mirror of
https://gitlab.com/libeigen/eigen.git
synced 2026-04-10 11:34:33 +08:00
Pulled in the latest changes from the Eigen trunk
This commit is contained in:
@@ -144,24 +144,16 @@ class Array
|
||||
}
|
||||
#endif
|
||||
|
||||
/** Constructs a vector or row-vector with given dimension. \only_for_vectors
|
||||
*
|
||||
* Note that this is only useful for dynamic-size vectors. For fixed-size vectors,
|
||||
* it is redundant to pass the dimension here, so it makes more sense to use the default
|
||||
* constructor Matrix() instead.
|
||||
*/
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE explicit Array(Index dim)
|
||||
: Base(dim, RowsAtCompileTime == 1 ? 1 : dim, ColsAtCompileTime == 1 ? 1 : dim)
|
||||
{
|
||||
Base::_check_template_params();
|
||||
EIGEN_STATIC_ASSERT_VECTOR_ONLY(Array)
|
||||
eigen_assert(dim >= 0);
|
||||
eigen_assert(SizeAtCompileTime == Dynamic || SizeAtCompileTime == dim);
|
||||
EIGEN_INITIALIZE_COEFFS_IF_THAT_OPTION_IS_ENABLED
|
||||
}
|
||||
|
||||
#ifndef EIGEN_PARSED_BY_DOXYGEN
|
||||
template<typename T>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE explicit Array(const T& x)
|
||||
{
|
||||
Base::_check_template_params();
|
||||
Base::template _init1<T>(x);
|
||||
}
|
||||
|
||||
template<typename T0, typename T1>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE Array(const T0& val0, const T1& val1)
|
||||
@@ -170,11 +162,23 @@ class Array
|
||||
this->template _init2<T0,T1>(val0, val1);
|
||||
}
|
||||
#else
|
||||
/** constructs an uninitialized matrix with \a rows rows and \a cols columns.
|
||||
/** \brief Constructs a fixed-sized array initialized with coefficients starting at \a data */
|
||||
EIGEN_DEVICE_FUNC explicit Array(const Scalar *data);
|
||||
/** Constructs a vector or row-vector with given dimension. \only_for_vectors
|
||||
*
|
||||
* This is useful for dynamic-size matrices. For fixed-size matrices,
|
||||
* Note that this is only useful for dynamic-size vectors. For fixed-size vectors,
|
||||
* it is redundant to pass the dimension here, so it makes more sense to use the default
|
||||
* constructor Array() instead.
|
||||
*/
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE explicit Array(Index dim);
|
||||
/** constructs an initialized 1x1 Array with the given coefficient */
|
||||
Array(const Scalar& value);
|
||||
/** constructs an uninitialized array with \a rows rows and \a cols columns.
|
||||
*
|
||||
* This is useful for dynamic-size arrays. For fixed-size arrays,
|
||||
* it is redundant to pass these parameters, so one should use the default constructor
|
||||
* Matrix() instead. */
|
||||
* Array() instead. */
|
||||
Array(Index rows, Index cols);
|
||||
/** constructs an initialized 2D vector with given coefficients */
|
||||
Array(const Scalar& val0, const Scalar& val1);
|
||||
@@ -202,8 +206,6 @@ class Array
|
||||
m_storage.data()[3] = val3;
|
||||
}
|
||||
|
||||
EIGEN_DEVICE_FUNC explicit Array(const Scalar *data);
|
||||
|
||||
/** Constructor copying the value of the expression \a other */
|
||||
template<typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
|
||||
@@ -202,6 +202,7 @@ 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_LA(atan, Atan)
|
||||
//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)
|
||||
|
||||
@@ -506,17 +506,6 @@ template<typename Derived> class DenseBase
|
||||
# endif
|
||||
#undef EIGEN_CURRENT_STORAGE_BASE_CLASS
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
|
||||
Block<Derived> corner(CornerType type, Index cRows, Index cCols);
|
||||
const Block<Derived> corner(CornerType type, Index cRows, Index cCols) const;
|
||||
template<int CRows, int CCols>
|
||||
Block<Derived, CRows, CCols> corner(CornerType type);
|
||||
template<int CRows, int CCols>
|
||||
const Block<Derived, CRows, CCols> corner(CornerType type) const;
|
||||
|
||||
#endif // EIGEN2_SUPPORT
|
||||
|
||||
|
||||
// disable the use of evalTo for dense objects with a nice compilation error
|
||||
template<typename Dest>
|
||||
|
||||
@@ -156,10 +156,8 @@ class DenseCoeffsBase<Derived,ReadOnlyAccessors> : public EigenBase<Derived>
|
||||
EIGEN_STRONG_INLINE CoeffReturnType
|
||||
operator[](Index index) const
|
||||
{
|
||||
#ifndef EIGEN2_SUPPORT
|
||||
EIGEN_STATIC_ASSERT(Derived::IsVectorAtCompileTime,
|
||||
THE_BRACKET_OPERATOR_IS_ONLY_FOR_VECTORS__USE_THE_PARENTHESIS_OPERATOR_INSTEAD)
|
||||
#endif
|
||||
eigen_assert(index >= 0 && index < size());
|
||||
return derived().coeff(index);
|
||||
}
|
||||
@@ -388,10 +386,8 @@ class DenseCoeffsBase<Derived, WriteAccessors> : public DenseCoeffsBase<Derived,
|
||||
EIGEN_STRONG_INLINE Scalar&
|
||||
operator[](Index index)
|
||||
{
|
||||
#ifndef EIGEN2_SUPPORT
|
||||
EIGEN_STATIC_ASSERT(Derived::IsVectorAtCompileTime,
|
||||
THE_BRACKET_OPERATOR_IS_ONLY_FOR_VECTORS__USE_THE_PARENTHESIS_OPERATOR_INSTEAD)
|
||||
#endif
|
||||
eigen_assert(index >= 0 && index < size());
|
||||
return derived().coeffRef(index);
|
||||
}
|
||||
|
||||
@@ -95,21 +95,6 @@ class DiagonalBase : public EigenBase<Derived>
|
||||
{
|
||||
return other.diagonal() * scalar;
|
||||
}
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
bool isApprox(const DiagonalBase<OtherDerived>& other, typename NumTraits<Scalar>::Real precision = NumTraits<Scalar>::dummy_precision()) const
|
||||
{
|
||||
return diagonal().isApprox(other.diagonal(), precision);
|
||||
}
|
||||
template<typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
bool isApprox(const MatrixBase<OtherDerived>& other, typename NumTraits<Scalar>::Real precision = NumTraits<Scalar>::dummy_precision()) const
|
||||
{
|
||||
return toDenseMatrix().isApprox(other, precision);
|
||||
}
|
||||
#endif
|
||||
};
|
||||
|
||||
template<typename Derived>
|
||||
|
||||
@@ -76,34 +76,6 @@ MatrixBase<Derived>::dot(const MatrixBase<OtherDerived>& other) const
|
||||
return internal::dot_nocheck<Derived,OtherDerived>::run(*this, other);
|
||||
}
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
/** \returns the dot product of *this with other, with the Eigen2 convention that the dot product is linear in the first variable
|
||||
* (conjugating the second variable). Of course this only makes a difference in the complex case.
|
||||
*
|
||||
* This method is only available in EIGEN2_SUPPORT mode.
|
||||
*
|
||||
* \only_for_vectors
|
||||
*
|
||||
* \sa dot()
|
||||
*/
|
||||
template<typename Derived>
|
||||
template<typename OtherDerived>
|
||||
typename internal::traits<Derived>::Scalar
|
||||
MatrixBase<Derived>::eigen2_dot(const MatrixBase<OtherDerived>& other) const
|
||||
{
|
||||
EIGEN_STATIC_ASSERT_VECTOR_ONLY(Derived)
|
||||
EIGEN_STATIC_ASSERT_VECTOR_ONLY(OtherDerived)
|
||||
EIGEN_STATIC_ASSERT_SAME_VECTOR_SIZE(Derived,OtherDerived)
|
||||
EIGEN_STATIC_ASSERT((internal::is_same<Scalar, typename OtherDerived::Scalar>::value),
|
||||
YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY)
|
||||
|
||||
eigen_assert(size() == other.size());
|
||||
|
||||
return internal::dot_nocheck<OtherDerived,Derived>::run(other,*this);
|
||||
}
|
||||
#endif
|
||||
|
||||
|
||||
//---------- implementation of L2 norm and related functions ----------
|
||||
|
||||
/** \returns, for vectors, the squared \em l2 norm of \c *this, and for matrices the Frobenius norm.
|
||||
|
||||
@@ -445,7 +445,7 @@ template<> struct gemv_selector<OnTheRight,ColMajor,true>
|
||||
if(!evalToDest)
|
||||
{
|
||||
#ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
int size = dest.size();
|
||||
Index size = dest.size();
|
||||
EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
#endif
|
||||
if(!alphaIsCompatible)
|
||||
@@ -510,7 +510,7 @@ template<> struct gemv_selector<OnTheRight,RowMajor,true>
|
||||
if(!DirectlyUseRhs)
|
||||
{
|
||||
#ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
int size = actualRhs.size();
|
||||
Index size = actualRhs.size();
|
||||
EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
#endif
|
||||
Map<typename _ActualRhsType::PlainObject>(actualRhsPtr, actualRhs.size()) = actualRhs;
|
||||
|
||||
8
Eigen/src/Core/GenericPacketMath.h
Executable file → Normal file
8
Eigen/src/Core/GenericPacketMath.h
Executable file → Normal file
@@ -234,10 +234,10 @@ template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline void pstore(
|
||||
template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline void pstoreu(Scalar* to, const Packet& from)
|
||||
{ (*to) = from; }
|
||||
|
||||
template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline Packet pgather(const Scalar* from, int /*stride*/)
|
||||
template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline Packet pgather(const Scalar* from, DenseIndex /*stride*/)
|
||||
{ return ploadu<Packet>(from); }
|
||||
|
||||
template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline void pscatter(Scalar* to, const Packet& from, int /*stride*/)
|
||||
template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline void pscatter(Scalar* to, const Packet& from, DenseIndex /*stride*/)
|
||||
{ pstore(to, from); }
|
||||
|
||||
/** \internal tries to do cache prefetching of \a addr */
|
||||
@@ -319,6 +319,10 @@ Packet pasin(const Packet& a) { using std::asin; return asin(a); }
|
||||
template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
|
||||
Packet pacos(const Packet& a) { using std::acos; return acos(a); }
|
||||
|
||||
/** \internal \returns the atan of \a a (coeff-wise) */
|
||||
template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
|
||||
Packet patan(const Packet& a) { using std::atan; return atan(a); }
|
||||
|
||||
/** \internal \returns the exp of \a a (coeff-wise) */
|
||||
template<typename Packet> EIGEN_DECLARE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS
|
||||
Packet pexp(const Packet& a) { using std::exp; return exp(a); }
|
||||
|
||||
@@ -45,6 +45,7 @@ namespace Eigen
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(asin,scalar_asin_op)
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(acos,scalar_acos_op)
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(tan,scalar_tan_op)
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(atan,scalar_atan_op)
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(exp,scalar_exp_op)
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(log,scalar_log_op)
|
||||
EIGEN_ARRAY_DECLARE_GLOBAL_UNARY(abs,scalar_abs_op)
|
||||
|
||||
@@ -110,14 +110,9 @@ template<typename PlainObjectType, int MapOptions, typename StrideType> class Ma
|
||||
EIGEN_DENSE_PUBLIC_INTERFACE(Map)
|
||||
|
||||
typedef typename Base::PointerType PointerType;
|
||||
#if EIGEN2_SUPPORT_STAGE <= STAGE30_FULL_EIGEN3_API
|
||||
typedef const Scalar* PointerArgType;
|
||||
inline PointerType cast_to_pointer_type(PointerArgType ptr) { return const_cast<PointerType>(ptr); }
|
||||
#else
|
||||
typedef PointerType PointerArgType;
|
||||
EIGEN_DEVICE_FUNC
|
||||
inline PointerType cast_to_pointer_type(PointerArgType ptr) { return ptr; }
|
||||
#endif
|
||||
|
||||
EIGEN_DEVICE_FUNC
|
||||
inline Index innerStride() const
|
||||
@@ -179,19 +174,6 @@ template<typename PlainObjectType, int MapOptions, typename StrideType> class Ma
|
||||
StrideType m_stride;
|
||||
};
|
||||
|
||||
template<typename _Scalar, int _Rows, int _Cols, int _Options, int _MaxRows, int _MaxCols>
|
||||
inline Array<_Scalar, _Rows, _Cols, _Options, _MaxRows, _MaxCols>
|
||||
::Array(const Scalar *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)
|
||||
{
|
||||
this->_set_noalias(Eigen::Map<const Matrix>(data));
|
||||
}
|
||||
|
||||
} // end namespace Eigen
|
||||
|
||||
|
||||
@@ -232,24 +232,17 @@ class Matrix
|
||||
}
|
||||
#endif
|
||||
|
||||
/** \brief Constructs a vector or row-vector with given dimension. \only_for_vectors
|
||||
*
|
||||
* Note that this is only useful for dynamic-size vectors. For fixed-size vectors,
|
||||
* it is redundant to pass the dimension here, so it makes more sense to use the default
|
||||
* constructor Matrix() instead.
