Pulled latest updates from the Eigen main trunk.

This commit is contained in:
Benoit Steiner
2014-06-10 10:23:32 -07:00
37 changed files with 337 additions and 151 deletions

View File

@@ -84,7 +84,7 @@ struct traits<Block<XprType, BlockRows, BlockCols, InnerPanel> > : traits<XprTyp
&& (InnerStrideAtCompileTime == 1)
? PacketAccessBit : 0,
MaskAlignedBit = (InnerPanel && (OuterStrideAtCompileTime!=Dynamic) && (((OuterStrideAtCompileTime * int(sizeof(Scalar))) % EIGEN_ALIGN_BYTES) == 0)) ? AlignedBit : 0,
FlagsLinearAccessBit = (RowsAtCompileTime == 1 || ColsAtCompileTime == 1) ? LinearAccessBit : 0,
FlagsLinearAccessBit = (RowsAtCompileTime == 1 || ColsAtCompileTime == 1 || (InnerPanel && (traits<XprType>::Flags&LinearAccessBit))) ? LinearAccessBit : 0,
FlagsLvalueBit = is_lvalue<XprType>::value ? LvalueBit : 0,
FlagsRowMajorBit = IsRowMajor ? RowMajorBit : 0,
Flags0 = traits<XprType>::Flags & ( (HereditaryBits & ~RowMajorBit) |

View File

@@ -250,6 +250,8 @@ template<typename Derived> class MapBase<Derived, WriteAccessors>
using Base::Base::operator=;
};
#undef EIGEN_STATIC_ASSERT_INDEX_BASED_ACCESS
} // end namespace Eigen
#endif // EIGEN_MAPBASE_H

View File

@@ -1,4 +1,5 @@
ADD_SUBDIRECTORY(SSE)
ADD_SUBDIRECTORY(AltiVec)
ADD_SUBDIRECTORY(NEON)
ADD_SUBDIRECTORY(AVX)
ADD_SUBDIRECTORY(Default)

View File

@@ -10,12 +10,6 @@
#ifndef EIGEN_GENERAL_BLOCK_PANEL_H
#define EIGEN_GENERAL_BLOCK_PANEL_H
#ifdef USE_IACA
#include "iacaMarks.h"
#else
#define IACA_START
#define IACA_END
#endif
namespace Eigen {
@@ -805,7 +799,6 @@ void gebp_kernel<LhsScalar,RhsScalar,Index,mr,nr,ConjugateLhs,ConjugateRhs>
blB += pk*4*RhsProgress;
blA += pk*3*Traits::LhsProgress;
IACA_END
}
// process remaining peeled loop
for(Index k=peeled_kc; k<depth; k++)

View File

@@ -263,7 +263,7 @@ EIGEN_DONT_INLINE void product_triangular_matrix_matrix<Scalar,Index,Mode,false,
Index mc = (std::min)(rows,blocking.mc()); // cache block size along the M direction
std::size_t sizeA = kc*mc;
std::size_t sizeB = kc*cols;
std::size_t sizeB = kc*cols+EIGEN_ALIGN_BYTES/sizeof(Scalar);
ei_declare_aligned_stack_constructed_variable(Scalar, blockA, sizeA, blocking.blockA());
ei_declare_aligned_stack_constructed_variable(Scalar, blockB, sizeB, blocking.blockB());

View File

@@ -348,13 +348,13 @@ namespace Eigen {
#elif defined(__clang__) // workaround clang bug (see http://forum.kde.org/viewtopic.php?f=74&t=102653)
#define EIGEN_INHERIT_ASSIGNMENT_EQUAL_OPERATOR(Derived) \
using Base::operator =; \
EIGEN_STRONG_INLINE Derived& operator=(const Derived& other) { Base::operator=(other); return *this; } \
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Derived& operator=(const Derived& other) { Base::operator=(other); return *this; } \
template <typename OtherDerived> \
EIGEN_STRONG_INLINE Derived& operator=(const DenseBase<OtherDerived>& other) { Base::operator=(other.derived()); return *this; }
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Derived& operator=(const DenseBase<OtherDerived>& other) { Base::operator=(other.derived()); return *this; }
#else
#define EIGEN_INHERIT_ASSIGNMENT_EQUAL_OPERATOR(Derived) \
using Base::operator =; \
EIGEN_STRONG_INLINE Derived& operator=(const Derived& other) \
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Derived& operator=(const Derived& other) \
{ \
Base::operator=(other); \
return *this; \

