PacketSize; }
+          if(Pack1>=2*PacketSize) { pstore(blockA+count, cj.pconj(B)); count+=PacketSize; }
+          if(Pack1>=3*PacketSize) { pstore(blockA+count, cj.pconj(C)); count+=PacketSize; }
+          if(Pack1>=4*PacketSize) { pstore(blockA+count, cj.pconj(D)); count+=PacketSize; }
+        }
+      }
+      else
+      {
+        for(Index k=0; k<depth; k++)
+        {
+          // TODO add a vectorized transpose here
+          Index w=0;
+          for(; w<Pack1-3; w+=4)
+          {
+            Scalar a(cj(lhs(i+w+0, k))),
+                   b(cj(lhs(i+w+1, k))),
+                   c(cj(lhs(i+w+2, k))),
+                   d(cj(lhs(i+w+3, k)));
+            blockA[count++] = a;
+            blockA[count++] = b;
+            blockA[count++] = c;
+            blockA[count++] = d;
+          }
+          if(Pack1%4)
+            for(;w<Pack1;++w)
+              blockA[count++] = cj(lhs(i+w, k));
+        }
+      }
       if(PanelMode) count += Pack1 * (stride-offset-depth);
     }
     if(rows-peeled_mc>=Pack2)
@@ -1172,9 +1205,10 @@ struct gemm_pack_rhs<Scalar, Index, nr, ColMajor, Conjugate, PanelMode>
 {
   typedef typename packet_traits<Scalar>::type Packet;
   enum { PacketSize = packet_traits<Scalar>::size };
-  void operator()(Scalar* blockB, const Scalar* rhs, Index rhsStride, Index depth, Index cols,
+  EIGEN_DONT_INLINE void operator()(Scalar* blockB, const Scalar* rhs, Index rhsStride, Index depth, Index cols,
                   Index stride=0, Index offset=0)
   {
+    EIGEN_ASM_COMMENT("EIGEN PRODUCT PACK RHS COLMAJOR");
     eigen_assert(((!PanelMode) && stride==0 && offset==0) || (PanelMode && stride>=depth && offset<=stride));
     conj_if<NumTraits<Scalar>::IsComplex && Conjugate> cj;
     Index packet_cols = (cols/nr) * nr;
@@ -1219,9 +1253,10 @@ template<typename Scalar, typename Index, int nr, bool Conjugate, bool PanelMode
 struct gemm_pack_rhs<Scalar, Index, nr, RowMajor, Conjugate, PanelMode>
 {
   enum { PacketSize = packet_traits<Scalar>::size };
-  void operator()(Scalar* blockB, const Scalar* rhs, Index rhsStride, Index depth, Index cols,
+  EIGEN_DONT_INLINE void operator()(Scalar* blockB, const Scalar* rhs, Index rhsStride, Index depth, Index cols,
                   Index stride=0, Index offset=0)
   {
+    EIGEN_ASM_COMMENT("EIGEN PRODUCT PACK RHS ROWMAJOR");
     eigen_assert(((!PanelMode) && stride==0 && offset==0) || (PanelMode && stride>=depth && offset<=stride));
     conj_if<NumTraits<Scalar>::IsComplex && Conjugate> cj;
     Index packet_cols = (cols/nr) * nr;

Eigen/src/Core/products/GeneralMatrixMatrix.h

@@ -78,7 +78,7 @@ static void run(Index rows, Index cols, Index depth,
   typedef gebp_traits<LhsScalar,RhsScalar> Traits;
 
   Index kc = blocking.kc();                 // cache block size along the K direction
-  Index mc = std::min(rows,blocking.mc());  // cache block size along the M direction
+  Index mc = (std::min)(rows,blocking.mc());  // cache block size along the M direction
   //Index nc = blocking.nc(); // cache block size along the N direction
 
   gemm_pack_lhs<LhsScalar, Index, Traits::mr, Traits::LhsProgress, LhsStorageOrder> pack_lhs;
@@ -94,15 +94,16 @@ static void run(Index rows, Index cols, Index depth,
     
     std::size_t sizeA = kc*mc;
     std::size_t sizeW = kc*Traits::WorkSpaceFactor;
-    LhsScalar* blockA = ei_aligned_stack_new(LhsScalar, sizeA);
-    RhsScalar* w = ei_aligned_stack_new(RhsScalar, sizeW);
+    ei_declare_aligned_stack_constructed_variable(LhsScalar, blockA, sizeA, 0);
+    ei_declare_aligned_stack_constructed_variable(RhsScalar, w, sizeW, 0);
+    
     RhsScalar* blockB = blocking.blockB();
     eigen_internal_assert(blockB!=0);
 
     // For each horizontal panel of the rhs, and corresponding vertical panel of the lhs...
     for(Index k=0; k<depth; k+=kc)
     {
-      const Index actual_kc = std::min(k+kc,depth)-k; // => rows of B', and cols of the A'
+      const Index actual_kc = (std::min)(k+kc,depth)-k; // => rows of B', and cols of the A'
 
       // In order to reduce the chance that a thread has to wait for the other,
       // let's start by packing A'.
@@ -139,7 +140,7 @@ static void run(Index rows, Index cols, Index depth,
       // Then keep going as usual with the remaining A'
       for(Index i=mc; i<rows; i+=mc)
       {
-        const Index actual_mc = std::min(i+mc,rows)-i;
+        const Index actual_mc = (std::min)(i+mc,rows)-i;
 
         // pack A_i,k to A'
         pack_lhs(blockA, &lhs(i,k), lhsStride, actual_kc, actual_mc);
@@ -154,9 +155,6 @@ static void run(Index rows, Index cols, Index depth,
         #pragma omp atomic
         --(info[j].users);
     }
-
-    ei_aligned_stack_delete(LhsScalar, blockA, kc*mc);
-    ei_aligned_stack_delete(RhsScalar, w, sizeW);
   }
   else
 #endif // EIGEN_HAS_OPENMP
@@ -167,15 +165,16 @@ static void run(Index rows, Index cols, Index depth,
     std::size_t sizeA = kc*mc;
     std::size_t sizeB = kc*cols;
     std::size_t sizeW = kc*Traits::WorkSpaceFactor;
-    LhsScalar *blockA = blocking.blockA()==0 ? ei_aligned_stack_new(LhsScalar, sizeA) : blocking.blockA();
-    RhsScalar *blockB = blocking.blockB()==0 ? ei_aligned_stack_new(RhsScalar, sizeB) : blocking.blockB();
-    RhsScalar *blockW = blocking.blockW()==0 ? ei_aligned_stack_new(RhsScalar, sizeW) : blocking.blockW();
+
+    ei_declare_aligned_stack_constructed_variable(LhsScalar, blockA, sizeA, blocking.blockA());
+    ei_declare_aligned_stack_constructed_variable(RhsScalar, blockB, sizeB, blocking.blockB());
+    ei_declare_aligned_stack_constructed_variable(RhsScalar, blockW, sizeW, blocking.blockW());
 
     // For each horizontal panel of the rhs, and corresponding panel of the lhs...
     // (==GEMM_VAR1)
     for(Index k2=0; k2<depth; k2+=kc)
     {
-      const Index actual_kc = std::min(k2+kc,depth)-k2;
+      const Index actual_kc = (std::min)(k2+kc,depth)-k2;
 
       // OK, here we have selected one horizontal panel of rhs and one vertical panel of lhs.
       // => Pack rhs's panel into a sequential chunk of memory (L2 caching)
@@ -188,7 +187,7 @@ static void run(Index rows, Index cols, Index depth,
       // (==GEPP_VAR1)
       for(Index i2=0; i2<rows; i2+=mc)
       {
-        const Index actual_mc = std::min(i2+mc,rows)-i2;
+        const Index actual_mc = (std::min)(i2+mc,rows)-i2;
 
         // We pack the lhs's block into a sequential chunk of memory (L1 caching)
         // Note that this block will be read a very high number of times, which is equal to the number of
@@ -200,10 +199,6 @@ static void run(Index rows, Index cols, Index depth,
 
       }
     }
-
-    if(blocking.blockA()==0) ei_aligned_stack_delete(LhsScalar, blockA, sizeA);
-    if(blocking.blockB()==0) ei_aligned_stack_delete(RhsScalar, blockB, sizeB);
-    if(blocking.blockW()==0) ei_aligned_stack_delete(RhsScalar, blockW, sizeW);
   }
 }
 
@@ -417,8 +412,8 @@ class GeneralProduct<Lhs, Rhs, GemmProduct>
     {
       eigen_assert(dst.rows()==m_lhs.rows() && dst.cols()==m_rhs.cols());
 
-      const ActualLhsType lhs = LhsBlasTraits::extract(m_lhs);
-      const ActualRhsType rhs = RhsBlasTraits::extract(m_rhs);
+      typename internal::add_const_on_value_type<ActualLhsType>::type lhs = LhsBlasTraits::extract(m_lhs);
+      typename internal::add_const_on_value_type<ActualRhsType>::type rhs = RhsBlasTraits::extract(m_rhs);
 
       Scalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(m_lhs)
                                  * RhsBlasTraits::extractScalarFactor(m_rhs);