|
||||
*/
|
||||
#ifndef EIGEN_PARSED_BY_DOXYGEN
|
||||
|
||||
// This constructor is for both 1x1 matrices and dynamic vectors
|
||||
template<typename T>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE explicit Matrix(Index dim)
|
||||
: Base(dim, RowsAtCompileTime == 1 ? 1 : dim, ColsAtCompileTime == 1 ? 1 : dim)
|
||||
EIGEN_STRONG_INLINE explicit Matrix(const T& x)
|
||||
{
|
||||
Base::_check_template_params();
|
||||
EIGEN_STATIC_ASSERT_VECTOR_ONLY(Matrix)
|
||||
eigen_assert(dim >= 0);
|
||||
eigen_assert(SizeAtCompileTime == Dynamic || SizeAtCompileTime == dim);
|
||||
EIGEN_INITIALIZE_COEFFS_IF_THAT_OPTION_IS_ENABLED
|
||||
Base::template _init1<T>(x);
|
||||
}
|
||||
|
||||
#ifndef EIGEN_PARSED_BY_DOXYGEN
|
||||
template<typename T0, typename T1>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE Matrix(const T0& x, const T1& y)
|
||||
@@ -258,13 +251,40 @@ class Matrix
|
||||
Base::template _init2<T0,T1>(x, y);
|
||||
}
|
||||
#else
|
||||
/** \brief Constructs a fixed-sized matrix initialized with coefficients starting at \a data */
|
||||
EIGEN_DEVICE_FUNC
|
||||
explicit Matrix(const Scalar *data);
|
||||
|
||||
/** \brief Constructs a vector or row-vector with given dimension. \only_for_vectors
|
||||
*
|
||||
* This is useful for dynamic-size vectors. For fixed-size vectors,
|
||||
* it is redundant to pass these parameters, so one should use the default constructor
|
||||
* Matrix() instead.
|
||||
*
|
||||
* \warning This constructor is disabled for fixed-size \c 1x1 matrices. For instance,
|
||||
* calling Matrix<double,1,1>(1) will call the initialization constructor: Matrix(const Scalar&).
|
||||
* For fixed-size \c 1x1 matrices it is thefore recommended to use the default
|
||||
* constructor Matrix() instead, especilly when using one of the non standard
|
||||
* \c EIGEN_INITIALIZE_MATRICES_BY_{ZERO,\c NAN} macros (see \ref TopicPreprocessorDirectives).
|
||||
*/
|
||||
EIGEN_STRONG_INLINE explicit Matrix(Index dim);
|
||||
/** \brief Constructs an initialized 1x1 matrix with the given coefficient */
|
||||
Matrix(const Scalar& x);
|
||||
/** \brief Constructs an uninitialized matrix with \a rows rows and \a cols columns.
|
||||
*
|
||||
* This is useful for dynamic-size matrices. For fixed-size matrices,
|
||||
* it is redundant to pass these parameters, so one should use the default constructor
|
||||
* Matrix() instead. */
|
||||
* Matrix() instead.
|
||||
*
|
||||
* \warning This constructor is disabled for fixed-size \c 1x2 and \c 2x1 vectors. For instance,
|
||||
* calling Matrix2f(2,1) will call the initialization constructor: Matrix(const Scalar& x, const Scalar& y).
|
||||
* For fixed-size \c 1x2 or \c 2x1 vectors it is thefore recommended to use the default
|
||||
* constructor Matrix() instead, especilly when using one of the non standard
|
||||
* \c EIGEN_INITIALIZE_MATRICES_BY_{ZERO,\c NAN} macros (see \ref TopicPreprocessorDirectives).
|
||||
*/
|
||||
EIGEN_DEVICE_FUNC
|
||||
Matrix(Index rows, Index cols);
|
||||
|
||||
/** \brief Constructs an initialized 2D vector with given coefficients */
|
||||
Matrix(const Scalar& x, const Scalar& y);
|
||||
#endif
|
||||
@@ -291,8 +311,6 @@ class Matrix
|
||||
m_storage.data()[3] = w;
|
||||
}
|
||||
|
||||
EIGEN_DEVICE_FUNC
|
||||
explicit Matrix(const Scalar *data);
|
||||
|
||||
/** \brief Constructor copying the value of the expression \a other */
|
||||
template<typename OtherDerived>
|
||||
@@ -362,13 +380,6 @@ class Matrix
|
||||
EIGEN_DEVICE_FUNC
|
||||
Matrix& operator=(const RotationBase<OtherDerived,ColsAtCompileTime>& r);
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename OtherDerived>
|
||||
explicit Matrix(const eigen2_RotationBase<OtherDerived,ColsAtCompileTime>& r);
|
||||
template<typename OtherDerived>
|
||||
Matrix& operator=(const eigen2_RotationBase<OtherDerived,ColsAtCompileTime>& r);
|
||||
#endif
|
||||
|
||||
// allow to extend Matrix outside Eigen
|
||||
#ifdef EIGEN_MATRIX_PLUGIN
|
||||
#include EIGEN_MATRIX_PLUGIN
|
||||
|
||||
@@ -221,11 +221,6 @@ template<typename Derived> class MatrixBase
|
||||
typename internal::scalar_product_traits<typename internal::traits<Derived>::Scalar,typename internal::traits<OtherDerived>::Scalar>::ReturnType
|
||||
dot(const MatrixBase<OtherDerived>& other) const;
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename OtherDerived>
|
||||
Scalar eigen2_dot(const MatrixBase<OtherDerived>& other) const;
|
||||
#endif
|
||||
|
||||
EIGEN_DEVICE_FUNC RealScalar squaredNorm() const;
|
||||
EIGEN_DEVICE_FUNC RealScalar norm() const;
|
||||
RealScalar stableNorm() const;
|
||||
@@ -269,17 +264,6 @@ template<typename Derived> class MatrixBase
|
||||
typename ConstDiagonalIndexReturnType<DynamicIndex>::Type diagonal(Index index) const;
|
||||
#endif
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<unsigned int Mode> typename internal::eigen2_part_return_type<Derived, Mode>::type part();
|
||||
template<unsigned int Mode> const typename internal::eigen2_part_return_type<Derived, Mode>::type part() const;
|
||||
|
||||
// huuuge hack. make Eigen2's matrix.part<Diagonal>() work in eigen3. Problem: Diagonal is now a class template instead
|
||||
// of an integer constant. Solution: overload the part() method template wrt template parameters list.
|
||||
template<template<typename T, int N> class U>
|
||||
const DiagonalWrapper<ConstDiagonalReturnType> part() const
|
||||
{ return diagonal().asDiagonal(); }
|
||||
#endif // EIGEN2_SUPPORT
|
||||
|
||||
template<unsigned int Mode> struct TriangularViewReturnType { typedef TriangularView<Derived, Mode> Type; };
|
||||
template<unsigned int Mode> struct ConstTriangularViewReturnType { typedef const TriangularView<const Derived, Mode> Type; };
|
||||
|
||||
@@ -373,24 +357,7 @@ template<typename Derived> class MatrixBase
|
||||
EIGEN_DEVICE_FUNC const FullPivLU<PlainObject> fullPivLu() const;
|
||||
EIGEN_DEVICE_FUNC const PartialPivLU<PlainObject> partialPivLu() const;
|
||||
|
||||
#if EIGEN2_SUPPORT_STAGE < STAGE20_RESOLVE_API_CONFLICTS
|
||||
const LU<PlainObject> lu() const;
|
||||
#endif
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
const LU<PlainObject> eigen2_lu() const;
|
||||
#endif
|
||||
|
||||
#if EIGEN2_SUPPORT_STAGE > STAGE20_RESOLVE_API_CONFLICTS
|
||||
const PartialPivLU<PlainObject> lu() const;
|
||||
#endif
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename ResultType>
|
||||
void computeInverse(MatrixBase<ResultType> *result) const {
|
||||
*result = this->inverse();
|
||||
}
|
||||
#endif
|
||||
|
||||
EIGEN_DEVICE_FUNC
|
||||
const internal::inverse_impl<Derived> inverse() const;
|
||||
@@ -419,10 +386,6 @@ template<typename Derived> class MatrixBase
|
||||
const HouseholderQR<PlainObject> householderQr() const;
|
||||
const ColPivHouseholderQR<PlainObject> colPivHouseholderQr() const;
|
||||
const FullPivHouseholderQR<PlainObject> fullPivHouseholderQr() const;
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
const QR<PlainObject> qr() const;
|
||||
#endif
|
||||
|
||||
EigenvaluesReturnType eigenvalues() const;
|
||||
RealScalar operatorNorm() const;
|
||||
@@ -431,10 +394,6 @@ template<typename Derived> class MatrixBase
|
||||
|
||||
JacobiSVD<PlainObject> jacobiSvd(unsigned int computationOptions = 0) const;
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
SVD<PlainObject> svd() const;
|
||||
#endif
|
||||
|
||||
/////////// Geometry module ///////////
|
||||
|
||||
#ifndef EIGEN_PARSED_BY_DOXYGEN
|
||||
@@ -458,13 +417,11 @@ template<typename Derived> class MatrixBase
|
||||
|
||||
Matrix<Scalar,3,1> eulerAngles(Index a0, Index a1, Index a2) const;
|
||||
|
||||
#if EIGEN2_SUPPORT_STAGE > STAGE20_RESOLVE_API_CONFLICTS
|
||||
ScalarMultipleReturnType operator*(const UniformScaling<Scalar>& s) const;
|
||||
// put this as separate enum value to work around possible GCC 4.3 bug (?)