View File

@@ -767,9 +767,9 @@ namespace internal {
#ifdef EIGEN_CPUID
inline bool cpuid_is_vendor(int abcd[4], const char* vendor)
inline bool cpuid_is_vendor(int abcd[4], const int vendor[3])
{
return abcd[1]==(reinterpret_cast<const int*>(vendor))[0] && abcd[3]==(reinterpret_cast<const int*>(vendor))[1] && abcd[2]==(reinterpret_cast<const int*>(vendor))[2];
return abcd[1]==vendor[0] && abcd[3]==vendor[1] && abcd[2]==vendor[2];
}
inline void queryCacheSizes_intel_direct(int& l1, int& l2, int& l3)
@@ -911,13 +911,16 @@ inline void queryCacheSizes(int& l1, int& l2, int& l3)
{
#ifdef EIGEN_CPUID
int abcd[4];
const int GenuineIntel[] = {0x756e6547, 0x49656e69, 0x6c65746e};
const int AuthenticAMD[] = {0x68747541, 0x69746e65, 0x444d4163};
const int AMDisbetter_[] = {0x69444d41, 0x74656273, 0x21726574}; // "AMDisbetter!"
// identify the CPU vendor
EIGEN_CPUID(abcd,0x0,0);
int max_std_funcs = abcd[1];
if(cpuid_is_vendor(abcd,"GenuineIntel"))
if(cpuid_is_vendor(abcd,GenuineIntel))
queryCacheSizes_intel(l1,l2,l3,max_std_funcs);
else if(cpuid_is_vendor(abcd,"AuthenticAMD") || cpuid_is_vendor(abcd,"AMDisbetter!"))
else if(cpuid_is_vendor(abcd,AuthenticAMD) || cpuid_is_vendor(abcd,AMDisbetter_))
queryCacheSizes_amd(l1,l2,l3);
else
// by default let's use Intel's API

View File

@@ -587,7 +587,7 @@ inline Derived& QuaternionBase<Derived>::setFromTwoVectors(const MatrixBase<Deri
// which yields a singular value problem
if (c < Scalar(-1)+NumTraits<Scalar>::dummy_precision())
{
c = max<Scalar>(c,-1);
c = (max)(c,Scalar(-1));
Matrix<Scalar,2,3> m; m << v0.transpose(), v1.transpose();
JacobiSVD<Matrix<Scalar,2,3> > svd(m, ComputeFullV);
Vector3 axis = svd.matrixV().col(2);

View File

@@ -194,9 +194,9 @@ public:
/** type of the matrix used to represent the linear part of the transformation */
typedef Matrix<Scalar,Dim,Dim,Options> LinearMatrixType;
/** type of read/write reference to the linear part of the transformation */
typedef Block<MatrixType,Dim,Dim,int(Mode)==(AffineCompact)> LinearPart;
typedef Block<MatrixType,Dim,Dim,int(Mode)==(AffineCompact) && (Options&RowMajor)==0> LinearPart;
/** type of read reference to the linear part of the transformation */
typedef const Block<ConstMatrixType,Dim,Dim,int(Mode)==(AffineCompact)> ConstLinearPart;
typedef const Block<ConstMatrixType,Dim,Dim,int(Mode)==(AffineCompact) && (Options&RowMajor)==0> ConstLinearPart;
/** type of read/write reference to the affine part of the transformation */
typedef typename internal::conditional<int(Mode)==int(AffineCompact),
MatrixType&,