Eigen/src/Core/products/GeneralMatrixMatrixTriangular.h

@@ -42,14 +42,14 @@ struct tribb_kernel;
 template <typename Index,
           typename LhsScalar, int LhsStorageOrder, bool ConjugateLhs,
           typename RhsScalar, int RhsStorageOrder, bool ConjugateRhs,
-                              int ResStorageOrder, int  UpLo>
+                              int ResStorageOrder, int  UpLo, int Version = Specialized>
 struct general_matrix_matrix_triangular_product;
 
 // as usual if the result is row major => we transpose the product
 template <typename Index, typename LhsScalar, int LhsStorageOrder, bool ConjugateLhs,
-                          typename RhsScalar, int RhsStorageOrder, bool ConjugateRhs, int  UpLo>
-struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,ConjugateLhs,RhsScalar,RhsStorageOrder,ConjugateRhs,RowMajor,UpLo>
-{  
+                          typename RhsScalar, int RhsStorageOrder, bool ConjugateRhs, int  UpLo, int Version>
+struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,ConjugateLhs,RhsScalar,RhsStorageOrder,ConjugateRhs,RowMajor,UpLo,Version>
+{
   typedef typename scalar_product_traits<LhsScalar, RhsScalar>::ReturnType ResScalar;
   static EIGEN_STRONG_INLINE void run(Index size, Index depth,const LhsScalar* lhs, Index lhsStride,
                                       const RhsScalar* rhs, Index rhsStride, ResScalar* res, Index resStride, ResScalar alpha)
@@ -63,8 +63,8 @@ struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,
 };
 
 template <typename Index, typename LhsScalar, int LhsStorageOrder, bool ConjugateLhs,
-                          typename RhsScalar, int RhsStorageOrder, bool ConjugateRhs, int  UpLo>
-struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,ConjugateLhs,RhsScalar,RhsStorageOrder,ConjugateRhs,ColMajor,UpLo>
+                          typename RhsScalar, int RhsStorageOrder, bool ConjugateRhs, int  UpLo, int Version>
+struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,ConjugateLhs,RhsScalar,RhsStorageOrder,ConjugateRhs,ColMajor,UpLo,Version>
 {
   typedef typename scalar_product_traits<LhsScalar, RhsScalar>::ReturnType ResScalar;
   static EIGEN_STRONG_INLINE void run(Index size, Index depth,const LhsScalar* _lhs, Index lhsStride,
@@ -83,10 +83,10 @@ struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,
     if(mc > Traits::nr)
       mc = (mc/Traits::nr)*Traits::nr;
 
-    LhsScalar* blockA = ei_aligned_stack_new(LhsScalar, kc*mc);
     std::size_t sizeW = kc*Traits::WorkSpaceFactor;
     std::size_t sizeB = sizeW + kc*size;
-    RhsScalar* allocatedBlockB = ei_aligned_stack_new(RhsScalar, sizeB);
+    ei_declare_aligned_stack_constructed_variable(LhsScalar, blockA, kc*mc, 0);
+    ei_declare_aligned_stack_constructed_variable(RhsScalar, allocatedBlockB, sizeB, 0);
     RhsScalar* blockB = allocatedBlockB + sizeW;
     
     gemm_pack_lhs<LhsScalar, Index, Traits::mr, Traits::LhsProgress, LhsStorageOrder> pack_lhs;
@@ -96,14 +96,14 @@ struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,
 
     for(Index k2=0; k2<depth; k2+=kc)
     {
-      const Index actual_kc = std::min(k2+kc,depth)-k2;
+      const Index actual_kc = (std::min)(k2+kc,depth)-k2;
 
       // note that the actual rhs is the transpose/adjoint of mat
       pack_rhs(blockB, &rhs(k2,0), rhsStride, actual_kc, size);
 
       for(Index i2=0; i2<size; i2+=mc)
       {
-        const Index actual_mc = std::min(i2+mc,size)-i2;
+        const Index actual_mc = (std::min)(i2+mc,size)-i2;
 
         pack_lhs(blockA, &lhs(i2, k2), lhsStride, actual_kc, actual_mc);
 
@@ -112,7 +112,7 @@ struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,
         //  2 - the actual_mc x actual_mc symmetric block => processed with a special kernel
         //  3 - after the diagonal => processed with gebp or skipped
         if (UpLo==Lower)
-          gebp(res+i2, resStride, blockA, blockB, actual_mc, actual_kc, std::min(size,i2), alpha,
+          gebp(res+i2, resStride, blockA, blockB, actual_mc, actual_kc, (std::min)(size,i2), alpha,
                -1, -1, 0, 0, allocatedBlockB);
 
         sybb(res+resStride*i2 + i2, resStride, blockA, blockB + actual_kc*i2, actual_mc, actual_kc, alpha, allocatedBlockB);
@@ -120,13 +120,11 @@ struct general_matrix_matrix_triangular_product<Index,LhsScalar,LhsStorageOrder,
         if (UpLo==Upper)
         {
           Index j2 = i2+actual_mc;
-          gebp(res+resStride*j2+i2, resStride, blockA, blockB+actual_kc*j2, actual_mc, actual_kc, std::max(Index(0), size-j2), alpha,
+          gebp(res+resStride*j2+i2, resStride, blockA, blockB+actual_kc*j2, actual_mc, actual_kc, (std::max)(Index(0), size-j2), alpha,
                -1, -1, 0, 0, allocatedBlockB);
         }
       }
     }
-    ei_aligned_stack_delete(LhsScalar, blockA, kc*mc);
-    ei_aligned_stack_delete(RhsScalar, allocatedBlockB, sizeB);
   }
 };
 
@@ -203,13 +201,13 @@ TriangularView<MatrixType,UpLo>& TriangularView<MatrixType,UpLo>::assignProduct(
   typedef internal::blas_traits<Lhs> LhsBlasTraits;
   typedef typename LhsBlasTraits::DirectLinearAccessType ActualLhs;
   typedef typename internal::remove_all<ActualLhs>::type _ActualLhs;
-  const ActualLhs actualLhs = LhsBlasTraits::extract(prod.lhs());
+  typename internal::add_const_on_value_type<ActualLhs>::type actualLhs = LhsBlasTraits::extract(prod.lhs());
   
   typedef typename internal::remove_all<typename ProductDerived::RhsNested>::type Rhs;
   typedef internal::blas_traits<Rhs> RhsBlasTraits;
   typedef typename RhsBlasTraits::DirectLinearAccessType ActualRhs;
   typedef typename internal::remove_all<ActualRhs>::type _ActualRhs;
-  const ActualRhs actualRhs = RhsBlasTraits::extract(prod.rhs());
+  typename internal::add_const_on_value_type<ActualRhs>::type actualRhs = RhsBlasTraits::extract(prod.rhs());
 
   typename ProductDerived::Scalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(prod.lhs().derived()) * RhsBlasTraits::extractScalarFactor(prod.rhs().derived());
 