|
||||
enum { HomogeneousReturnTypeDirection = ColsAtCompileTime==1?Vertical:Horizontal };
|
||||
typedef Homogeneous<Derived, HomogeneousReturnTypeDirection> HomogeneousReturnType;
|
||||
HomogeneousReturnType homogeneous() const;
|
||||
#endif
|
||||
|
||||
enum {
|
||||
SizeMinusOne = SizeAtCompileTime==Dynamic ? Dynamic : SizeAtCompileTime-1
|
||||
@@ -513,41 +470,6 @@ template<typename Derived> class MatrixBase
|
||||
const MatrixPowerReturnValue<Derived> pow(const RealScalar& p) const;
|
||||
const MatrixComplexPowerReturnValue<Derived> pow(const std::complex<RealScalar>& p) const;
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename ProductDerived, typename Lhs, typename Rhs>
|
||||
Derived& operator+=(const Flagged<ProductBase<ProductDerived, Lhs,Rhs>, 0,
|
||||
EvalBeforeAssigningBit>& other);
|
||||
|
||||
template<typename ProductDerived, typename Lhs, typename Rhs>
|
||||
Derived& operator-=(const Flagged<ProductBase<ProductDerived, Lhs,Rhs>, 0,
|
||||
EvalBeforeAssigningBit>& other);
|
||||
|
||||
/** \deprecated because .lazy() is deprecated
|
||||
* Overloaded for cache friendly product evaluation */
|
||||
template<typename OtherDerived>
|
||||
Derived& lazyAssign(const Flagged<OtherDerived, 0, EvalBeforeAssigningBit>& other)
|
||||
{ return lazyAssign(other._expression()); }
|
||||
|
||||
template<unsigned int Added>
|
||||
const Flagged<Derived, Added, 0> marked() const;
|
||||
const Flagged<Derived, 0, EvalBeforeAssigningBit> lazy() const;
|
||||
|
||||
inline const Cwise<Derived> cwise() const;
|
||||
inline Cwise<Derived> cwise();
|
||||
|
||||
VectorBlock<Derived> start(Index size);
|
||||
const VectorBlock<const Derived> start(Index size) const;
|
||||
VectorBlock<Derived> end(Index size);
|
||||
const VectorBlock<const Derived> end(Index size) const;
|
||||
template<int Size> VectorBlock<Derived,Size> start();
|
||||
template<int Size> const VectorBlock<const Derived,Size> start() const;
|
||||
template<int Size> VectorBlock<Derived,Size> end();
|
||||
template<int Size> const VectorBlock<const Derived,Size> end() const;
|
||||
|
||||
Minor<Derived> minor(Index row, Index col);
|
||||
const Minor<Derived> minor(Index row, Index col) const;
|
||||
#endif
|
||||
|
||||
protected:
|
||||
EIGEN_DEVICE_FUNC MatrixBase() : Base() {}
|
||||
|
||||
|
||||
@@ -85,13 +85,6 @@ template<typename T> struct GenericNumTraits
|
||||
}
|
||||
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 {
|
||||
HasFloatingPoint = !IsInteger
|
||||
};
|
||||
typedef NonInteger FloatingPoint;
|
||||
#endif
|
||||
};
|
||||
|
||||
template<typename T> struct NumTraits : GenericNumTraits<T>
|
||||
|
||||
@@ -66,11 +66,11 @@ class PermutationBase : public EigenBase<Derived>
|
||||
MaxRowsAtCompileTime = Traits::MaxRowsAtCompileTime,
|
||||
MaxColsAtCompileTime = Traits::MaxColsAtCompileTime
|
||||
};
|
||||
typedef typename Traits::Scalar Scalar;
|
||||
typedef typename Traits::StorageIndexType StorageIndexType;
|
||||
typedef typename Traits::Index Index;
|
||||
typedef Matrix<Scalar,RowsAtCompileTime,ColsAtCompileTime,0,MaxRowsAtCompileTime,MaxColsAtCompileTime>
|
||||
typedef Matrix<StorageIndexType,RowsAtCompileTime,ColsAtCompileTime,0,MaxRowsAtCompileTime,MaxColsAtCompileTime>
|
||||
DenseMatrixType;
|
||||
typedef PermutationMatrix<IndicesType::SizeAtCompileTime,IndicesType::MaxSizeAtCompileTime,Index>
|
||||
typedef PermutationMatrix<IndicesType::SizeAtCompileTime,IndicesType::MaxSizeAtCompileTime,StorageIndexType>
|
||||
PlainPermutationType;
|
||||
using Base::derived;
|
||||
#endif
|
||||
@@ -147,7 +147,7 @@ class PermutationBase : public EigenBase<Derived>
|
||||
/** Sets *this to be the identity permutation matrix */
|
||||
void setIdentity()
|
||||
{
|
||||
for(Index i = 0; i < size(); ++i)
|
||||
for(StorageIndexType i = 0; i < size(); ++i)
|
||||
indices().coeffRef(i) = i;
|
||||
}
|
||||
|
||||
@@ -173,8 +173,8 @@ class PermutationBase : public EigenBase<Derived>
|
||||
eigen_assert(i>=0 && j>=0 && i<size() && j<size());
|
||||
for(Index k = 0; k < size(); ++k)
|
||||
{
|
||||
if(indices().coeff(k) == i) indices().coeffRef(k) = j;
|
||||
else if(indices().coeff(k) == j) indices().coeffRef(k) = i;
|
||||
if(indices().coeff(k) == i) indices().coeffRef(k) = StorageIndexType(j);
|
||||
else if(indices().coeff(k) == j) indices().coeffRef(k) = StorageIndexType(i);
|
||||
}
|
||||
return derived();
|
||||
}
|
||||
@@ -262,7 +262,7 @@ class PermutationBase : public EigenBase<Derived>
|
||||
*
|
||||
* \param SizeAtCompileTime the number of rows/cols, or Dynamic
|
||||
* \param MaxSizeAtCompileTime the maximum number of rows/cols, or Dynamic. This optional parameter defaults to SizeAtCompileTime. Most of the time, you should not have to specify it.
|
||||
* \param IndexType the interger type of the indices
|
||||
* \param StorageIndexType the integer type of the indices
|
||||
*
|
||||
* This class represents a permutation matrix, internally stored as a vector of integers.
|
||||
*
|
||||
@@ -270,17 +270,18 @@ class PermutationBase : public EigenBase<Derived>
|
||||
*/
|
||||
|
||||
namespace internal {
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType>
|
||||
struct traits<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, IndexType> >
|
||||
: traits<Matrix<IndexType,SizeAtCompileTime,SizeAtCompileTime,0,MaxSizeAtCompileTime,MaxSizeAtCompileTime> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType>
|
||||
struct traits<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, _StorageIndexType> >
|
||||
: traits<Matrix<_StorageIndexType,SizeAtCompileTime,SizeAtCompileTime,0,MaxSizeAtCompileTime,MaxSizeAtCompileTime> >
|
||||
{
|
||||
typedef IndexType Index;
|
||||
typedef Matrix<IndexType, SizeAtCompileTime, 1, 0, MaxSizeAtCompileTime, 1> IndicesType;
|
||||
typedef Matrix<_StorageIndexType, SizeAtCompileTime, 1, 0, MaxSizeAtCompileTime, 1> IndicesType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
typedef _StorageIndexType StorageIndexType;
|
||||
};
|
||||
}
|
||||
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType>
|
||||
class PermutationMatrix : public PermutationBase<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, IndexType> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType>
|
||||
class PermutationMatrix : public PermutationBase<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, _StorageIndexType> >
|
||||
{
|
||||
typedef PermutationBase<PermutationMatrix> Base;
|
||||
typedef internal::traits<PermutationMatrix> Traits;
|
||||
@@ -288,6 +289,8 @@ class PermutationMatrix : public PermutationBase<PermutationMatrix<SizeAtCompile
|
||||
|
||||
#ifndef EIGEN_PARSED_BY_DOXYGEN
|
||||
typedef typename Traits::IndicesType IndicesType;
|
||||
typedef typename Traits::StorageIndexType StorageIndexType;
|
||||
typedef typename Traits::Index Index;
|
||||
#endif
|
||||
|
||||
inline PermutationMatrix()
|
||||
@@ -295,7 +298,7 @@ class PermutationMatrix : public PermutationBase<PermutationMatrix<SizeAtCompile
|
||||
|
||||
/** Constructs an uninitialized permutation matrix of given size.
|
||||
*/
|
||||
inline PermutationMatrix(int size) : m_indices(size)
|
||||
inline PermutationMatrix(Index size) : m_indices(size)
|
||||
{}
|
||||
|
||||
/** Copy constructor. */
|
||||
@@ -384,18 +387,19 @@ class PermutationMatrix : public PermutationBase<PermutationMatrix<SizeAtCompile
|
||||
|
||||
|
||||
namespace internal {
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType, int _PacketAccess>
|
||||
struct traits<Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, IndexType>,_PacketAccess> >
|
||||
: traits<Matrix<IndexType,SizeAtCompileTime,SizeAtCompileTime,0,MaxSizeAtCompileTime,MaxSizeAtCompileTime> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType, int _PacketAccess>
|
||||
struct traits<Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, _StorageIndexType>,_PacketAccess> >
|
||||
: traits<Matrix<_StorageIndexType,SizeAtCompileTime,SizeAtCompileTime,0,MaxSizeAtCompileTime,MaxSizeAtCompileTime> >
|
||||
{
|
||||
typedef IndexType Index;
|
||||
typedef Map<const Matrix<IndexType, SizeAtCompileTime, 1, 0, MaxSizeAtCompileTime, 1>, _PacketAccess> IndicesType;
|
||||
typedef Map<const Matrix<_StorageIndexType, SizeAtCompileTime, 1, 0, MaxSizeAtCompileTime, 1>, _PacketAccess> IndicesType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
typedef _StorageIndexType StorageIndexType;
|
||||
};
|
||||
}
|
||||
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType, int _PacketAccess>
|
||||
class Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, IndexType>,_PacketAccess>
|
||||
: public PermutationBase<Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, IndexType>,_PacketAccess> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType, int _PacketAccess>
|
||||
class Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, _StorageIndexType>,_PacketAccess>
|
||||
: public PermutationBase<Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, _StorageIndexType>,_PacketAccess> >
|
||||
{
|
||||
typedef PermutationBase<Map> Base;
|
||||
typedef internal::traits<Map> Traits;
|
||||
@@ -403,14 +407,15 @@ class Map<PermutationMatrix<SizeAtCompileTime, MaxSizeAtCompileTime, IndexType>,
|
||||
|
||||
#ifndef EIGEN_PARSED_BY_DOXYGEN
|
||||
typedef typename Traits::IndicesType IndicesType;
|
||||
typedef typename IndicesType::Scalar Index;
|
||||
typedef typename IndicesType::Scalar StorageIndexType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
#endif
|
||||
|
||||
inline Map(const Index* indicesPtr)
|
||||
inline Map(const StorageIndexType* indicesPtr)
|
||||
: m_indices(indicesPtr)
|
||||
{}
|
||||
|
||||
inline Map(const Index* indicesPtr, Index size)
|
||||
inline Map(const StorageIndexType* indicesPtr, Index size)
|
||||
: m_indices(indicesPtr,size)
|
||||
{}
|
||||
|
||||
@@ -466,7 +471,8 @@ struct traits<PermutationWrapper<_IndicesType> >
|
||||
{
|
||||
typedef PermutationStorage StorageKind;
|
||||
typedef typename _IndicesType::Scalar Scalar;
|
||||
typedef typename _IndicesType::Scalar Index;
|
||||
typedef typename _IndicesType::Scalar StorageIndexType;
|
||||
typedef typename _IndicesType::Index Index;
|
||||
typedef _IndicesType IndicesType;
|
||||
enum {
|
||||
RowsAtCompileTime = _IndicesType::SizeAtCompileTime,
|
||||
|
||||
@@ -681,6 +681,7 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
|
||||
FLOATING_POINT_ARGUMENT_PASSED__INTEGER_WAS_EXPECTED)
|
||||
resize(nbRows,nbCols);
|
||||
}
|
||||
|
||||
template<typename T0, typename T1>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init2(const Scalar& val0, const Scalar& val1, typename internal::enable_if<Base::SizeAtCompileTime==2,T0>::type* = 0)
|
||||
@@ -689,6 +690,82 @@ class PlainObjectBase : public internal::dense_xpr_base<Derived>::type
|
||||
m_storage.data()[0] = val0;
|
||||
m_storage.data()[1] = val1;
|
||||
}
|
||||
|
||||
template<typename T0, typename T1>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init2(const Index& val0, const Index& val1,
|
||||
typename internal::enable_if< (!internal::is_same<Index,Scalar>::value)
|
||||
&& (internal::is_same<T0,Index>::value)
|
||||
&& (internal::is_same<T1,Index>::value)
|
||||
&& Base::SizeAtCompileTime==2,T1>::type* = 0)
|
||||
{
|
||||
EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(PlainObjectBase, 2)
|
||||
m_storage.data()[0] = Scalar(val0);
|
||||
m_storage.data()[1] = Scalar(val1);
|
||||
}
|
||||
|
||||
template<typename T>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(Index size, typename internal::enable_if<Base::SizeAtCompileTime!=1 || !internal::is_convertible<T, Scalar>::value,T>::type* = 0)
|
||||
{
|
||||
// NOTE MSVC 2008 complains if we directly put bool(NumTraits<T>::IsInteger) as the EIGEN_STATIC_ASSERT argument.