View File

@@ -113,7 +113,7 @@ umeyama(const MatrixBase<Derived>& src, const MatrixBase<OtherDerived>& dst, boo
const Index n = src.cols(); // number of measurements
// required for demeaning ...
const RealScalar one_over_n = 1 / static_cast<RealScalar>(n);
const RealScalar one_over_n = RealScalar(1) / static_cast<RealScalar>(n);
// computation of mean
const VectorType src_mean = src.rowwise().sum() * one_over_n;
@@ -136,16 +136,16 @@ umeyama(const MatrixBase<Derived>& src, const MatrixBase<OtherDerived>& dst, boo
// Eq. (39)
VectorType S = VectorType::Ones(m);
if (sigma.determinant()<0) S(m-1) = -1;
if (sigma.determinant()<Scalar(0)) S(m-1) = Scalar(-1);
// Eq. (40) and (43)
const VectorType& d = svd.singularValues();
Index rank = 0; for (Index i=0; i<m; ++i) if (!internal::isMuchSmallerThan(d.coeff(i),d.coeff(0))) ++rank;
if (rank == m-1) {
if ( svd.matrixU().determinant() * svd.matrixV().determinant() > 0 ) {
if ( svd.matrixU().determinant() * svd.matrixV().determinant() > Scalar(0) ) {
Rt.block(0,0,m,m).noalias() = svd.matrixU()*svd.matrixV().transpose();
} else {
const Scalar s = S(m-1); S(m-1) = -1;
const Scalar s = S(m-1); S(m-1) = Scalar(-1);
Rt.block(0,0,m,m).noalias() = svd.matrixU() * S.asDiagonal() * svd.matrixV().transpose();
S(m-1) = s;
}
@@ -156,7 +156,7 @@ umeyama(const MatrixBase<Derived>& src, const MatrixBase<OtherDerived>& dst, boo
if (with_scaling)
{
// Eq. (42)
const Scalar c = 1/src_var * svd.singularValues().dot(S);
const Scalar c = Scalar(1)/src_var * svd.singularValues().dot(S);
// Eq. (41)
Rt.col(m).head(m) = dst_mean;

View File

@@ -20,10 +20,11 @@ namespace Eigen {
*
* \param MatrixType the type of the matrix of which we are computing the LU decomposition
*
* This class represents a LU decomposition of any matrix, with complete pivoting: the matrix A
* is decomposed as A = PLUQ where L is unit-lower-triangular, U is upper-triangular, and P and Q
* are permutation matrices. This is a rank-revealing LU decomposition. The eigenvalues (diagonal
* coefficients) of U are sorted in such a way that any zeros are at the end.
* This class represents a LU decomposition of any matrix, with complete pivoting: the matrix A is
* decomposed as \f$ A = P^{-1} L U Q^{-1} \f$ where L is unit-lower-triangular, U is
* upper-triangular, and P and Q are permutation matrices. This is a rank-revealing LU
* decomposition. The eigenvalues (diagonal coefficients) of U are sorted in such a way that any
* zeros are at the end.
*
* This decomposition provides the generic approach to solving systems of linear equations, computing
* the rank, invertibility, inverse, kernel, and determinant.
@@ -511,8 +512,8 @@ typename internal::traits<MatrixType>::Scalar FullPivLU<MatrixType>::determinant
}
/** \returns the matrix represented by the decomposition,
* i.e., it returns the product: P^{-1} L U Q^{-1}.
* This function is provided for debug purpose. */
* i.e., it returns the product: \f$ P^{-1} L U Q^{-1} \f$.
* This function is provided for debug purposes. */
template<typename MatrixType>
MatrixType FullPivLU<MatrixType>::reconstructedMatrix() const
{

View File

@@ -1139,7 +1139,7 @@ EIGEN_DONT_INLINE typename SparseMatrix<_Scalar,_Options,_Index>::Scalar& Sparse
m_data.value(p) = m_data.value(p-1);
--p;
}
eigen_assert((p<=startId || m_data.index(p-1)!=inner) && "you cannot insert an element that already exist, you must call coeffRef to this end");
eigen_assert((p<=startId || m_data.index(p-1)!=inner) && "you cannot insert an element that already exists, you must call coeffRef to this end");
m_innerNonZeros[outer]++;

View File

@@ -11,7 +11,7 @@
#ifndef EIGEN_STDDEQUE_H
#define EIGEN_STDDEQUE_H
#include "Eigen/src/StlSupport/details.h"
#include "details.h"
// Define the explicit instantiation (e.g. necessary for the Intel compiler)
#if defined(__INTEL_COMPILER) || defined(__GNUC__)

View File

@@ -10,7 +10,7 @@
#ifndef EIGEN_STDLIST_H
#define EIGEN_STDLIST_H
#include "Eigen/src/StlSupport/details.h"
#include "details.h"
// Define the explicit instantiation (e.g. necessary for the Intel compiler)
#if defined(__INTEL_COMPILER) || defined(__GNUC__)

View File

@@ -11,7 +11,7 @@
#ifndef EIGEN_STDVECTOR_H
#define EIGEN_STDVECTOR_H
#include "Eigen/src/StlSupport/details.h"
#include "details.h"
/**
* This section contains a convenience MACRO which allows an easy specialization of