Eigen/src/Core/products/GeneralMatrixMatrixTriangular_MKL.h

@@ -0,0 +1,142 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *   Level 3 BLAS SYRK/HERK implementation.
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_GENERAL_MATRIX_MATRIX_TRIANGULAR_MKL_H
+#define EIGEN_GENERAL_MATRIX_MATRIX_TRIANGULAR_MKL_H
+
+namespace internal {
+
+template <typename Index, typename Scalar, int AStorageOrder, bool ConjugateA, int ResStorageOrder, int  UpLo>
+struct general_matrix_matrix_rankupdate :
+       general_matrix_matrix_triangular_product<
+         Index,Scalar,AStorageOrder,ConjugateA,Scalar,AStorageOrder,ConjugateA,ResStorageOrder,UpLo,BuiltIn> {};
+
+
+// try to go to BLAS specialization
+#define EIGEN_MKL_RANKUPDATE_SPECIALIZE(Scalar) \
+template <typename Index, int LhsStorageOrder, bool ConjugateLhs, \
+                          int RhsStorageOrder, bool ConjugateRhs, int  UpLo> \
+struct general_matrix_matrix_triangular_product<Index,Scalar,LhsStorageOrder,ConjugateLhs, \
+               Scalar,RhsStorageOrder,ConjugateRhs,ColMajor,UpLo,Specialized> { \
+  static EIGEN_STRONG_INLINE void run(Index size, Index depth,const Scalar* lhs, Index lhsStride, \
+                          const Scalar* rhs, Index rhsStride, Scalar* res, Index resStride, Scalar alpha) \
+  { \
+    if (lhs==rhs) { \
+      general_matrix_matrix_rankupdate<Index,Scalar,LhsStorageOrder,ConjugateLhs,ColMajor,UpLo> \
+      ::run(size,depth,lhs,lhsStride,rhs,rhsStride,res,resStride,alpha); \
+    } else { \
+      general_matrix_matrix_triangular_product<Index, \
+        Scalar, LhsStorageOrder, ConjugateLhs, \
+        Scalar, RhsStorageOrder, ConjugateRhs, \
+        ColMajor, UpLo, BuiltIn> \
+      ::run(size,depth,lhs,lhsStride,rhs,rhsStride,res,resStride,alpha); \
+    } \
+  } \
+};
+
+EIGEN_MKL_RANKUPDATE_SPECIALIZE(double)
+//EIGEN_MKL_RANKUPDATE_SPECIALIZE(dcomplex)
+EIGEN_MKL_RANKUPDATE_SPECIALIZE(float)
+//EIGEN_MKL_RANKUPDATE_SPECIALIZE(scomplex)
+
+// SYRK for float/double
+#define EIGEN_MKL_RANKUPDATE_R(EIGTYPE, MKLTYPE, MKLFUNC) \
+template <typename Index, int AStorageOrder, bool ConjugateA, int  UpLo> \
+struct general_matrix_matrix_rankupdate<Index,EIGTYPE,AStorageOrder,ConjugateA,ColMajor,UpLo> { \
+  enum { \
+    IsLower = (UpLo&Lower) == Lower, \
+    LowUp = IsLower ? Lower : Upper, \
+    conjA = ((AStorageOrder==ColMajor) && ConjugateA) ? 1 : 0 \
+  }; \
+  static EIGEN_STRONG_INLINE void run(Index size, Index depth,const EIGTYPE* lhs, Index lhsStride, \
+                          const EIGTYPE* rhs, Index rhsStride, EIGTYPE* res, Index resStride, EIGTYPE alpha) \
+  { \
+  /* typedef Matrix<EIGTYPE, Dynamic, Dynamic, RhsStorageOrder> MatrixRhs;*/ \
+\
+   MKL_INT lda=lhsStride, ldc=resStride, n=size, k=depth; \
+   char uplo=(IsLower) ? 'L' : 'U', trans=(AStorageOrder==RowMajor) ? 'T':'N'; \
+   MKLTYPE alpha_, beta_; \
+\
+/* Set alpha_ & beta_ */ \
+   assign_scalar_eig2mkl<MKLTYPE, EIGTYPE>(alpha_, alpha); \
+   assign_scalar_eig2mkl<MKLTYPE, EIGTYPE>(beta_, EIGTYPE(1)); \
+   MKLFUNC(&uplo, &trans, &n, &k, &alpha_, lhs, &lda, &beta_, res, &ldc); \
+  } \
+};
+
+// HERK for complex data
+#define EIGEN_MKL_RANKUPDATE_C(EIGTYPE, MKLTYPE, RTYPE, MKLFUNC) \
+template <typename Index, int AStorageOrder, bool ConjugateA, int  UpLo> \
+struct general_matrix_matrix_rankupdate<Index,EIGTYPE,AStorageOrder,ConjugateA,ColMajor,UpLo> { \
+  enum { \
+    IsLower = (UpLo&Lower) == Lower, \
+    LowUp = IsLower ? Lower : Upper, \
+    conjA = (((AStorageOrder==ColMajor) && ConjugateA) || ((AStorageOrder==RowMajor) && !ConjugateA)) ? 1 : 0 \
+  }; \
+  static EIGEN_STRONG_INLINE void run(Index size, Index depth,const EIGTYPE* lhs, Index lhsStride, \
+                          const EIGTYPE* rhs, Index rhsStride, EIGTYPE* res, Index resStride, EIGTYPE alpha) \
+  { \
+   typedef Matrix<EIGTYPE, Dynamic, Dynamic, AStorageOrder> MatrixType; \
+\
+   MKL_INT lda=lhsStride, ldc=resStride, n=size, k=depth; \
+   char uplo=(IsLower) ? 'L' : 'U', trans=(AStorageOrder==RowMajor) ? 'C':'N'; \
+   RTYPE alpha_, beta_; \
+   const EIGTYPE* a_ptr; \
+\
+/* Set alpha_ & beta_ */ \
+/*   assign_scalar_eig2mkl<MKLTYPE, EIGTYPE>(alpha_, alpha); */\
+/*   assign_scalar_eig2mkl<MKLTYPE, EIGTYPE>(beta_, EIGTYPE(1));*/ \
+   alpha_ = alpha.real(); \
+   beta_ = 1.0; \
+/* Copy with conjugation in some cases*/ \
+   MatrixType a; \
+   if (conjA) { \
+     Map<const MatrixType, 0, OuterStride<> > mapA(lhs,n,k,OuterStride<>(lhsStride)); \
+     a = mapA.conjugate(); \
+     lda = a.outerStride(); \
+     a_ptr = a.data(); \
+   } else a_ptr=lhs; \
+   MKLFUNC(&uplo, &trans, &n, &k, &alpha_, (MKLTYPE*)a_ptr, &lda, &beta_, (MKLTYPE*)res, &ldc); \
+  } \
+};
+
+
+EIGEN_MKL_RANKUPDATE_R(double, double, dsyrk)
+EIGEN_MKL_RANKUPDATE_R(float,  float,  ssyrk)
+
+//EIGEN_MKL_RANKUPDATE_C(dcomplex, MKL_Complex16, double, zherk)
+//EIGEN_MKL_RANKUPDATE_C(scomplex, MKL_Complex8,  double, cherk)
+
+
+} // end namespace internal
+
+#endif // EIGEN_GENERAL_MATRIX_MATRIX_TRIANGULAR_MKL_H

Eigen/src/Core/products/GeneralMatrixMatrix_MKL.h

@@ -0,0 +1,114 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *   General matrix-matrix product functionality based on ?GEMM.
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_GENERAL_MATRIX_MATRIX_MKL_H
+#define EIGEN_GENERAL_MATRIX_MATRIX_MKL_H
+
+namespace internal {
+
+/**********************************************************************
+* This file implements general matrix-matrix multiplication using BLAS
+* gemm function via partial specialization of
+* general_matrix_matrix_product::run(..) method for float, double,
+* std::complex<float> and std::complex<double> types
+**********************************************************************/
+
+// gemm specialization
+
+#define GEMM_SPECIALIZATION(EIGTYPE, EIGPREFIX, MKLTYPE, MKLPREFIX) \
+template< \
+  typename Index, \
+  int LhsStorageOrder, bool ConjugateLhs, \
+  int RhsStorageOrder, bool ConjugateRhs> \
+struct general_matrix_matrix_product<Index,EIGTYPE,LhsStorageOrder,ConjugateLhs,EIGTYPE,RhsStorageOrder,ConjugateRhs,ColMajor> \
+{ \
+static void run(Index rows, Index cols, Index depth, \
+  const EIGTYPE* _lhs, Index lhsStride, \
+  const EIGTYPE* _rhs, Index rhsStride, \
+  EIGTYPE* res, Index resStride, \
+  EIGTYPE alpha, \
+  level3_blocking<EIGTYPE, EIGTYPE>& blocking, \
+  GemmParallelInfo<Index>* info = 0) \
+{ \
+  using std::conj; \
+\
+  char transa, transb; \
+  MKL_INT m, n, k, lda, ldb, ldc; \
+  const EIGTYPE *a, *b; \
+  MKLTYPE alpha_, beta_; \
+  MatrixX##EIGPREFIX a_tmp, b_tmp; \
+  EIGTYPE myone(1);\
+\
+/* Set transpose options */ \
+  transa = (LhsStorageOrder==RowMajor) ? ((ConjugateLhs) ? 'C' : 'T') : 'N'; \
+  transb = (RhsStorageOrder==RowMajor) ? ((ConjugateRhs) ? 'C' : 'T') : 'N'; \
+\
+/* Set m, n, k */ \
+  m = (MKL_INT)rows;  \
+  n = (MKL_INT)cols;  \
+  k = (MKL_INT)depth; \
+\
+/* Set alpha_ & beta_ */ \
+  assign_scalar_eig2mkl(alpha_, alpha); \
+  assign_scalar_eig2mkl(beta_, myone); \
+\
+/* Set lda, ldb, ldc */ \
+  lda = (MKL_INT)lhsStride; \
+  ldb = (MKL_INT)rhsStride; \
+  ldc = (MKL_INT)resStride; \
+\
+/* Set a, b, c */ \
+  if ((LhsStorageOrder==ColMajor) && (ConjugateLhs)) { \
+    Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > lhs(_lhs,m,k,OuterStride<>(lhsStride)); \
+    a_tmp = lhs.conjugate(); \
+    a = a_tmp.data(); \
+    lda = a_tmp.outerStride(); \
+  } else a = _lhs; \
+\
+  if ((RhsStorageOrder==ColMajor) && (ConjugateRhs)) { \
+    Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > rhs(_rhs,k,n,OuterStride<>(rhsStride)); \
+    b_tmp = rhs.conjugate(); \
+    b = b_tmp.data(); \
+    ldb = b_tmp.outerStride(); \
+  } else b = _rhs; \
+\
+  MKLPREFIX##gemm(&transa, &transb, &m, &n, &k, &alpha_, (const MKLTYPE*)a, &lda, (const MKLTYPE*)b, &ldb, &beta_, (MKLTYPE*)res, &ldc); \
+}};
+
+GEMM_SPECIALIZATION(double,   d,  double,        d)
+GEMM_SPECIALIZATION(float,    f,  float,         s)
+GEMM_SPECIALIZATION(dcomplex, cd, MKL_Complex16, z)
+GEMM_SPECIALIZATION(scomplex, cf, MKL_Complex8,  c)
+
+} //end of namespase
+
+#endif // EIGEN_GENERAL_MATRIX_MATRIX_MKL_H