|
||||
const bool is_integer = NumTraits<T>::IsInteger;
|
||||
EIGEN_STATIC_ASSERT(is_integer,
|
||||
FLOATING_POINT_ARGUMENT_PASSED__INTEGER_WAS_EXPECTED)
|
||||
resize(size);
|
||||
}
|
||||
template<typename T>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const Scalar& val0, typename internal::enable_if<Base::SizeAtCompileTime==1 && internal::is_convertible<T, Scalar>::value,T>::type* = 0)
|
||||
{
|
||||
EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(PlainObjectBase, 1)
|
||||
m_storage.data()[0] = val0;
|
||||
}
|
||||
|
||||
template<typename T>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const Index& val0,
|
||||
typename internal::enable_if< (!internal::is_same<Index,Scalar>::value)
|
||||
&& (internal::is_same<Index,T>::value)
|
||||
&& Base::SizeAtCompileTime==1
|
||||
&& internal::is_convertible<T, Scalar>::value,T*>::type* = 0)
|
||||
{
|
||||
EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(PlainObjectBase, 1)
|
||||
m_storage.data()[0] = Scalar(val0);
|
||||
}
|
||||
|
||||
template<typename T>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const Scalar* data){
|
||||
this->_set_noalias(ConstMapType(data));
|
||||
}
|
||||
|
||||
template<typename T, typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const DenseBase<OtherDerived>& other){
|
||||
this->_set_noalias(other);
|
||||
}
|
||||
|
||||
template<typename T, typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const EigenBase<OtherDerived>& other){
|
||||
this->derived() = other;
|
||||
}
|
||||
|
||||
template<typename T, typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const ReturnByValue<OtherDerived>& other)
|
||||
{
|
||||
resize(other.rows(), other.cols());
|
||||
other.evalTo(this->derived());
|
||||
}
|
||||
|
||||
template<typename T, typename OtherDerived, int ColsAtCompileTime>
|
||||
EIGEN_DEVICE_FUNC
|
||||
EIGEN_STRONG_INLINE void _init1(const RotationBase<OtherDerived,ColsAtCompileTime>& r)
|
||||
{
|
||||
this->derived() = r;
|
||||
}
|
||||
|
||||
template<typename MatrixTypeA, typename MatrixTypeB, bool SwapPointers>
|
||||
friend struct internal::matrix_swap_impl;
|
||||
|
||||
@@ -131,17 +131,13 @@ class ProductBase : public MatrixBase<Derived>
|
||||
const Diagonal<FullyLazyCoeffBaseProductType,Dynamic> diagonal(Index index) const
|
||||
{ return FullyLazyCoeffBaseProductType(m_lhs, m_rhs).diagonal(index); }
|
||||
|
||||
// restrict coeff accessors to 1x1 expressions. No need to care about mutators here since this isnt a Lvalue expression
|
||||
// restrict coeff accessors to 1x1 expressions. No need to care about mutators here since this isn't an Lvalue expression
|
||||
typename Base::CoeffReturnType coeff(Index row, Index col) const
|
||||
{
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
return lhs().row(row).cwiseProduct(rhs().col(col).transpose()).sum();
|
||||
#else
|
||||
EIGEN_STATIC_ASSERT_SIZE_1x1(Derived)
|
||||
eigen_assert(this->rows() == 1 && this->cols() == 1);
|
||||
Matrix<Scalar,1,1> result = *this;
|
||||
return result.coeff(row,col);
|
||||
#endif
|
||||
}
|
||||
|
||||
typename Base::CoeffReturnType coeff(Index i) const
|
||||
|
||||
@@ -19,17 +19,17 @@ template<typename PlainObjectType, int Options = 0,
|
||||
/** \class Ref
|
||||
* \ingroup Core_Module
|
||||
*
|
||||
* \brief A matrix or vector expression mapping an existing expressions
|
||||
* \brief A matrix or vector expression mapping an existing expression
|
||||
*
|
||||
* \tparam PlainObjectType the equivalent matrix type of the mapped data
|
||||
* \tparam Options specifies whether the pointer is \c #Aligned, or \c #Unaligned.
|
||||
* The default is \c #Unaligned.
|
||||
* \tparam StrideType optionally specifies strides. By default, Ref implies a contiguous storage along the inner dimension (inner stride==1),
|
||||
* but accept a variable outer stride (leading dimension).
|
||||
* but accepts a variable outer stride (leading dimension).
|
||||
* This can be overridden by specifying strides.
|
||||
* The type passed here must be a specialization of the Stride template, see examples below.
|
||||
*
|
||||
* This class permits to write non template functions taking Eigen's object as parameters while limiting the number of copies.
|
||||
* This class provides a way to write non-template functions taking Eigen objects as parameters while limiting the number of copies.
|
||||
* A Ref<> object can represent either a const expression or a l-value:
|
||||
* \code
|
||||
* // in-out argument:
|
||||
@@ -39,10 +39,10 @@ template<typename PlainObjectType, int Options = 0,
|
||||
* void foo2(const Ref<const VectorXf>& x);
|
||||
* \endcode
|
||||
*
|
||||
* In the in-out case, the input argument must satisfies the constraints of the actual Ref<> type, otherwise a compilation issue will be triggered.
|
||||
* In the in-out case, the input argument must satisfy the constraints of the actual Ref<> type, otherwise a compilation issue will be triggered.
|
||||
* By default, a Ref<VectorXf> can reference any dense vector expression of float having a contiguous memory layout.
|
||||
* Likewise, a Ref<MatrixXf> can reference any column major dense matrix expression of float whose column's elements are contiguously stored with
|
||||
* the possibility to have a constant space inbetween each column, i.e.: the inner stride mmust be equal to 1, but the outer-stride (or leading dimension),
|
||||
* Likewise, a Ref<MatrixXf> can reference any column-major dense matrix expression of float whose column's elements are contiguously stored with
|
||||
* the possibility to have a constant space in-between each column, i.e. the inner stride must be equal to 1, but the outer stride (or leading dimension)
|
||||
* can be greater than the number of rows.
|
||||
*
|
||||
* In the const case, if the input expression does not match the above requirement, then it is evaluated into a temporary before being passed to the function.
|
||||
@@ -58,15 +58,15 @@ template<typename PlainObjectType, int Options = 0,
|
||||
* foo2(A.col().segment(2,4)); // No temporary
|
||||
* \endcode
|
||||
*
|
||||
* The range of inputs that can be referenced without temporary can be enlarged using the last two template parameter.
|
||||
* The range of inputs that can be referenced without temporary can be enlarged using the last two template parameters.
|
||||
* Here is an example accepting an innerstride!=1:
|
||||
* \code
|
||||
* // in-out argument:
|
||||
* void foo3(Ref<VectorXf,0,InnerStride<> > x);
|
||||
* foo3(A.row()); // OK
|
||||
* \endcode
|
||||
* The downside here is that the function foo3 might be significantly slower than foo1 because it won't be able to exploit vectorization, and will involved more
|
||||
* expensive address computations even if the input is contiguously stored in memory. To overcome this issue, one might propose to overloads internally calling a
|
||||
* The downside here is that the function foo3 might be significantly slower than foo1 because it won't be able to exploit vectorization, and will involve more
|
||||
* expensive address computations even if the input is contiguously stored in memory. To overcome this issue, one might propose to overload internally calling a
|
||||
* template function, e.g.:
|
||||
* \code
|
||||
* // in the .h:
|
||||
|
||||
@@ -177,31 +177,6 @@ template<typename MatrixType, unsigned int UpLo> class SelfAdjointView
|
||||
EigenvaluesReturnType eigenvalues() const;
|
||||
EIGEN_DEVICE_FUNC
|
||||
RealScalar operatorNorm() const;
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
SelfAdjointView& operator=(const MatrixBase<OtherDerived>& other)
|
||||
{
|
||||
enum {
|
||||
OtherPart = UpLo == Upper ? StrictlyLower : StrictlyUpper
|
||||
};
|
||||
m_matrix.const_cast_derived().template triangularView<UpLo>() = other;
|
||||
m_matrix.const_cast_derived().template triangularView<OtherPart>() = other.adjoint();
|
||||
return *this;
|
||||
}
|
||||
template<typename OtherMatrixType, unsigned int OtherMode>
|
||||
EIGEN_DEVICE_FUNC
|
||||
SelfAdjointView& operator=(const TriangularView<OtherMatrixType, OtherMode>& other)
|
||||
{
|
||||
enum {
|
||||
OtherPart = UpLo == Upper ? StrictlyLower : StrictlyUpper
|
||||
};
|
||||
m_matrix.const_cast_derived().template triangularView<UpLo>() = other.toDenseMatrix();
|
||||
m_matrix.const_cast_derived().template triangularView<OtherPart>() = other.toDenseMatrix().adjoint();
|
||||
return *this;
|
||||
}
|
||||
#endif
|
||||
|
||||
protected:
|
||||
MatrixTypeNested m_matrix;
|
||||
|
||||
@@ -53,7 +53,8 @@ class TranspositionsBase
|
||||
public:
|
||||
|
||||
typedef typename Traits::IndicesType IndicesType;
|
||||
typedef typename IndicesType::Scalar Index;
|
||||
typedef typename IndicesType::Scalar StorageIndexType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
|
||||
Derived& derived() { return *static_cast<Derived*>(this); }
|
||||
const Derived& derived() const { return *static_cast<const Derived*>(this); }
|
||||
@@ -81,17 +82,17 @@ class TranspositionsBase
|
||||
inline Index size() const { return indices().size(); }
|
||||
|
||||
/** Direct access to the underlying index vector */
|
||||
inline const Index& coeff(Index i) const { return indices().coeff(i); }
|
||||
inline const StorageIndexType& coeff(Index i) const { return indices().coeff(i); }
|
||||
/** Direct access to the underlying index vector */
|
||||
inline Index& coeffRef(Index i) { return indices().coeffRef(i); }
|
||||
inline StorageIndexType& coeffRef(Index i) { return indices().coeffRef(i); }
|
||||
/** Direct access to the underlying index vector */
|
||||
inline const Index& operator()(Index i) const { return indices()(i); }
|
||||
inline const StorageIndexType& operator()(Index i) const { return indices()(i); }
|
||||
/** Direct access to the underlying index vector */
|
||||
inline Index& operator()(Index i) { return indices()(i); }
|
||||
inline StorageIndexType& operator()(Index i) { return indices()(i); }
|
||||
/** Direct access to the underlying index vector */
|
||||
inline const Index& operator[](Index i) const { return indices()(i); }
|
||||
inline const StorageIndexType& operator[](Index i) const { return indices()(i); }
|
||||
/** Direct access to the underlying index vector */
|
||||