Eigen/src/Core/products/GeneralMatrixVector.h

@@ -40,8 +40,8 @@ namespace internal {
  *  |cplx |real |cplx | invalid, the caller has to do tmp: = A * B; C += alpha*tmp
  *  |cplx |real |real | optimal case, vectorization possible via real-cplx mul
  */
-template<typename Index, typename LhsScalar, bool ConjugateLhs, typename RhsScalar, bool ConjugateRhs>
-struct general_matrix_vector_product<Index,LhsScalar,ColMajor,ConjugateLhs,RhsScalar,ConjugateRhs>
+template<typename Index, typename LhsScalar, bool ConjugateLhs, typename RhsScalar, bool ConjugateRhs, int Version>
+struct general_matrix_vector_product<Index,LhsScalar,ColMajor,ConjugateLhs,RhsScalar,ConjugateRhs,Version>
 {
 typedef typename scalar_product_traits<LhsScalar, RhsScalar>::ReturnType ResScalar;
 
@@ -99,7 +99,7 @@ EIGEN_DONT_INLINE static void run(
   
   // How many coeffs of the result do we have to skip to be aligned.
   // Here we assume data are at least aligned on the base scalar type.
-  Index alignedStart = first_aligned(res,size);
+  Index alignedStart = internal::first_aligned(res,size);
   Index alignedSize = ResPacketSize>1 ? alignedStart + ((size-alignedStart) & ~ResPacketAlignedMask) : 0;
   const Index peeledSize  = peels>1 ? alignedStart + ((alignedSize-alignedStart) & ~PeelAlignedMask) : alignedStart;
 
@@ -109,7 +109,7 @@ EIGEN_DONT_INLINE static void run(
                        : FirstAligned;
 
   // we cannot assume the first element is aligned because of sub-matrices
-  const Index lhsAlignmentOffset = first_aligned(lhs,size);
+  const Index lhsAlignmentOffset = internal::first_aligned(lhs,size);
 
   // find how many columns do we have to skip to be aligned with the result (if possible)
   Index skipColumns = 0;
@@ -134,7 +134,7 @@ EIGEN_DONT_INLINE static void run(
     }
     else
     {
-      skipColumns = std::min(skipColumns,cols);
+      skipColumns = (std::min)(skipColumns,cols);
       // note that the skiped columns are processed later.
     }
 
@@ -296,8 +296,8 @@ EIGEN_DONT_INLINE static void run(
  *  - alpha is always a complex (or converted to a complex)
  *  - no vectorization
  */
-template<typename Index, typename LhsScalar, bool ConjugateLhs, typename RhsScalar, bool ConjugateRhs>
-struct general_matrix_vector_product<Index,LhsScalar,RowMajor,ConjugateLhs,RhsScalar,ConjugateRhs>
+template<typename Index, typename LhsScalar, bool ConjugateLhs, typename RhsScalar, bool ConjugateRhs, int Version>
+struct general_matrix_vector_product<Index,LhsScalar,RowMajor,ConjugateLhs,RhsScalar,ConjugateRhs,Version>
 {
 typedef typename scalar_product_traits<LhsScalar, RhsScalar>::ReturnType ResScalar;
 
@@ -351,7 +351,7 @@ EIGEN_DONT_INLINE static void run(
   // How many coeffs of the result do we have to skip to be aligned.
   // Here we assume data are at least aligned on the base scalar type
   // if that's not the case then vectorization is discarded, see below.
-  Index alignedStart = first_aligned(rhs, depth);
+  Index alignedStart = internal::first_aligned(rhs, depth);
   Index alignedSize = RhsPacketSize>1 ? alignedStart + ((depth-alignedStart) & ~RhsPacketAlignedMask) : 0;
   const Index peeledSize  = peels>1 ? alignedStart + ((alignedSize-alignedStart) & ~PeelAlignedMask) : alignedStart;
 
@@ -361,7 +361,7 @@ EIGEN_DONT_INLINE static void run(
                          : FirstAligned;
 
   // we cannot assume the first element is aligned because of sub-matrices
-  const Index lhsAlignmentOffset = first_aligned(lhs,depth);
+  const Index lhsAlignmentOffset = internal::first_aligned(lhs,depth);
 
   // find how many rows do we have to skip to be aligned with rhs (if possible)
   Index skipRows = 0;
@@ -386,7 +386,7 @@ EIGEN_DONT_INLINE static void run(
     }
     else
     {
-      skipRows = std::min(skipRows,Index(rows));
+      skipRows = (std::min)(skipRows,Index(rows));
       // note that the skiped columns are processed later.
     }
     eigen_internal_assert(  alignmentPattern==NoneAligned

Eigen/src/Core/products/GeneralMatrixVector_MKL.h

@@ -0,0 +1,127 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *   General matrix-vector product functionality based on ?GEMV.
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_GENERAL_MATRIX_VECTOR_MKL_H
+#define EIGEN_GENERAL_MATRIX_VECTOR_MKL_H
+
+namespace internal {
+
+/**********************************************************************
+* This file implements general matrix-vector multiplication using BLAS
+* gemv function via partial specialization of
+* general_matrix_vector_product::run(..) method for float, double,
+* std::complex<float> and std::complex<double> types
+**********************************************************************/
+
+// gemv specialization
+
+template<typename Index, typename LhsScalar, int LhsStorageOrder, bool ConjugateLhs, typename RhsScalar, bool ConjugateRhs>
+struct general_matrix_vector_product_gemv :
+  general_matrix_vector_product<Index,LhsScalar,LhsStorageOrder,ConjugateLhs,RhsScalar,ConjugateRhs,BuiltIn> {};
+
+#define EIGEN_MKL_GEMV_SPECIALIZE(Scalar) \
+template<typename Index, bool ConjugateLhs, bool ConjugateRhs> \
+struct general_matrix_vector_product<Index,Scalar,ColMajor,ConjugateLhs,Scalar,ConjugateRhs,Specialized> { \
+static EIGEN_DONT_INLINE void run( \
+  Index rows, Index cols, \
+  const Scalar* lhs, Index lhsStride, \
+  const Scalar* rhs, Index rhsIncr, \
+  Scalar* res, Index resIncr, Scalar alpha) \
+{ \
+  if (ConjugateLhs) { \
+    general_matrix_vector_product<Index,Scalar,ColMajor,ConjugateLhs,Scalar,ConjugateRhs,BuiltIn>::run( \
+      rows, cols, lhs, lhsStride, rhs, rhsIncr, res, resIncr, alpha); \
+  } else { \
+    general_matrix_vector_product_gemv<Index,Scalar,ColMajor,ConjugateLhs,Scalar,ConjugateRhs>::run( \
+      rows, cols, lhs, lhsStride, rhs, rhsIncr, res, resIncr, alpha); \
+  } \
+} \
+}; \
+template<typename Index, bool ConjugateLhs, bool ConjugateRhs> \
+struct general_matrix_vector_product<Index,Scalar,RowMajor,ConjugateLhs,Scalar,ConjugateRhs,Specialized> { \
+static EIGEN_DONT_INLINE void run( \
+  Index rows, Index cols, \
+  const Scalar* lhs, Index lhsStride, \
+  const Scalar* rhs, Index rhsIncr, \
+  Scalar* res, Index resIncr, Scalar alpha) \
+{ \
+    general_matrix_vector_product_gemv<Index,Scalar,RowMajor,ConjugateLhs,Scalar,ConjugateRhs>::run( \
+      rows, cols, lhs, lhsStride, rhs, rhsIncr, res, resIncr, alpha); \
+} \
+}; \
+
+EIGEN_MKL_GEMV_SPECIALIZE(double)
+EIGEN_MKL_GEMV_SPECIALIZE(float)
+EIGEN_MKL_GEMV_SPECIALIZE(dcomplex)
+EIGEN_MKL_GEMV_SPECIALIZE(scomplex)
+
+#define EIGEN_MKL_GEMV_SPECIALIZATION(EIGTYPE,MKLTYPE,MKLPREFIX) \
+template<typename Index, int LhsStorageOrder, bool ConjugateLhs, bool ConjugateRhs> \
+struct general_matrix_vector_product_gemv<Index,EIGTYPE,LhsStorageOrder,ConjugateLhs,EIGTYPE,ConjugateRhs> \
+{ \
+typedef Matrix<EIGTYPE,Dynamic,1,ColMajor> GEMVVector;\
+\
+static EIGEN_DONT_INLINE void run( \
+  Index rows, Index cols, \
+  const EIGTYPE* lhs, Index lhsStride, \
+  const EIGTYPE* rhs, Index rhsIncr, \
+  EIGTYPE* res, Index resIncr, EIGTYPE alpha) \
+{ \
+  MKL_INT m=rows, n=cols, lda=lhsStride, incx=rhsIncr, incy=resIncr; \
+  MKLTYPE alpha_, beta_; \
+  const EIGTYPE *x_ptr, myone(1); \
+  char trans=(LhsStorageOrder==ColMajor) ? 'N' : (ConjugateLhs) ? 'C' : 'T'; \
+  if (LhsStorageOrder==RowMajor) { \
+    m=cols; \
+    n=rows; \
+  }\
+  assign_scalar_eig2mkl(alpha_, alpha); \
+  assign_scalar_eig2mkl(beta_, myone); \
+  GEMVVector x_tmp; \
+  if (ConjugateRhs) { \
+    Map<const GEMVVector, 0, InnerStride<> > map_x(rhs,cols,1,InnerStride<>(incx)); \
+    x_tmp=map_x.conjugate(); \
+    x_ptr=x_tmp.data(); \
+    incx=1; \
+  } else x_ptr=rhs; \
+  MKLPREFIX##gemv(&trans, &m, &n, &alpha_, (const MKLTYPE*)lhs, &lda, (const MKLTYPE*)x_ptr, &incx, &beta_, (MKLTYPE*)res, &incy); \
+}\
+};
+
+EIGEN_MKL_GEMV_SPECIALIZATION(double,   double,        d)
+EIGEN_MKL_GEMV_SPECIALIZATION(float,    float,         s)
+EIGEN_MKL_GEMV_SPECIALIZATION(dcomplex, MKL_Complex16, z)
+EIGEN_MKL_GEMV_SPECIALIZATION(scomplex, MKL_Complex8,  c)
+
+} //end of namespase
+
+#endif // EIGEN_GENERAL_MATRIX_VECTOR_MKL_H

Eigen/src/Core/products/Parallelizer.h

@@ -30,7 +30,7 @@ namespace internal {
 /** \internal */
 inline void manage_multi_threading(Action action, int* v)
 {
-  static int m_maxThreads = -1;
+  static EIGEN_UNUSED int m_maxThreads = -1;
 
   if(action==SetAction)
   {
@@ -85,7 +85,9 @@ template<typename Index> struct GemmParallelInfo
 template<bool Condition, typename Functor, typename Index>
 void parallelize_gemm(const Functor& func, Index rows, Index cols, bool transpose)
 {
-#ifndef EIGEN_HAS_OPENMP
+  // TODO when EIGEN_USE_BLAS is defined,
+  // we should still enable OMP for other scalar types
+#if !(defined (EIGEN_HAS_OPENMP)) || defined (EIGEN_USE_BLAS)
   // FIXME the transpose variable is only needed to properly split
   // the matrix product when multithreading is enabled. This is a temporary
   // fix to support row-major destination matrices. This whole

Eigen/src/Core/products/SelfadjointMatrixMatrix.h

@@ -114,7 +114,7 @@ struct symm_pack_rhs
     }
 
     // second part: diagonal block
-    for(Index j2=k2; j2<std::min(k2+rows,packet_cols); j2+=nr)
+    for(Index j2=k2; j2<(std::min)(k2+rows,packet_cols); j2+=nr)
     {
       // again we can split vertically in three different parts (transpose, symmetric, normal)
       // transpose
@@ -179,7 +179,7 @@ struct symm_pack_rhs
     for(Index j2=packet_cols; j2<cols; ++j2)
     {
       // transpose
-      Index half = std::min(end_k,j2);
+      Index half = (std::min)(end_k,j2);
       for(Index k=k2; k<half; k++)
       {
         blockB[count] = conj(rhs(j2,k));
@@ -261,12 +261,12 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,true,ConjugateLhs
     Index nc = cols;  // cache block size along the N direction
     computeProductBlockingSizes<Scalar,Scalar>(kc, mc, nc);
     // kc must smaller than mc
-    kc = std::min(kc,mc);
+    kc = (std::min)(kc,mc);
 
-    Scalar* blockA = ei_aligned_stack_new(Scalar, kc*mc);
     std::size_t sizeW = kc*Traits::WorkSpaceFactor;
     std::size_t sizeB = sizeW + kc*cols;
-    Scalar* allocatedBlockB = ei_aligned_stack_new(Scalar, sizeB);
+    ei_declare_aligned_stack_constructed_variable(Scalar, blockA, kc*mc, 0);
+    ei_declare_aligned_stack_constructed_variable(Scalar, allocatedBlockB, sizeB, 0);
     Scalar* blockB = allocatedBlockB + sizeW;
 
     gebp_kernel<Scalar, Scalar, Index, Traits::mr, Traits::nr, ConjugateLhs, ConjugateRhs> gebp_kernel;
@@ -276,7 +276,7 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,true,ConjugateLhs
 
     for(Index k2=0; k2<size; k2+=kc)
     {
-      const Index actual_kc = std::min(k2+kc,size)-k2;
+      const Index actual_kc = (std::min)(k2+kc,size)-k2;
 
       // we have selected one row panel of rhs and one column panel of lhs
       // pack rhs's panel into a sequential chunk of memory
@@ -289,7 +289,7 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,true,ConjugateLhs
       //  3 - the panel below the diagonal block => generic packed copy
       for(Index i2=0; i2<k2; i2+=mc)
       {
-        const Index actual_mc = std::min(i2+mc,k2)-i2;
+        const Index actual_mc = (std::min)(i2+mc,k2)-i2;
         // transposed packed copy
         pack_lhs_transposed(blockA, &lhs(k2, i2), lhsStride, actual_kc, actual_mc);
 
@@ -297,7 +297,7 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,true,ConjugateLhs
       }
       // the block diagonal
       {
-        const Index actual_mc = std::min(k2+kc,size)-k2;
+        const Index actual_mc = (std::min)(k2+kc,size)-k2;
         // symmetric packed copy
         pack_lhs(blockA, &lhs(k2,k2), lhsStride, actual_kc, actual_mc);
 
@@ -306,16 +306,13 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,true,ConjugateLhs
 
       for(Index i2=k2+kc; i2<size; i2+=mc)
       {
-        const Index actual_mc = std::min(i2+mc,size)-i2;
+        const Index actual_mc = (std::min)(i2+mc,size)-i2;
         gemm_pack_lhs<Scalar, Index, Traits::mr, Traits::LhsProgress, LhsStorageOrder,false>()
           (blockA, &lhs(i2, k2), lhsStride, actual_kc, actual_mc);
 
         gebp_kernel(res+i2, resStride, blockA, blockB, actual_mc, actual_kc, cols, alpha);
       }
     }
-
-    ei_aligned_stack_delete(Scalar, blockA, kc*mc);
-    ei_aligned_stack_delete(Scalar, allocatedBlockB, sizeB);
   }
 };
 
@@ -343,11 +340,10 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,false,ConjugateLh
     Index mc = rows;  // cache block size along the M direction
     Index nc = cols;  // cache block size along the N direction
     computeProductBlockingSizes<Scalar,Scalar>(kc, mc, nc);
-
-    Scalar* blockA = ei_aligned_stack_new(Scalar, kc*mc);
     std::size_t sizeW = kc*Traits::WorkSpaceFactor;
     std::size_t sizeB = sizeW + kc*cols;
-    Scalar* allocatedBlockB = ei_aligned_stack_new(Scalar, sizeB);
+    ei_declare_aligned_stack_constructed_variable(Scalar, blockA, kc*mc, 0);
+    ei_declare_aligned_stack_constructed_variable(Scalar, allocatedBlockB, sizeB, 0);
     Scalar* blockB = allocatedBlockB + sizeW;
 
     gebp_kernel<Scalar, Scalar, Index, Traits::mr, Traits::nr, ConjugateLhs, ConjugateRhs> gebp_kernel;
@@ -356,22 +352,19 @@ struct product_selfadjoint_matrix<Scalar,Index,LhsStorageOrder,false,ConjugateLh
 
     for(Index k2=0; k2<size; k2+=kc)
     {
-      const Index actual_kc = std::min(k2+kc,size)-k2;
+      const Index actual_kc = (std::min)(k2+kc,size)-k2;
 
       pack_rhs(blockB, _rhs, rhsStride, actual_kc, cols, k2);
 
       // => GEPP
       for(Index i2=0; i2<rows; i2+=mc)
       {
-        const Index actual_mc = std::min(i2+mc,rows)-i2;
+        const Index actual_mc = (std::min)(i2+mc,rows)-i2;
         pack_lhs(blockA, &lhs(i2, k2), lhsStride, actual_kc, actual_mc);
 
         gebp_kernel(res+i2, resStride, blockA, blockB, actual_mc, actual_kc, cols, alpha);
       }
     }
-
-    ei_aligned_stack_delete(Scalar, blockA, kc*mc);
-    ei_aligned_stack_delete(Scalar, allocatedBlockB, sizeB);
   }
 };
 
@@ -407,8 +400,8 @@ struct SelfadjointProductMatrix<Lhs,LhsMode,false,Rhs,RhsMode,false>
   {
     eigen_assert(dst.rows()==m_lhs.rows() && dst.cols()==m_rhs.cols());
 