inline Index& operator[](Index i) { return indices()(i); }
|
||||
inline StorageIndexType& operator[](Index i) { return indices()(i); }
|
||||
|
||||
/** const version of indices(). */
|
||||
const IndicesType& indices() const { return derived().indices(); }
|
||||
@@ -99,7 +100,7 @@ class TranspositionsBase
|
||||
IndicesType& indices() { return derived().indices(); }
|
||||
|
||||
/** Resizes to given size. */
|
||||
inline void resize(int newSize)
|
||||
inline void resize(Index newSize)
|
||||
{
|
||||
indices().resize(newSize);
|
||||
}
|
||||
@@ -107,7 +108,7 @@ class TranspositionsBase
|
||||
/** Sets \c *this to represents an identity transformation */
|
||||
void setIdentity()
|
||||
{
|
||||
for(int i = 0; i < indices().size(); ++i)
|
||||
for(StorageIndexType i = 0; i < indices().size(); ++i)
|
||||
coeffRef(i) = i;
|
||||
}
|
||||
|
||||
@@ -144,23 +145,26 @@ class TranspositionsBase
|
||||
};
|
||||
|
||||
namespace internal {
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType>
|
||||
struct traits<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,IndexType> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType>
|
||||
struct traits<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,_StorageIndexType> >
|
||||
{
|
||||
typedef IndexType Index;
|
||||
typedef Matrix<Index, SizeAtCompileTime, 1, 0, MaxSizeAtCompileTime, 1> IndicesType;
|
||||
typedef Matrix<_StorageIndexType, SizeAtCompileTime, 1, 0, MaxSizeAtCompileTime, 1> IndicesType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
typedef _StorageIndexType StorageIndexType;
|
||||
};
|
||||
}
|
||||
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType>
|
||||
class Transpositions : public TranspositionsBase<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,IndexType> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType>
|
||||
class Transpositions : public TranspositionsBase<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,_StorageIndexType> >
|
||||
{
|
||||
typedef internal::traits<Transpositions> Traits;
|
||||
public:
|
||||
|
||||
typedef TranspositionsBase<Transpositions> Base;
|
||||
typedef typename Traits::IndicesType IndicesType;
|
||||
typedef typename IndicesType::Scalar Index;
|
||||
typedef typename IndicesType::Scalar StorageIndexType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
|
||||
|
||||
inline Transpositions() {}
|
||||
|
||||
@@ -215,30 +219,32 @@ class Transpositions : public TranspositionsBase<Transpositions<SizeAtCompileTim
|
||||
|
||||
|
||||
namespace internal {
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType, int _PacketAccess>
|
||||
struct traits<Map<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,IndexType>,_PacketAccess> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType, int _PacketAccess>
|
||||
struct traits<Map<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,_StorageIndexType>,_PacketAccess> >
|
||||
{
|
||||
typedef IndexType Index;
|
||||
typedef Map<const Matrix<Index,SizeAtCompileTime,1,0,MaxSizeAtCompileTime,1>, _PacketAccess> IndicesType;
|
||||
typedef Map<const Matrix<_StorageIndexType,SizeAtCompileTime,1,0,MaxSizeAtCompileTime,1>, _PacketAccess> IndicesType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
typedef _StorageIndexType StorageIndexType;
|
||||
};
|
||||
}
|
||||
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename IndexType, int PacketAccess>
|
||||
class Map<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,IndexType>,PacketAccess>
|
||||
: public TranspositionsBase<Map<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,IndexType>,PacketAccess> >
|
||||
template<int SizeAtCompileTime, int MaxSizeAtCompileTime, typename _StorageIndexType, int PacketAccess>
|
||||
class Map<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,_StorageIndexType>,PacketAccess>
|
||||
: public TranspositionsBase<Map<Transpositions<SizeAtCompileTime,MaxSizeAtCompileTime,_StorageIndexType>,PacketAccess> >
|
||||
{
|
||||
typedef internal::traits<Map> Traits;
|
||||
public:
|
||||
|
||||
typedef TranspositionsBase<Map> Base;
|
||||
typedef typename Traits::IndicesType IndicesType;
|
||||
typedef typename IndicesType::Scalar Index;
|
||||
typedef typename IndicesType::Scalar StorageIndexType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
|
||||
inline Map(const Index* indicesPtr)
|
||||
inline Map(const StorageIndexType* indicesPtr)
|
||||
: m_indices(indicesPtr)
|
||||
{}
|
||||
|
||||
inline Map(const Index* indicesPtr, Index size)
|
||||
inline Map(const StorageIndexType* indicesPtr, Index size)
|
||||
: m_indices(indicesPtr,size)
|
||||
{}
|
||||
|
||||
@@ -275,7 +281,8 @@ namespace internal {
|
||||
template<typename _IndicesType>
|
||||
struct traits<TranspositionsWrapper<_IndicesType> >
|
||||
{
|
||||
typedef typename _IndicesType::Scalar Index;
|
||||
typedef typename _IndicesType::Scalar StorageIndexType;
|
||||
typedef typename _IndicesType::Index Index;
|
||||
typedef _IndicesType IndicesType;
|
||||
};
|
||||
}
|
||||
@@ -289,7 +296,8 @@ class TranspositionsWrapper
|
||||
|
||||
typedef TranspositionsBase<TranspositionsWrapper> Base;
|
||||
typedef typename Traits::IndicesType IndicesType;
|
||||
typedef typename IndicesType::Scalar Index;
|
||||
typedef typename IndicesType::Scalar StorageIndexType;
|
||||
typedef typename IndicesType::Index Index;
|
||||
|
||||
inline TranspositionsWrapper(IndicesType& a_indices)
|
||||
: m_indices(a_indices)
|
||||
@@ -363,24 +371,25 @@ struct transposition_matrix_product_retval
|
||||
{
|
||||
typedef typename remove_all<typename MatrixType::Nested>::type MatrixTypeNestedCleaned;
|
||||
typedef typename TranspositionType::Index Index;
|
||||
typedef typename TranspositionType::StorageIndexType StorageIndexType;
|
||||
|
||||
transposition_matrix_product_retval(const TranspositionType& tr, const MatrixType& matrix)
|
||||
: m_transpositions(tr), m_matrix(matrix)
|
||||
{}
|
||||
|
||||
inline int rows() const { return m_matrix.rows(); }
|
||||
inline int cols() const { return m_matrix.cols(); }
|
||||
inline Index rows() const { return m_matrix.rows(); }
|
||||
inline Index cols() const { return m_matrix.cols(); }
|
||||
|
||||
template<typename Dest> inline void evalTo(Dest& dst) const
|
||||
{
|
||||
const int size = m_transpositions.size();
|
||||
Index j = 0;
|
||||
const Index size = m_transpositions.size();
|
||||
StorageIndexType j = 0;
|
||||
|
||||
if(!(is_same<MatrixTypeNestedCleaned,Dest>::value && extract_data(dst) == extract_data(m_matrix)))
|
||||
dst = m_matrix;
|
||||
|
||||
for(int k=(Transposed?size-1:0) ; Transposed?k>=0:k<size ; Transposed?--k:++k)
|
||||
if((j=m_transpositions.coeff(k))!=k)
|
||||
for(Index k=(Transposed?size-1:0) ; Transposed?k>=0:k<size ; Transposed?--k:++k)
|
||||
if(Index(j=m_transpositions.coeff(k))!=k)
|
||||
{
|
||||
if(Side==OnTheLeft)
|
||||
dst.row(k).swap(dst.row(j));
|
||||
|
||||
@@ -323,54 +323,25 @@ template<typename _MatrixType, unsigned int _Mode> class TriangularView
|
||||
/** Efficient triangular matrix times vector/matrix product */
|
||||
template<typename OtherDerived>
|
||||
EIGEN_DEVICE_FUNC
|
||||
TriangularProduct<Mode,true,MatrixType,false,OtherDerived, OtherDerived::IsVectorAtCompileTime>
|
||||
TriangularProduct<Mode, true, MatrixType, false, OtherDerived, OtherDerived::ColsAtCompileTime==1>
|
||||
operator*(const MatrixBase<OtherDerived>& rhs) const
|
||||
{
|
||||
return TriangularProduct
|
||||
<Mode,true,MatrixType,false,OtherDerived,OtherDerived::IsVectorAtCompileTime>
|
||||
<Mode, true, MatrixType, false, OtherDerived, OtherDerived::ColsAtCompileTime==1>
|
||||
(m_matrix, rhs.derived());
|
||||
}
|
||||
|
||||
/** Efficient vector/matrix times triangular matrix product */
|
||||
template<typename OtherDerived> friend
|
||||
EIGEN_DEVICE_FUNC
|
||||
TriangularProduct<Mode,false,OtherDerived,OtherDerived::IsVectorAtCompileTime,MatrixType,false>
|
||||
TriangularProduct<Mode, false, OtherDerived, OtherDerived::RowsAtCompileTime==1, MatrixType, false>
|
||||
operator*(const MatrixBase<OtherDerived>& lhs, const TriangularView& rhs)
|
||||
{
|
||||
return TriangularProduct
|
||||
<Mode,false,OtherDerived,OtherDerived::IsVectorAtCompileTime,MatrixType,false>
|
||||
<Mode, false, OtherDerived, OtherDerived::RowsAtCompileTime==1, MatrixType, false>
|
||||
(lhs.derived(),rhs.m_matrix);
|
||||
}
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename OtherDerived>
|
||||
struct eigen2_product_return_type
|
||||
{
|
||||
typedef typename TriangularView<MatrixType,Mode>::DenseMatrixType DenseMatrixType;
|
||||
typedef typename OtherDerived::PlainObject::DenseType OtherPlainObject;
|
||||
typedef typename ProductReturnType<DenseMatrixType, OtherPlainObject>::Type ProdRetType;
|
||||
typedef typename ProdRetType::PlainObject type;
|
||||
};
|
||||
template<typename OtherDerived>
|
||||
const typename eigen2_product_return_type<OtherDerived>::type
|
||||
operator*(const EigenBase<OtherDerived>& rhs) const
|
||||
{
|
||||
typename OtherDerived::PlainObject::DenseType rhsPlainObject;
|
||||
rhs.evalTo(rhsPlainObject);
|
||||
return this->toDenseMatrix() * rhsPlainObject;
|
||||
}
|
||||
template<typename OtherMatrixType>
|
||||
bool isApprox(const TriangularView<OtherMatrixType, Mode>& other, typename NumTraits<Scalar>::Real precision = NumTraits<Scalar>::dummy_precision()) const
|
||||
{
|
||||
return this->toDenseMatrix().isApprox(other.toDenseMatrix(), precision);
|
||||
}
|
||||
template<typename OtherDerived>
|
||||
bool isApprox(const MatrixBase<OtherDerived>& other, typename NumTraits<Scalar>::Real precision = NumTraits<Scalar>::dummy_precision()) const
|
||||
{
|
||||
return this->toDenseMatrix().isApprox(other, precision);
|
||||
}
|
||||
#endif // EIGEN2_SUPPORT
|
||||
|
||||
template<int Side, typename Other>
|
||||
EIGEN_DEVICE_FUNC
|
||||
inline const internal::triangular_solve_retval<Side,TriangularView, Other>
|
||||
@@ -780,41 +751,6 @@ void TriangularBase<Derived>::evalToLazy(MatrixBase<DenseDerived> &other) const
|
||||
* Implementation of MatrixBase methods
|
||||
***************************************************************************/
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
|
||||
// implementation of part<>(), including the SelfAdjoint case.