-    const ActualLhsType lhs = LhsBlasTraits::extract(m_lhs);
-    const ActualRhsType rhs = RhsBlasTraits::extract(m_rhs);
+    typename internal::add_const_on_value_type<ActualLhsType>::type lhs = LhsBlasTraits::extract(m_lhs);
+    typename internal::add_const_on_value_type<ActualRhsType>::type rhs = RhsBlasTraits::extract(m_rhs);
 
     Scalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(m_lhs)
                                * RhsBlasTraits::extractScalarFactor(m_rhs);

Eigen/src/Core/products/SelfadjointMatrixMatrix_MKL.h

@@ -0,0 +1,291 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *   Self adjoint matrix * matrix product functionality based on ?SYMM/?HEMM.
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_SELFADJOINT_MATRIX_MATRIX_MKL_H
+#define EIGEN_SELFADJOINT_MATRIX_MATRIX_MKL_H
+
+namespace internal {
+
+
+/* Optimized selfadjoint matrix * matrix (?SYMM/?HEMM) product */
+
+#define EIGEN_MKL_SYMM_L(EIGTYPE, MKLTYPE, EIGPREFIX, MKLPREFIX) \
+template <typename Index, \
+          int LhsStorageOrder, bool ConjugateLhs, \
+          int RhsStorageOrder, bool ConjugateRhs> \
+struct product_selfadjoint_matrix<EIGTYPE,Index,LhsStorageOrder,true,ConjugateLhs,RhsStorageOrder,false,ConjugateRhs,ColMajor> \
+{\
+\
+  static EIGEN_DONT_INLINE void run( \
+    Index rows, Index cols, \
+    const EIGTYPE* _lhs, Index lhsStride, \
+    const EIGTYPE* _rhs, Index rhsStride, \
+    EIGTYPE* res,        Index resStride, \
+    EIGTYPE alpha) \
+  { \
+    char side='L', uplo='L'; \
+    MKL_INT m, n, lda, ldb, ldc; \
+    const EIGTYPE *a, *b; \
+    MKLTYPE alpha_, beta_; \
+    MatrixX##EIGPREFIX b_tmp; \
+    EIGTYPE myone(1);\
+\
+/* Set transpose options */ \
+/* Set m, n, k */ \
+    m = (MKL_INT)rows;  \
+    n = (MKL_INT)cols;  \
+\
+/* Set alpha_ & beta_ */ \
+    assign_scalar_eig2mkl(alpha_, alpha); \
+    assign_scalar_eig2mkl(beta_, myone); \
+\
+/* Set lda, ldb, ldc */ \
+    lda = (MKL_INT)lhsStride; \
+    ldb = (MKL_INT)rhsStride; \
+    ldc = (MKL_INT)resStride; \
+\
+/* Set a, b, c */ \
+    if (LhsStorageOrder==RowMajor) uplo='U'; \
+    a = _lhs; \
+\
+    if (RhsStorageOrder==RowMajor) { \
+      Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > rhs(_rhs,n,m,OuterStride<>(rhsStride)); \
+      b_tmp = rhs.adjoint(); \
+      b = b_tmp.data(); \
+      ldb = b_tmp.outerStride(); \
+    } else b = _rhs; \
+\
+    MKLPREFIX##symm(&side, &uplo, &m, &n, &alpha_, (const MKLTYPE*)a, &lda, (const MKLTYPE*)b, &ldb, &beta_, (MKLTYPE*)res, &ldc); \
+\
+  } \
+};
+
+
+#define EIGEN_MKL_HEMM_L(EIGTYPE, MKLTYPE, EIGPREFIX, MKLPREFIX) \
+template <typename Index, \
+          int LhsStorageOrder, bool ConjugateLhs, \
+          int RhsStorageOrder, bool ConjugateRhs> \
+struct product_selfadjoint_matrix<EIGTYPE,Index,LhsStorageOrder,true,ConjugateLhs,RhsStorageOrder,false,ConjugateRhs,ColMajor> \
+{\
+  static EIGEN_DONT_INLINE void run( \
+    Index rows, Index cols, \
+    const EIGTYPE* _lhs, Index lhsStride, \
+    const EIGTYPE* _rhs, Index rhsStride, \
+    EIGTYPE* res,        Index resStride, \
+    EIGTYPE alpha) \
+  { \
+    char side='L', uplo='L'; \
+    MKL_INT m, n, lda, ldb, ldc; \
+    const EIGTYPE *a, *b; \
+    MKLTYPE alpha_, beta_; \
+    MatrixX##EIGPREFIX b_tmp; \
+    Matrix<EIGTYPE, Dynamic, Dynamic, LhsStorageOrder> a_tmp; \
+    EIGTYPE myone(1); \
+\
+/* Set transpose options */ \
+/* Set m, n, k */ \
+    m = (MKL_INT)rows; \
+    n = (MKL_INT)cols; \
+\
+/* Set alpha_ & beta_ */ \
+    assign_scalar_eig2mkl(alpha_, alpha); \
+    assign_scalar_eig2mkl(beta_, myone); \
+\
+/* Set lda, ldb, ldc */ \
+    lda = (MKL_INT)lhsStride; \
+    ldb = (MKL_INT)rhsStride; \
+    ldc = (MKL_INT)resStride; \
+\
+/* Set a, b, c */ \
+    if (((LhsStorageOrder==ColMajor) && ConjugateLhs) || ((LhsStorageOrder==RowMajor) && (!ConjugateLhs))) { \
+      Map<const Matrix<EIGTYPE, Dynamic, Dynamic, LhsStorageOrder>, 0, OuterStride<> > lhs(_lhs,m,m,OuterStride<>(lhsStride)); \
+      a_tmp = lhs.conjugate(); \
+      a = a_tmp.data(); \
+      lda = a_tmp.outerStride(); \
+    } else a = _lhs; \
+    if (LhsStorageOrder==RowMajor) uplo='U'; \
+\
+    if (RhsStorageOrder==ColMajor && (!ConjugateRhs)) { \
+       b = _rhs; } \
+    else { \
+      if (RhsStorageOrder==ColMajor && ConjugateRhs) { \
+        Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > rhs(_rhs,m,n,OuterStride<>(rhsStride)); \
+        b_tmp = rhs.conjugate(); \
+      } else \
+      if (ConjugateRhs) { \
+        Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > rhs(_rhs,n,m,OuterStride<>(rhsStride)); \
+        b_tmp = rhs.adjoint(); \
+      } else { \
+        Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > rhs(_rhs,n,m,OuterStride<>(rhsStride)); \
+        b_tmp = rhs.transpose(); \
+      } \
+      b = b_tmp.data(); \
+      ldb = b_tmp.outerStride(); \
+    } \
+\
+    MKLPREFIX##hemm(&side, &uplo, &m, &n, &alpha_, (const MKLTYPE*)a, &lda, (const MKLTYPE*)b, &ldb, &beta_, (MKLTYPE*)res, &ldc); \
+\
+  } \
+};
+
+EIGEN_MKL_SYMM_L(double, double, d, d)
+EIGEN_MKL_SYMM_L(float, float, f, s)
+EIGEN_MKL_HEMM_L(dcomplex, MKL_Complex16, cd, z)
+EIGEN_MKL_HEMM_L(scomplex, MKL_Complex8, cf, c)
+
+
+/* Optimized matrix * selfadjoint matrix (?SYMM/?HEMM) product */
+
+#define EIGEN_MKL_SYMM_R(EIGTYPE, MKLTYPE, EIGPREFIX, MKLPREFIX) \
+template <typename Index, \
+          int LhsStorageOrder, bool ConjugateLhs, \
+          int RhsStorageOrder, bool ConjugateRhs> \
+struct product_selfadjoint_matrix<EIGTYPE,Index,LhsStorageOrder,false,ConjugateLhs,RhsStorageOrder,true,ConjugateRhs,ColMajor> \
+{\
+\
+  static EIGEN_DONT_INLINE void run( \
+    Index rows, Index cols, \
+    const EIGTYPE* _lhs, Index lhsStride, \
+    const EIGTYPE* _rhs, Index rhsStride, \
+    EIGTYPE* res,        Index resStride, \
+    EIGTYPE alpha) \
+  { \
+    char side='R', uplo='L'; \
+    MKL_INT m, n, lda, ldb, ldc; \
+    const EIGTYPE *a, *b; \
+    MKLTYPE alpha_, beta_; \
+    MatrixX##EIGPREFIX b_tmp; \
+    EIGTYPE myone(1);\
+\
+/* Set m, n, k */ \
+    m = (MKL_INT)rows;  \
+    n = (MKL_INT)cols;  \
+\
+/* Set alpha_ & beta_ */ \
+    assign_scalar_eig2mkl(alpha_, alpha); \
+    assign_scalar_eig2mkl(beta_, myone); \
+\
+/* Set lda, ldb, ldc */ \
+    lda = (MKL_INT)rhsStride; \
+    ldb = (MKL_INT)lhsStride; \
+    ldc = (MKL_INT)resStride; \
+\
+/* Set a, b, c */ \
+    if (RhsStorageOrder==RowMajor) uplo='U'; \
+    a = _rhs; \
+\
+    if (LhsStorageOrder==RowMajor) { \
+      Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > lhs(_lhs,n,m,OuterStride<>(rhsStride)); \
+      b_tmp = lhs.adjoint(); \
+      b = b_tmp.data(); \
+      ldb = b_tmp.outerStride(); \
+    } else b = _lhs; \
+\
+    MKLPREFIX##symm(&side, &uplo, &m, &n, &alpha_, (const MKLTYPE*)a, &lda, (const MKLTYPE*)b, &ldb, &beta_, (MKLTYPE*)res, &ldc); \
+\
+  } \
+};
+
+
+#define EIGEN_MKL_HEMM_R(EIGTYPE, MKLTYPE, EIGPREFIX, MKLPREFIX) \
+template <typename Index, \
+          int LhsStorageOrder, bool ConjugateLhs, \
+          int RhsStorageOrder, bool ConjugateRhs> \
+struct product_selfadjoint_matrix<EIGTYPE,Index,LhsStorageOrder,false,ConjugateLhs,RhsStorageOrder,true,ConjugateRhs,ColMajor> \
+{\
+  static EIGEN_DONT_INLINE void run( \
+    Index rows, Index cols, \
+    const EIGTYPE* _lhs, Index lhsStride, \
+    const EIGTYPE* _rhs, Index rhsStride, \
+    EIGTYPE* res,        Index resStride, \
+    EIGTYPE alpha) \
+  { \
+    char side='R', uplo='L'; \
+    MKL_INT m, n, lda, ldb, ldc; \
+    const EIGTYPE *a, *b; \
+    MKLTYPE alpha_, beta_; \
+    MatrixX##EIGPREFIX b_tmp; \
+    Matrix<EIGTYPE, Dynamic, Dynamic, RhsStorageOrder> a_tmp; \
+    EIGTYPE myone(1); \
+\
+/* Set m, n, k */ \
+    m = (MKL_INT)rows; \
+    n = (MKL_INT)cols; \
+\
+/* Set alpha_ & beta_ */ \
+    assign_scalar_eig2mkl(alpha_, alpha); \
+    assign_scalar_eig2mkl(beta_, myone); \
+\
+/* Set lda, ldb, ldc */ \
+    lda = (MKL_INT)rhsStride; \
+    ldb = (MKL_INT)lhsStride; \
+    ldc = (MKL_INT)resStride; \
+\
+/* Set a, b, c */ \
+    if (((RhsStorageOrder==ColMajor) && ConjugateRhs) || ((RhsStorageOrder==RowMajor) && (!ConjugateRhs))) { \
+      Map<const Matrix<EIGTYPE, Dynamic, Dynamic, RhsStorageOrder>, 0, OuterStride<> > rhs(_rhs,n,n,OuterStride<>(rhsStride)); \
+      a_tmp = rhs.conjugate(); \
+      a = a_tmp.data(); \
+      lda = a_tmp.outerStride(); \
+    } else a = _rhs; \
+    if (RhsStorageOrder==RowMajor) uplo='U'; \
+\
+    if (LhsStorageOrder==ColMajor && (!ConjugateLhs)) { \
+       b = _lhs; } \
+    else { \
+      if (LhsStorageOrder==ColMajor && ConjugateLhs) { \
+        Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > lhs(_lhs,m,n,OuterStride<>(lhsStride)); \
+        b_tmp = lhs.conjugate(); \
+      } else \
+      if (ConjugateLhs) { \
+        Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > lhs(_lhs,n,m,OuterStride<>(lhsStride)); \
+        b_tmp = lhs.adjoint(); \
+      } else { \
+        Map<const MatrixX##EIGPREFIX, 0, OuterStride<> > lhs(_lhs,n,m,OuterStride<>(lhsStride)); \
+        b_tmp = lhs.transpose(); \
+      } \
+      b = b_tmp.data(); \
+      ldb = b_tmp.outerStride(); \
+    } \
+\
+    MKLPREFIX##hemm(&side, &uplo, &m, &n, &alpha_, (const MKLTYPE*)a, &lda, (const MKLTYPE*)b, &ldb, &beta_, (MKLTYPE*)res, &ldc); \
+  } \
+};
+
+EIGEN_MKL_SYMM_R(double, double, d, d)
+EIGEN_MKL_SYMM_R(float, float, f, s)
+EIGEN_MKL_HEMM_R(dcomplex, MKL_Complex16, cd, z)
+EIGEN_MKL_HEMM_R(scomplex, MKL_Complex8, cf, c)
+
+} // end namespace internal
+
+#endif // EIGEN_SELFADJOINT_MATRIX_MATRIX_MKL_H

Eigen/src/Core/products/SelfadjointMatrixVector.h

@@ -32,8 +32,15 @@ namespace internal {
  * the number of load/stores of the result by a factor 2 and to reduce
  * the instruction dependency.
  */
-template<typename Scalar, typename Index, int StorageOrder, int UpLo, bool ConjugateLhs, bool ConjugateRhs>
-static EIGEN_DONT_INLINE void product_selfadjoint_vector(
+
+template<typename Scalar, typename Index, int StorageOrder, int UpLo, bool ConjugateLhs, bool ConjugateRhs, int Version=Specialized>
+struct selfadjoint_matrix_vector_product;
+
+template<typename Scalar, typename Index, int StorageOrder, int UpLo, bool ConjugateLhs, bool ConjugateRhs, int Version>
+struct selfadjoint_matrix_vector_product
+
+{
+static EIGEN_DONT_INLINE void run(
   Index size,
   const Scalar*  lhs, Index lhsStride,
   const Scalar* _rhs, Index rhsIncr,
@@ -62,17 +69,15 @@ static EIGEN_DONT_INLINE void product_selfadjoint_vector(
   // FIXME this copy is now handled outside product_selfadjoint_vector, so it could probably be removed.
   // if the rhs is not sequentially stored in memory we copy it to a temporary buffer,
   // this is because we need to extract packets
-  const Scalar* EIGEN_RESTRICT rhs = _rhs;
+  ei_declare_aligned_stack_constructed_variable(Scalar,rhs,size,rhsIncr==1 ? const_cast<Scalar*>(_rhs) : 0);  
   if (rhsIncr!=1)
   {
-    Scalar* r = ei_aligned_stack_new(Scalar, size);
     const Scalar* it = _rhs;
     for (Index i=0; i<size; ++i, it+=rhsIncr)
-      r[i] = *it;
-    rhs = r;
+      rhs[i] = *it;
   }
 
-  Index bound = std::max(Index(0),size-8) & 0xfffffffe;
+  Index bound = (std::max)(Index(0),size-8) & 0xfffffffe;
   if (FirstTriangular)
     bound = size - bound;
 
@@ -87,14 +92,14 @@ static EIGEN_DONT_INLINE void product_selfadjoint_vector(
     Scalar t1 = cjAlpha * rhs[j+1];
     Packet ptmp1 = pset1<Packet>(t1);
 
-    Scalar t2 = 0;
+    Scalar t2(0);
     Packet ptmp2 = pset1<Packet>(t2);
-    Scalar t3 = 0;
+    Scalar t3(0);
     Packet ptmp3 = pset1<Packet>(t3);
 
     size_t starti = FirstTriangular ? 0 : j+2;
     size_t endi   = FirstTriangular ? j : size;
-    size_t alignedStart = (starti) + first_aligned(&res[starti], endi-starti);
+    size_t alignedStart = (starti) + internal::first_aligned(&res[starti], endi-starti);
     size_t alignedEnd = alignedStart + ((endi-alignedStart)/(PacketSize))*(PacketSize);
 
     // TODO make sure this product is a real * complex and that the rhs is properly conjugated if needed
@@ -150,7 +155,7 @@ static EIGEN_DONT_INLINE void product_selfadjoint_vector(
     register const Scalar* EIGEN_RESTRICT A0 = lhs + j*lhsStride;
 
     Scalar t1 = cjAlpha * rhs[j];
-    Scalar t2 = 0;
+    Scalar t2(0);
     // TODO make sure this product is a real * complex and that the rhs is properly conjugated if needed
     res[j] += cjd.pmul(internal::real(A0[j]), t1);
     for (Index i=FirstTriangular ? 0 : j+1; i<(FirstTriangular ? j : size); i++)
@@ -160,10 +165,8 @@ static EIGEN_DONT_INLINE void product_selfadjoint_vector(
     }
     res[j] += alpha * t2;
   }
-
-  if(rhsIncr!=1)
-    ei_aligned_stack_delete(Scalar, const_cast<Scalar*>(rhs), size);
 }
+};
 
 } // end namespace internal 
 
@@ -198,8 +201,8 @@ struct SelfadjointProductMatrix<Lhs,LhsMode,false,Rhs,0,true>
     
     eigen_assert(dest.rows()==m_lhs.rows() && dest.cols()==m_rhs.cols());
 
-    const ActualLhsType lhs = LhsBlasTraits::extract(m_lhs);
-    const ActualRhsType rhs = RhsBlasTraits::extract(m_rhs);
+    typename internal::add_const_on_value_type<ActualLhsType>::type lhs = LhsBlasTraits::extract(m_lhs);
+    typename internal::add_const_on_value_type<ActualRhsType>::type rhs = RhsBlasTraits::extract(m_rhs);
 