|
||||
|
||||
namespace internal {
|
||||
template<typename MatrixType, unsigned int Mode>
|
||||
struct eigen2_part_return_type
|
||||
{
|
||||
typedef TriangularView<MatrixType, Mode> type;
|
||||
};
|
||||
|
||||
template<typename MatrixType>
|
||||
struct eigen2_part_return_type<MatrixType, SelfAdjoint>
|
||||
{
|
||||
typedef SelfAdjointView<MatrixType, Upper> type;
|
||||
};
|
||||
}
|
||||
|
||||
/** \deprecated use MatrixBase::triangularView() */
|
||||
template<typename Derived>
|
||||
template<unsigned int Mode>
|
||||
const typename internal::eigen2_part_return_type<Derived, Mode>::type MatrixBase<Derived>::part() const
|
||||
{
|
||||
return derived();
|
||||
}
|
||||
|
||||
/** \deprecated use MatrixBase::triangularView() */
|
||||
template<typename Derived>
|
||||
template<unsigned int Mode>
|
||||
typename internal::eigen2_part_return_type<Derived, Mode>::type MatrixBase<Derived>::part()
|
||||
{
|
||||
return derived();
|
||||
}
|
||||
#endif
|
||||
|
||||
/**
|
||||
* \returns an expression of a triangular view extracted from the current matrix
|
||||
*
|
||||
|
||||
@@ -560,9 +560,7 @@ template<typename ExpressionType, int Direction> class VectorwiseOp
|
||||
|
||||
/////////// Geometry module ///////////
|
||||
|
||||
#if EIGEN2_SUPPORT_STAGE > STAGE20_RESOLVE_API_CONFLICTS
|
||||
Homogeneous<ExpressionType,Direction> homogeneous() const;
|
||||
#endif
|
||||
|
||||
typedef typename ExpressionType::PlainObject CrossReturnType;
|
||||
template<typename OtherDerived>
|
||||
|
||||
@@ -194,7 +194,7 @@ DenseBase<Derived>::minCoeff(IndexType* index) const
|
||||
EIGEN_STATIC_ASSERT_VECTOR_ONLY(Derived)
|
||||
internal::min_coeff_visitor<Derived> minVisitor;
|
||||
this->visit(minVisitor);
|
||||
*index = (RowsAtCompileTime==1) ? minVisitor.col : minVisitor.row;
|
||||
*index = IndexType((RowsAtCompileTime==1) ? minVisitor.col : minVisitor.row);
|
||||
return minVisitor.res;
|
||||
}
|
||||
|
||||
|
||||
@@ -92,7 +92,7 @@ template<> EIGEN_STRONG_INLINE Packet4cf ploaddup<Packet4cf>(const std::complex<
|
||||
template<> EIGEN_STRONG_INLINE void pstore <std::complex<float> >(std::complex<float>* to, const Packet4cf& from) { EIGEN_DEBUG_ALIGNED_STORE pstore(&numext::real_ref(*to), from.v); }
|
||||
template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<float> >(std::complex<float>* to, const Packet4cf& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu(&numext::real_ref(*to), from.v); }
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4cf pgather<std::complex<float>, Packet4cf>(const std::complex<float>* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4cf pgather<std::complex<float>, Packet4cf>(const std::complex<float>* from, DenseIndex stride)
|
||||
{
|
||||
return Packet4cf(_mm256_set_ps(std::imag(from[3*stride]), std::real(from[3*stride]),
|
||||
std::imag(from[2*stride]), std::real(from[2*stride]),
|
||||
@@ -100,7 +100,7 @@ template<> EIGEN_DEVICE_FUNC inline Packet4cf pgather<std::complex<float>, Packe
|
||||
std::imag(from[0*stride]), std::real(from[0*stride])));
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet4cf>(std::complex<float>* to, const Packet4cf& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet4cf>(std::complex<float>* to, const Packet4cf& from, DenseIndex stride)
|
||||
{
|
||||
__m128 low = _mm256_extractf128_ps(from.v, 0);
|
||||
to[stride*0] = std::complex<float>(_mm_cvtss_f32(_mm_shuffle_ps(low, low, 0)),
|
||||
@@ -310,13 +310,13 @@ template<> EIGEN_STRONG_INLINE Packet2cd ploaddup<Packet2cd>(const std::complex<
|
||||
template<> EIGEN_STRONG_INLINE void pstore <std::complex<double> >(std::complex<double> * to, const Packet2cd& from) { EIGEN_DEBUG_ALIGNED_STORE pstore((double*)to, from.v); }
|
||||
template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<double> >(std::complex<double> * to, const Packet2cd& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu((double*)to, from.v); }
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cd pgather<std::complex<double>, Packet2cd>(const std::complex<double>* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cd pgather<std::complex<double>, Packet2cd>(const std::complex<double>* from, DenseIndex stride)
|
||||
{
|
||||
return Packet2cd(_mm256_set_pd(std::imag(from[1*stride]), std::real(from[1*stride]),
|
||||
std::imag(from[0*stride]), std::real(from[0*stride])));
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<double>, Packet2cd>(std::complex<double>* to, const Packet2cd& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<double>, Packet2cd>(std::complex<double>* to, const Packet2cd& from, DenseIndex stride)
|
||||
{
|
||||
__m128d low = _mm256_extractf128_pd(from.v, 0);
|
||||
to[stride*0] = std::complex<double>(_mm_cvtsd_f64(low), _mm_cvtsd_f64(_mm_shuffle_pd(low, low, 1)));
|
||||
|
||||
@@ -226,17 +226,17 @@ template<> EIGEN_STRONG_INLINE void pstoreu<int>(int* to, const Packet8i&
|
||||
|
||||
// NOTE: leverage _mm256_i32gather_ps and _mm256_i32gather_pd if AVX2 instructions are available
|
||||
// NOTE: for the record the following seems to be slower: return _mm256_i32gather_ps(from, _mm256_set1_epi32(stride), 4);
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet8f pgather<float, Packet8f>(const float* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet8f pgather<float, Packet8f>(const float* from, DenseIndex stride)
|
||||
{
|
||||
return _mm256_set_ps(from[7*stride], from[6*stride], from[5*stride], from[4*stride],
|
||||
from[3*stride], from[2*stride], from[1*stride], from[0*stride]);
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4d pgather<double, Packet4d>(const double* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4d pgather<double, Packet4d>(const double* from, DenseIndex stride)
|
||||
{
|
||||
return _mm256_set_pd(from[3*stride], from[2*stride], from[1*stride], from[0*stride]);
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet8f>(float* to, const Packet8f& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet8f>(float* to, const Packet8f& from, DenseIndex stride)
|
||||
{
|
||||
__m128 low = _mm256_extractf128_ps(from, 0);
|
||||
to[stride*0] = _mm_cvtss_f32(low);
|
||||
@@ -250,7 +250,7 @@ template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet8f>(float* to, co
|
||||
to[stride*6] = _mm_cvtss_f32(_mm_shuffle_ps(high, high, 2));
|
||||
to[stride*7] = _mm_cvtss_f32(_mm_shuffle_ps(high, high, 3));
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<double, Packet4d>(double* to, const Packet4d& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<double, Packet4d>(double* to, const Packet4d& from, DenseIndex stride)
|
||||
{
|
||||
__m128d low = _mm256_extractf128_pd(from, 0);
|
||||
to[stride*0] = _mm_cvtsd_f64(low);
|
||||
|
||||
@@ -16,13 +16,14 @@ namespace internal {
|
||||
|
||||
static Packet4ui p4ui_CONJ_XOR = vec_mergeh((Packet4ui)p4i_ZERO, (Packet4ui)p4f_ZERO_);//{ 0x00000000, 0x80000000, 0x00000000, 0x80000000 };
|
||||
static Packet16uc p16uc_COMPLEX_RE = vec_sld((Packet16uc) vec_splat((Packet4ui)p16uc_FORWARD, 0), (Packet16uc) vec_splat((Packet4ui)p16uc_FORWARD, 2), 8);//{ 0,1,2,3, 0,1,2,3, 8,9,10,11, 8,9,10,11 };
|
||||
static Packet16uc p16uc_COMPLEX_IM = vec_sld((Packet16uc) vec_splat((Packet4ui)p16uc_FORWARD, 1), (Packet16uc) vec_splat((Packet4ui)p16uc_FORWARD, 3), 8);//{ 4,5,6,7, 4,5,6,7, 12,13,14,15, 12,13,14,15 };
|
||||
static Packet16uc p16uc_COMPLEX_IM = vec_sld(p16uc_DUPLICATE, (Packet16uc) vec_splat((Packet4ui)p16uc_FORWARD, 3), 8);//{ 4,5,6,7, 4,5,6,7, 12,13,14,15, 12,13,14,15 };
|
||||
static Packet16uc p16uc_COMPLEX_REV = vec_sld(p16uc_REVERSE, p16uc_REVERSE, 8);//{ 4,5,6,7, 0,1,2,3, 12,13,14,15, 8,9,10,11 };
|
||||
static Packet16uc p16uc_COMPLEX_REV2 = vec_sld(p16uc_FORWARD, p16uc_FORWARD, 8);//{ 8,9,10,11, 12,13,14,15, 0,1,2,3, 4,5,6,7 };
|
||||
static Packet16uc p16uc_PSET_HI = (Packet16uc) vec_mergeh((Packet4ui) vec_splat((Packet4ui)p16uc_FORWARD, 0), (Packet4ui) vec_splat((Packet4ui)p16uc_FORWARD, 1));//{ 0,1,2,3, 4,5,6,7, 0,1,2,3, 4,5,6,7 };
|
||||
static Packet16uc p16uc_PSET_LO = (Packet16uc) vec_mergeh((Packet4ui) vec_splat((Packet4ui)p16uc_FORWARD, 2), (Packet4ui) vec_splat((Packet4ui)p16uc_FORWARD, 3));//{ 8,9,10,11, 12,13,14,15, 8,9,10,11, 12,13,14,15 };
|
||||
static Packet16uc p16uc_COMPLEX_TRANSPOSE_0 = { 0,1,2,3, 4,5,6,7, 16,17,18,19, 20,21,22,23};
|
||||
static Packet16uc p16uc_COMPLEX_TRANSPOSE_1 = { 8,9,10,11, 12,13,14,15, 24,25,26,27, 28,29,30,31};
|
||||
static Packet16uc p16uc_PSET_HI = (Packet16uc) vec_mergeh((Packet4ui)p16uc_COMPLEX_RE, (Packet4ui)p16uc_COMPLEX_IM);//{ 0,1,2,3, 4,5,6,7, 0,1,2,3, 4,5,6,7 };
|
||||
static Packet16uc p16uc_PSET_LO = (Packet16uc) vec_mergel((Packet4ui)p16uc_COMPLEX_RE, (Packet4ui)p16uc_COMPLEX_IM);//{ 8,9,10,11, 12,13,14,15, 8,9,10,11, 12,13,14,15 };
|
||||
static Packet16uc p16uc_COMPLEX_MASK16 = vec_sld((Packet16uc)p4i_ZERO, vec_splat((Packet16uc) vec_abs(p4i_MINUS16), 3), 8);//{ 0,0,0,0, 0,0,0,0, 16,16,16,16, 16,16,16,16};
|
||||
static Packet16uc p16uc_COMPLEX_TRANSPOSE_0 = vec_add(p16uc_PSET_HI, p16uc_COMPLEX_MASK16);//{ 0,1,2,3, 4,5,6,7, 16,17,18,19, 20,21,22,23};
|
||||
static Packet16uc p16uc_COMPLEX_TRANSPOSE_1 = vec_add(p16uc_PSET_LO, p16uc_COMPLEX_MASK16);//{ 8,9,10,11, 12,13,14,15, 24,25,26,27, 28,29,30,31};
|
||||
|
||||
//---------- float ----------
|
||||
struct Packet2cf
|
||||
@@ -68,14 +69,14 @@ template<> EIGEN_STRONG_INLINE Packet2cf pset1<Packet2cf>(const std::complex<flo
|
||||
return res;
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packet2cf>(const std::complex<float>* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packet2cf>(const std::complex<float>* from, DenseIndex stride)
|
||||
{
|
||||
std::complex<float> EIGEN_ALIGN16 af[2];
|
||||
af[0] = from[0*stride];
|
||||
af[1] = from[1*stride];
|
||||
return Packet2cf(vec_ld(0, (const float*)af));
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet2cf>(std::complex<float>* to, const Packet2cf& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet2cf>(std::complex<float>* to, const Packet2cf& from, DenseIndex stride)
|
||||
{
|
||||
std::complex<float> EIGEN_ALIGN16 af[2];
|
||||
vec_st(from.v, 0, (float*)af);
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
// This file is part of Eigen, a lightweight C++ template library
|
||||
// for linear algebra.