     Scalar actualAlpha = alpha * LhsBlasTraits::extractScalarFactor(m_lhs)
                                * RhsBlasTraits::extractScalarFactor(m_rhs);
@@ -211,45 +214,33 @@ struct SelfadjointProductMatrix<Lhs,LhsMode,false,Rhs,0,true>
     
     internal::gemv_static_vector_if<ResScalar,Dest::SizeAtCompileTime,Dest::MaxSizeAtCompileTime,!EvalToDest> static_dest;
     internal::gemv_static_vector_if<RhsScalar,_ActualRhsType::SizeAtCompileTime,_ActualRhsType::MaxSizeAtCompileTime,!UseRhs> static_rhs;
+
+    ei_declare_aligned_stack_constructed_variable(ResScalar,actualDestPtr,dest.size(),
+                                                  EvalToDest ? dest.data() : static_dest.data());
+                                                  
+    ei_declare_aligned_stack_constructed_variable(RhsScalar,actualRhsPtr,rhs.size(),
+        UseRhs ? const_cast<RhsScalar*>(rhs.data()) : static_rhs.data());
     
-    bool freeDestPtr = false;
-    ResScalar* actualDestPtr;
-    if(EvalToDest)
-      actualDestPtr = dest.data();
-    else
+    if(!EvalToDest)
     {
       #ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
       int size = dest.size();
       EIGEN_DENSE_STORAGE_CTOR_PLUGIN
       #endif
-      if((actualDestPtr=static_dest.data())==0)
-      {
-        freeDestPtr = true;
-        actualDestPtr = ei_aligned_stack_new(ResScalar,dest.size());
-      }
       MappedDest(actualDestPtr, dest.size()) = dest;
     }
       
-    bool freeRhsPtr = false;
-    RhsScalar* actualRhsPtr;
-    if(UseRhs)
-      actualRhsPtr = const_cast<RhsScalar*>(rhs.data());
-    else
+    if(!UseRhs)
     {
       #ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
       int size = rhs.size();
       EIGEN_DENSE_STORAGE_CTOR_PLUGIN
       #endif
-      if((actualRhsPtr=static_rhs.data())==0)
-      {
-        freeRhsPtr = true;
-        actualRhsPtr = ei_aligned_stack_new(RhsScalar,rhs.size());
-      }
       Map<typename _ActualRhsType::PlainObject>(actualRhsPtr, rhs.size()) = rhs;
     }
       
       
-    internal::product_selfadjoint_vector<Scalar, Index, (internal::traits<_ActualLhsType>::Flags&RowMajorBit) ? RowMajor : ColMajor, int(LhsUpLo), bool(LhsBlasTraits::NeedToConjugate), bool(RhsBlasTraits::NeedToConjugate)>
+    internal::selfadjoint_matrix_vector_product<Scalar, Index, (internal::traits<_ActualLhsType>::Flags&RowMajorBit) ? RowMajor : ColMajor, int(LhsUpLo), bool(LhsBlasTraits::NeedToConjugate), bool(RhsBlasTraits::NeedToConjugate)>::run
       (
         lhs.rows(),                             // size
         &lhs.coeffRef(0,0),  lhs.outerStride(), // lhs info
@@ -259,11 +250,7 @@ struct SelfadjointProductMatrix<Lhs,LhsMode,false,Rhs,0,true>
       );
     
     if(!EvalToDest)
-    {
       dest = MappedDest(actualDestPtr, dest.size());
-      if(freeDestPtr) ei_aligned_stack_delete(ResScalar, actualDestPtr, dest.size());
-    }
-    if(freeRhsPtr) ei_aligned_stack_delete(RhsScalar, actualRhsPtr, rhs.size());
   }
 };
 

Eigen/src/Core/products/SelfadjointMatrixVector_MKL.h

@@ -0,0 +1,110 @@
+/*
+ Copyright (c) 2011, Intel Corporation. All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without modification,
+ are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of Intel Corporation nor the names of its contributors may
+   be used to endorse or promote products derived from this software without
+   specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ ********************************************************************************
+ *   Content : Eigen bindings to Intel(R) MKL
+ *   Selfadjoint matrix-vector product functionality based on ?SYMV/HEMV.
+ ********************************************************************************
+*/
+
+#ifndef EIGEN_SELFADJOINT_MATRIX_VECTOR_MKL_H
+#define EIGEN_SELFADJOINT_MATRIX_VECTOR_MKL_H
+
+namespace internal {
+
+/**********************************************************************
+* This file implements selfadjoint matrix-vector multiplication using BLAS
+**********************************************************************/
+
+// symv/hemv specialization
+
+template<typename Scalar, typename Index, int StorageOrder, int UpLo, bool ConjugateLhs, bool ConjugateRhs>
+struct selfadjoint_matrix_vector_product_symv :
+  selfadjoint_matrix_vector_product<Scalar,Index,StorageOrder,UpLo,ConjugateLhs,ConjugateRhs,BuiltIn> {};
+
+#define EIGEN_MKL_SYMV_SPECIALIZE(Scalar) \
+template<typename Index, int StorageOrder, int UpLo, bool ConjugateLhs, bool ConjugateRhs> \
+struct selfadjoint_matrix_vector_product<Scalar,Index,StorageOrder,UpLo,ConjugateLhs,ConjugateRhs,Specialized> { \
+static EIGEN_DONT_INLINE void run( \
+  Index size, const Scalar*  lhs, Index lhsStride, \
+  const Scalar* _rhs, Index rhsIncr, Scalar* res, Scalar alpha) { \
+    enum {\
+      IsColMajor = StorageOrder==ColMajor \
+    }; \
+    if (IsColMajor == ConjugateLhs) {\
+      selfadjoint_matrix_vector_product<Scalar,Index,StorageOrder,UpLo,ConjugateLhs,ConjugateRhs,BuiltIn>::run( \
+        size, lhs, lhsStride, _rhs, rhsIncr, res, alpha);  \
+    } else {\
+      selfadjoint_matrix_vector_product_symv<Scalar,Index,StorageOrder,UpLo,ConjugateLhs,ConjugateRhs>::run( \
+        size, lhs, lhsStride, _rhs, rhsIncr, res, alpha);  \
+    }\
+  } \
+}; \
+
+EIGEN_MKL_SYMV_SPECIALIZE(double)
+EIGEN_MKL_SYMV_SPECIALIZE(float)
+EIGEN_MKL_SYMV_SPECIALIZE(dcomplex)
+EIGEN_MKL_SYMV_SPECIALIZE(scomplex)
+
+#define EIGEN_MKL_SYMV_SPECIALIZATION(EIGTYPE,MKLTYPE,MKLFUNC) \
+template<typename Index, int StorageOrder, int UpLo, bool ConjugateLhs, bool ConjugateRhs> \
+struct selfadjoint_matrix_vector_product_symv<EIGTYPE,Index,StorageOrder,UpLo,ConjugateLhs,ConjugateRhs> \
+{ \
+typedef Matrix<EIGTYPE,Dynamic,1,ColMajor> SYMVVector;\
+\
+static EIGEN_DONT_INLINE void run( \
+Index size, const EIGTYPE*  lhs, Index lhsStride, \
+const EIGTYPE* _rhs, Index rhsIncr, EIGTYPE* res, EIGTYPE alpha) \
+{ \
+  enum {\
+    IsRowMajor = StorageOrder==RowMajor ? 1 : 0, \
+    IsLower = UpLo == Lower ? 1 : 0, \
+  }; \
+  MKL_INT n=size, lda=lhsStride, incx=rhsIncr, incy=1; \
+  MKLTYPE alpha_, beta_; \
+  const EIGTYPE *x_ptr, myone(1); \
+  char uplo=(IsRowMajor) ? (IsLower ? 'U' : 'L') : (IsLower ? 'L' : 'U'); \
+  assign_scalar_eig2mkl(alpha_, alpha); \
+  assign_scalar_eig2mkl(beta_, myone); \
+  SYMVVector x_tmp; \
+  if (ConjugateRhs) { \
+    Map<const SYMVVector, 0, InnerStride<> > map_x(_rhs,size,1,InnerStride<>(incx)); \
+    x_tmp=map_x.conjugate(); \
+    x_ptr=x_tmp.data(); \
+    incx=1; \
+  } else x_ptr=_rhs; \
+  MKLFUNC(&uplo, &n, &alpha_, (const MKLTYPE*)lhs, &lda, (const MKLTYPE*)x_ptr, &incx, &beta_, (MKLTYPE*)res, &incy); \
+}\
+};
+
+EIGEN_MKL_SYMV_SPECIALIZATION(double,   double,        dsymv)
+EIGEN_MKL_SYMV_SPECIALIZATION(float,    float,         ssymv)
+EIGEN_MKL_SYMV_SPECIALIZATION(dcomplex, MKL_Complex16, zhemv)
+EIGEN_MKL_SYMV_SPECIALIZATION(scomplex, MKL_Complex8,  chemv)
+
+} //end of namespase
+
+#endif // EIGEN_SELFADJOINT_MATRIX_VECTOR_MKL_H