|
||||
//
|
||||
// Copyright (C) 2008 Konstantinos Margaritis <markos@codex.gr>
|
||||
// Copyright (C) 2008-2014 Konstantinos Margaritis <markos@freevec.org>
|
||||
//
|
||||
// This Source Code Form is subject to the terms of the Mozilla
|
||||
// Public License v. 2.0. If a copy of the MPL was not distributed
|
||||
@@ -62,17 +62,17 @@ typedef __vector unsigned char Packet16uc;
|
||||
// Define global static constants:
|
||||
static Packet4f p4f_COUNTDOWN = { 0.0, 1.0, 2.0, 3.0 };
|
||||
static Packet4i p4i_COUNTDOWN = { 0, 1, 2, 3 };
|
||||
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 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); //{ 0,1,2,3, 4,5,6,7, 8,9,10,11, 12,13,14,15}
|
||||
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);
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(ONE,1);
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(MINUS16,-16);
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(MINUS1,-1);
|
||||
static Packet4f p4f_ONE = vec_ctf(p4i_ONE, 0);
|
||||
static Packet4f p4f_ZERO_ = (Packet4f) vec_sl((Packet4ui)p4i_MINUS1, (Packet4ui)p4i_MINUS1);
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4f(ZERO, 0); //{ 0.0, 0.0, 0.0, 0.0}
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(ZERO, 0); //{ 0, 0, 0, 0,}
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(ONE,1); //{ 1, 1, 1, 1}
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(MINUS16,-16); //{ -16, -16, -16, -16}
|
||||
static _EIGEN_DECLARE_CONST_FAST_Packet4i(MINUS1,-1); //{ -1, -1, -1, -1}
|
||||
static Packet4f p4f_ONE = vec_ctf(p4i_ONE, 0); //{ 1.0, 1.0, 1.0, 1.0}
|
||||
static Packet4f p4f_ZERO_ = (Packet4f) vec_sl((Packet4ui)p4i_MINUS1, (Packet4ui)p4i_MINUS1); //{ 0x80000000, 0x80000000, 0x80000000, 0x80000000}
|
||||
|
||||
template<> struct packet_traits<float> : default_packet_traits
|
||||
{
|
||||
@@ -82,8 +82,10 @@ template<> struct packet_traits<float> : default_packet_traits
|
||||
Vectorizable = 1,
|
||||
AlignedOnScalar = 1,
|
||||
size=4,
|
||||
HasHalfPacket=0,
|
||||
|
||||
// FIXME check the Has*
|
||||
HasDiv = 1,
|
||||
HasSin = 0,
|
||||
HasCos = 0,
|
||||
HasLog = 0,
|
||||
@@ -190,7 +192,7 @@ pbroadcast4<Packet4i>(const int *a,
|
||||
a3 = vec_splat(a3, 3);
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const float* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const float* from, DenseIndex stride)
|
||||
{
|
||||
float EIGEN_ALIGN16 af[4];
|
||||
af[0] = from[0*stride];
|
||||
@@ -199,7 +201,7 @@ template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const floa
|
||||
af[3] = from[3*stride];
|
||||
return vec_ld(0, af);
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* from, DenseIndex stride)
|
||||
{
|
||||
int EIGEN_ALIGN16 ai[4];
|
||||
ai[0] = from[0*stride];
|
||||
@@ -208,7 +210,7 @@ template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* f
|
||||
ai[3] = from[3*stride];
|
||||
return vec_ld(0, ai);
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, const Packet4f& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, const Packet4f& from, DenseIndex stride)
|
||||
{
|
||||
float EIGEN_ALIGN16 af[4];
|
||||
vec_st(from, 0, af);
|
||||
@@ -217,7 +219,7 @@ template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, co
|
||||
to[2*stride] = af[2];
|
||||
to[3*stride] = af[3];
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<int, Packet4i>(int* to, const Packet4i& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<int, Packet4i>(int* to, const Packet4i& from, DenseIndex stride)
|
||||
{
|
||||
int EIGEN_ALIGN16 ai[4];
|
||||
vec_st(from, 0, ai);
|
||||
|
||||
@@ -111,7 +111,7 @@ template<> EIGEN_STRONG_INLINE Packet2cf ploaddup<Packet2cf>(const std::complex<
|
||||
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); }
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packet2cf>(const std::complex<float>* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packet2cf>(const std::complex<float>* from, DenseIndex stride)
|
||||
{
|
||||
Packet4f res;
|
||||
res = vsetq_lane_f32(std::real(from[0*stride]), res, 0);
|
||||
@@ -121,7 +121,7 @@ template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packe
|
||||
return Packet2cf(res);
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet2cf>(std::complex<float>* to, const Packet2cf& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet2cf>(std::complex<float>* to, const Packet2cf& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = std::complex<float>(vgetq_lane_f32(from.v, 0), vgetq_lane_f32(from.v, 1));
|
||||
to[stride*1] = std::complex<float>(vgetq_lane_f32(from.v, 2), vgetq_lane_f32(from.v, 3));
|
||||
|
||||
@@ -222,7 +222,7 @@ template<> EIGEN_STRONG_INLINE void pstore<int>(int* to, const Packet4i& f
|
||||
template<> EIGEN_STRONG_INLINE void pstoreu<float>(float* to, const Packet4f& from) { EIGEN_DEBUG_UNALIGNED_STORE vst1q_f32(to, from); }
|
||||
template<> EIGEN_STRONG_INLINE void pstoreu<int>(int* to, const Packet4i& from) { EIGEN_DEBUG_UNALIGNED_STORE vst1q_s32(to, from); }
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const float* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const float* from, DenseIndex stride)
|
||||
{
|
||||
Packet4f res;
|
||||
res = vsetq_lane_f32(from[0*stride], res, 0);
|
||||
@@ -231,7 +231,7 @@ template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const floa
|
||||
res = vsetq_lane_f32(from[3*stride], res, 3);
|
||||
return res;
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* from, DenseIndex stride)
|
||||
{
|
||||
Packet4i res;
|
||||
res = vsetq_lane_s32(from[0*stride], res, 0);
|
||||
@@ -241,14 +241,14 @@ template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* f
|
||||
return res;
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, const Packet4f& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, const Packet4f& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = vgetq_lane_f32(from, 0);
|
||||
to[stride*1] = vgetq_lane_f32(from, 1);
|
||||
to[stride*2] = vgetq_lane_f32(from, 2);
|
||||
to[stride*3] = vgetq_lane_f32(from, 3);
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<int, Packet4i>(int* to, const Packet4i& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<int, Packet4i>(int* to, const Packet4i& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = vgetq_lane_s32(from, 0);
|
||||
to[stride*1] = vgetq_lane_s32(from, 1);
|
||||
|
||||
@@ -115,13 +115,13 @@ template<> EIGEN_STRONG_INLINE void pstore <std::complex<float> >(std::complex<f
|
||||
template<> EIGEN_STRONG_INLINE void pstoreu<std::complex<float> >(std::complex<float> * to, const Packet2cf& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu(&numext::real_ref(*to), Packet4f(from.v)); }
|
||||
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packet2cf>(const std::complex<float>* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2cf pgather<std::complex<float>, Packet2cf>(const std::complex<float>* from, DenseIndex stride)
|
||||
{
|
||||
return Packet2cf(_mm_set_ps(std::imag(from[1*stride]), std::real(from[1*stride]),
|
||||
std::imag(from[0*stride]), std::real(from[0*stride])));
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet2cf>(std::complex<float>* to, const Packet2cf& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<std::complex<float>, Packet2cf>(std::complex<float>* to, const Packet2cf& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = std::complex<float>(_mm_cvtss_f32(_mm_shuffle_ps(from.v, from.v, 0)),
|
||||
_mm_cvtss_f32(_mm_shuffle_ps(from.v, from.v, 1)));
|
||||
|
||||
@@ -387,32 +387,32 @@ template<> EIGEN_STRONG_INLINE void pstoreu<double>(double* to, const Packet2d&
|
||||
template<> EIGEN_STRONG_INLINE void pstoreu<float>(float* to, const Packet4f& from) { EIGEN_DEBUG_UNALIGNED_STORE pstoreu(reinterpret_cast<double*>(to), Packet2d(_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), Packet2d(_mm_castsi128_pd(from))); }
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const float* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4f pgather<float, Packet4f>(const float* from, DenseIndex stride)
|
||||
{
|
||||
return _mm_set_ps(from[3*stride], from[2*stride], from[1*stride], from[0*stride]);
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2d pgather<double, Packet2d>(const double* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet2d pgather<double, Packet2d>(const double* from, DenseIndex stride)
|
||||
{
|
||||
return _mm_set_pd(from[1*stride], from[0*stride]);
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline Packet4i pgather<int, Packet4i>(const int* from, DenseIndex stride)
|
||||
{
|
||||
return _mm_set_epi32(from[3*stride], from[2*stride], from[1*stride], from[0*stride]);
|
||||
}
|
||||
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, const Packet4f& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<float, Packet4f>(float* to, const Packet4f& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = _mm_cvtss_f32(from);
|
||||
to[stride*1] = _mm_cvtss_f32(_mm_shuffle_ps(from, from, 1));
|
||||
to[stride*2] = _mm_cvtss_f32(_mm_shuffle_ps(from, from, 2));
|
||||
to[stride*3] = _mm_cvtss_f32(_mm_shuffle_ps(from, from, 3));
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<double, Packet2d>(double* to, const Packet2d& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<double, Packet2d>(double* to, const Packet2d& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = _mm_cvtsd_f64(from);
|
||||
to[stride*1] = _mm_cvtsd_f64(_mm_shuffle_pd(from, from, 1));
|
||||
}
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<int, Packet4i>(int* to, const Packet4i& from, int stride)
|
||||
template<> EIGEN_DEVICE_FUNC inline void pscatter<int, Packet4i>(int* to, const Packet4i& from, DenseIndex stride)
|
||||
{
|
||||
to[stride*0] = _mm_cvtsi128_si32(from);
|
||||
to[stride*1] = _mm_cvtsi128_si32(_mm_shuffle_epi32(from, 1));
|
||||
|
||||
@@ -92,7 +92,7 @@ struct linspaced_op_impl<Scalar,true>
|
||||
|
||||
template<typename Index>
|
||||
EIGEN_STRONG_INLINE const Packet packetOp(Index i) const
|
||||
{ return internal::padd(m_lowPacket, pmul(m_stepPacket, padd(pset1<Packet>(i),m_interPacket))); }
|
||||
{ return internal::padd(m_lowPacket, pmul(m_stepPacket, padd(pset1<Packet>(Scalar(i)),m_interPacket))); }
|
||||
|
||||
const Scalar m_low;
|
||||
const Scalar m_step;
|
||||
@@ -112,7 +112,7 @@ template <typename Scalar, bool RandomAccess> struct functor_traits< linspaced_o
|
||||
template <typename Scalar, bool RandomAccess> struct linspaced_op
|
||||
{
|
||||
typedef typename packet_traits<Scalar>::type Packet;
|
||||
linspaced_op(const Scalar& low, const Scalar& high, DenseIndex num_steps) : impl((num_steps==1 ? high : low), (num_steps==1 ? Scalar() : (high-low)/(num_steps-1))) {}
|
||||
linspaced_op(const Scalar& low, const Scalar& high, DenseIndex num_steps) : impl((num_steps==1 ? high : low), (num_steps==1 ? Scalar() : (high-low)/Scalar(num_steps-1))) {}
|
||||
|
||||
template<typename Index>
|
||||
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Scalar operator() (Index i) const { return impl(i); }
|
||||
|
||||
@@ -320,6 +320,26 @@ struct functor_traits<scalar_asin_op<Scalar> >
|
||||
};
|
||||
};
|
||||
|
||||
|
||||
/** \internal
|
||||
* \brief Template functor to compute the atan of a scalar
|
||||
* \sa class CwiseUnaryOp, ArrayBase::atan()
|
||||
*/
|
||||
template<typename Scalar> struct scalar_atan_op {
|
||||
EIGEN_EMPTY_STRUCT_CTOR(scalar_atan_op)
|
||||
inline const Scalar operator() (const Scalar& a) const { using std::atan; return atan(a); }
|
||||
typedef typename packet_traits<Scalar>::type Packet;
|
||||
inline Packet packetOp(const Packet& a) const { return internal::patan(a); }
|
||||
};
|
||||
template<typename Scalar>
|
||||
struct functor_traits<scalar_atan_op<Scalar> >
|
||||
{
|
||||
enum {
|
||||
Cost = 5 * NumTraits<Scalar>::MulCost,
|
||||
PacketAccess = packet_traits<Scalar>::HasATan
|
||||
};
|
||||
};
|
||||
|
||||
/** \internal
|
||||
* \brief Template functor to compute the inverse of a scalar
|
||||
* \sa class CwiseUnaryOp, Cwise::inverse()
|
||||
|
||||
@@ -218,7 +218,7 @@ struct SelfadjointProductMatrix<Lhs,LhsMode,false,Rhs,0,true>
|
||||
if(!EvalToDest)
|
||||
{
|
||||
#ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
int size = dest.size();
|
||||
Index size = dest.size();
|
||||
EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
#endif
|
||||
MappedDest(actualDestPtr, dest.size()) = dest;
|
||||
@@ -227,7 +227,7 @@ struct SelfadjointProductMatrix<Lhs,LhsMode,false,Rhs,0,true>
|
||||
if(!UseRhs)
|
||||
{
|
||||
#ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
int size = rhs.size();
|
||||
Index size = rhs.size();
|
||||
EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
#endif
|
||||
Map<typename _ActualRhsType::PlainObject>(actualRhsPtr, rhs.size()) = rhs;
|
||||
|
||||
@@ -322,7 +322,7 @@ template<> struct trmv_selector<RowMajor>
|
||||
if(!DirectlyUseRhs)
|
||||
{
|
||||
#ifdef EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
int size = actualRhs.size();
|
||||
Index size = actualRhs.size();
|
||||
EIGEN_DENSE_STORAGE_CTOR_PLUGIN
|
||||
#endif
|
||||
Map<typename _ActualRhsType::PlainObject>(actualRhsPtr, actualRhs.size()) = actualRhs;
|
||||
|
||||
@@ -129,7 +129,6 @@ struct triangular_matrix_vector_product_trmv<Index,Mode,EIGTYPE,ConjLhs,EIGTYPE,
|
||||
MKLPREFIX##axpy(&n, &alpha_,(const MKLTYPE*)x, &incx, (MKLTYPE*)_res, &incy); \
|
||||
/* Non-square case - doesn't fit to MKL ?TRMV. Fall to default triangular product*/ \
|
||||
if (size<(std::max)(rows,cols)) { \
|
||||
typedef Matrix<EIGTYPE, Dynamic, Dynamic> MatrixLhs; \
|
||||
if (ConjRhs) x_tmp = rhs.conjugate(); else x_tmp = rhs; \
|
||||
x = x_tmp.data(); \
|
||||
if (size<rows) { \
|
||||
@@ -214,7 +213,6 @@ struct triangular_matrix_vector_product_trmv<Index,Mode,EIGTYPE,ConjLhs,EIGTYPE,
|
||||
MKLPREFIX##axpy(&n, &alpha_,(const MKLTYPE*)x, &incx, (MKLTYPE*)_res, &incy); \
|
||||
/* Non-square case - doesn't fit to MKL ?TRMV. Fall to default triangular product*/ \
|
||||
if (size<(std::max)(rows,cols)) { \
|
||||
typedef Matrix<EIGTYPE, Dynamic, Dynamic> MatrixLhs; \
|
||||
if (ConjRhs) x_tmp = rhs.conjugate(); else x_tmp = rhs; \
|
||||
x = x_tmp.data(); \
|
||||
if (size<rows) { \
|
||||
|
||||
@@ -236,35 +236,12 @@ template<typename Scalar> class Rotation2D;
|
||||
template<typename Scalar> class AngleAxis;
|
||||
template<typename Scalar,int Dim> class Translation;
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename Derived, int _Dim> class eigen2_RotationBase;
|
||||
template<typename Lhs, typename Rhs> class eigen2_Cross;
|
||||
template<typename Scalar> class eigen2_Quaternion;
|
||||
template<typename Scalar> class eigen2_Rotation2D;
|
||||
template<typename Scalar> class eigen2_AngleAxis;
|
||||
template<typename Scalar,int Dim> class eigen2_Transform;
|
||||
template <typename _Scalar, int _AmbientDim> class eigen2_ParametrizedLine;
|
||||
template <typename _Scalar, int _AmbientDim> class eigen2_Hyperplane;
|
||||
template<typename Scalar,int Dim> class eigen2_Translation;
|
||||
template<typename Scalar,int Dim> class eigen2_Scaling;
|
||||
#endif
|
||||
|
||||
#if EIGEN2_SUPPORT_STAGE < STAGE20_RESOLVE_API_CONFLICTS
|
||||
template<typename Scalar> class Quaternion;
|
||||
template<typename Scalar,int Dim> class Transform;
|
||||
template <typename _Scalar, int _AmbientDim> class ParametrizedLine;
|
||||
template <typename _Scalar, int _AmbientDim> class Hyperplane;
|
||||
template<typename Scalar,int Dim> class Scaling;
|
||||
#endif
|
||||
|
||||
#if EIGEN2_SUPPORT_STAGE > STAGE20_RESOLVE_API_CONFLICTS
|
||||
template<typename Scalar, int Options = AutoAlign> class Quaternion;
|
||||
template<typename Scalar,int Dim,int Mode,int _Options=AutoAlign> class Transform;
|
||||
template <typename _Scalar, int _AmbientDim, int Options=AutoAlign> class ParametrizedLine;
|
||||
template <typename _Scalar, int _AmbientDim, int Options=AutoAlign> class Hyperplane;
|
||||
template<typename Scalar> class UniformScaling;
|
||||
template<typename MatrixType,int Direction> class Homogeneous;
|
||||
#endif
|
||||
|
||||
// MatrixFunctions module
|
||||
template<typename Derived> struct MatrixExponentialReturnValue;
|
||||
@@ -283,18 +260,6 @@ struct stem_function
|
||||
};
|
||||
}
|
||||
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
template<typename ExpressionType> class Cwise;
|
||||
template<typename MatrixType> class Minor;
|
||||
template<typename MatrixType> class LU;
|
||||
template<typename MatrixType> class QR;
|
||||
template<typename MatrixType> class SVD;
|
||||
namespace internal {
|
||||
template<typename MatrixType, unsigned int Mode> struct eigen2_part_return_type;
|
||||
}
|
||||
#endif
|
||||
|
||||
} // end namespace Eigen
|
||||
|
||||
#endif // EIGEN_FORWARDDECLARATIONS_H
|
||||
|
||||
@@ -457,4 +457,16 @@ namespace Eigen {
|
||||
const RHS \
|
||||
>
|
||||
|
||||
#ifdef EIGEN_EXCEPTIONS
|
||||
# define EIGEN_THROW_X(X) throw X
|
||||
# define EIGEN_THROW throw
|
||||
# define EIGEN_TRY try
|
||||
# define EIGEN_CATCH(X) catch (X)
|
||||
#else
|
||||
# define EIGEN_THROW_X(X) std::abort()
|
||||
# define EIGEN_THROW std::abort()
|
||||
# define EIGEN_TRY if (true)
|
||||
# define EIGEN_CATCH(X) else
|
||||
#endif
|
||||
|
||||
#endif // EIGEN_MACROS_H
|
||||
|
||||
@@ -338,15 +338,6 @@ template<> inline void* conditional_aligned_realloc<false>(void* ptr, size_t new
|
||||
*** Construction/destruction of array elements ***
|
||||
*****************************************************************************/
|
||||
|
||||
/** \internal Constructs the elements of an array.
|
||||
* The \a size parameter tells on how many objects to call the constructor of T.
|
||||
*/
|
||||
template<typename T> inline T* construct_elements_of_array(T *ptr, size_t size)
|
||||
{
|
||||
for (size_t i=0; i < size; ++i) ::new (ptr + i) T;
|
||||
return ptr;
|
||||
}
|
||||
|
||||
/** \internal Destructs the elements of an array.
|
||||
* The \a size parameters tells on how many objects to call the destructor of T.
|
||||
*/
|
||||
@@ -357,6 +348,24 @@ template<typename T> inline void destruct_elements_of_array(T *ptr, size_t size)
|
||||
while(size) ptr[--size].~T();
|
||||
}
|
||||
|
||||
/** \internal Constructs the elements of an array.
|
||||
* The \a size parameter tells on how many objects to call the constructor of T.
|
||||
*/
|
||||
template<typename T> inline T* construct_elements_of_array(T *ptr, size_t size)
|
||||
{
|
||||
size_t i;
|
||||
EIGEN_TRY
|
||||
{
|
||||
for (i = 0; i < size; ++i) ::new (ptr + i) T;
|
||||
return ptr;
|
||||
}
|
||||
EIGEN_CATCH(...)
|
||||
{
|
||||
destruct_elements_of_array(ptr, i);
|
||||
EIGEN_THROW;
|
||||
}
|
||||
}
|
||||
|
||||
/*****************************************************************************
|
||||
*** Implementation of aligned new/delete-like functions ***
|
||||
*****************************************************************************/
|
||||
@@ -376,14 +385,30 @@ template<typename T> inline T* aligned_new(size_t size)
|
||||
{
|
||||
check_size_for_overflow<T>(size);
|
||||
T *result = reinterpret_cast<T*>(aligned_malloc(sizeof(T)*size));
|
||||
return construct_elements_of_array(result, size);
|
||||
EIGEN_TRY
|
||||
{
|
||||
return construct_elements_of_array(result, size);
|
||||
}
|
||||
EIGEN_CATCH(...)
|
||||
{
|
||||
aligned_free(result);
|
||||
EIGEN_THROW;
|
||||
}
|
||||
}
|
||||
|
||||
template<typename T, bool Align> inline T* conditional_aligned_new(size_t size)
|
||||
{
|
||||
check_size_for_overflow<T>(size);
|
||||
T *result = reinterpret_cast<T*>(conditional_aligned_malloc<Align>(sizeof(T)*size));
|
||||
return construct_elements_of_array(result, size);
|
||||
EIGEN_TRY
|
||||
{
|
||||
return construct_elements_of_array(result, size);
|
||||
}
|
||||
EIGEN_CATCH(...)
|
||||
{
|
||||
conditional_aligned_free<Align>(result);
|
||||
EIGEN_THROW;
|
||||
}
|
||||
}
|
||||
|
||||
/** \internal Deletes objects constructed with aligned_new
|
||||
@@ -412,7 +437,17 @@ template<typename T, bool Align> inline T* conditional_aligned_realloc_new(T* pt
|
||||
destruct_elements_of_array(pts+new_size, old_size-new_size);
|
||||
T *result = reinterpret_cast<T*>(conditional_aligned_realloc<Align>(reinterpret_cast<void*>(pts), sizeof(T)*new_size, sizeof(T)*old_size));
|
||||
if(new_size > old_size)
|
||||
construct_elements_of_array(result+old_size, new_size-old_size);
|
||||
{
|
||||
EIGEN_TRY
|
||||
{
|
||||
construct_elements_of_array(result+old_size, new_size-old_size);
|
||||
}
|
||||
EIGEN_CATCH(...)
|
||||
{
|
||||
conditional_aligned_free<Align>(result);
|
||||
EIGEN_THROW;
|
||||
}
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
@@ -422,7 +457,17 @@ template<typename T, bool Align> inline T* conditional_aligned_new_auto(size_t s
|
||||
check_size_for_overflow<T>(size);
|
||||
T *result = reinterpret_cast<T*>(conditional_aligned_malloc<Align>(sizeof(T)*size));
|
||||
if(NumTraits<T>::RequireInitialization)
|
||||
construct_elements_of_array(result, size);
|
||||
{
|
||||
EIGEN_TRY
|
||||
{
|
||||
construct_elements_of_array(result, size);
|
||||
}
|
||||
EIGEN_CATCH(...)
|
||||
{
|
||||
conditional_aligned_free<Align>(result);
|
||||
EIGEN_THROW;
|
||||
}
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
@@ -434,7 +479,17 @@ template<typename T, bool Align> inline T* conditional_aligned_realloc_new_auto(
|
||||
destruct_elements_of_array(pts+new_size, old_size-new_size);
|
||||
T *result = reinterpret_cast<T*>(conditional_aligned_realloc<Align>(reinterpret_cast<void*>(pts), sizeof(T)*new_size, sizeof(T)*old_size));
|
||||
if(NumTraits<T>::RequireInitialization && (new_size > old_size))
|
||||
construct_elements_of_array(result+old_size, new_size-old_size);
|
||||
{
|
||||
EIGEN_TRY
|
||||
{
|
||||
construct_elements_of_array(result+old_size, new_size-old_size);
|
||||
}
|
||||
EIGEN_CATCH(...)
|
||||
{
|
||||
conditional_aligned_free<Align>(result);
|
||||
EIGEN_THROW;
|
||||
}
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
@@ -552,7 +607,7 @@ template<typename T> struct smart_memmove_helper<T,false> {
|
||||
// you can overwrite Eigen's default behavior regarding alloca by defining EIGEN_ALLOCA
|
||||
// to the appropriate stack allocation function
|
||||
#ifndef EIGEN_ALLOCA
|
||||
#if (defined __linux__) || (defined __APPLE__)
|
||||
#if (defined __linux__) || (defined __APPLE__) || (defined alloca)
|
||||
#define EIGEN_ALLOCA alloca
|
||||
#elif defined(_MSC_VER)
|
||||
#define EIGEN_ALLOCA _alloca
|
||||
@@ -607,12 +662,9 @@ template<typename T> class aligned_stack_memory_handler
|
||||
* The underlying stack allocation function can controlled with the EIGEN_ALLOCA preprocessor token.
|
||||
*/
|
||||
#ifdef EIGEN_ALLOCA
|
||||
// The native alloca() that comes with llvm aligns buffer on 16 bytes even when AVX is enabled.
|
||||
#if defined(__arm__) || defined(_WIN32) || EIGEN_ALIGN_BYTES > 16
|
||||
#define EIGEN_ALIGNED_ALLOCA(SIZE) reinterpret_cast<void*>((reinterpret_cast<size_t>(EIGEN_ALLOCA(SIZE+EIGEN_ALIGN_BYTES)) & ~(size_t(EIGEN_ALIGN_BYTES-1))) + EIGEN_ALIGN_BYTES)
|
||||
#else
|
||||
#define EIGEN_ALIGNED_ALLOCA EIGEN_ALLOCA
|
||||
#endif
|
||||
// We always manually re-align the result of EIGEN_ALLOCA.
|
||||
// If alloca is already aligned, the compiler should be smart enough to optimize away the re-alignment.
|
||||
#define EIGEN_ALIGNED_ALLOCA(SIZE) reinterpret_cast<void*>((reinterpret_cast<size_t>(EIGEN_ALLOCA(SIZE+EIGEN_ALIGN_BYTES-1)) + EIGEN_ALIGN_BYTES-1) & ~(size_t(EIGEN_ALIGN_BYTES-1)))
|
||||
|
||||
#define ei_declare_aligned_stack_constructed_variable(TYPE,NAME,SIZE,BUFFER) \
|
||||
Eigen::internal::check_size_for_overflow<TYPE>(SIZE); \
|
||||
@@ -637,20 +689,11 @@ template<typename T> class aligned_stack_memory_handler
|
||||
*****************************************************************************/
|
||||
|
||||
#if EIGEN_ALIGN
|
||||
#ifdef EIGEN_EXCEPTIONS
|
||||
#define EIGEN_MAKE_ALIGNED_OPERATOR_NEW_NOTHROW(NeedsToAlign) \
|
||||
#define EIGEN_MAKE_ALIGNED_OPERATOR_NEW_NOTHROW(NeedsToAlign) \
|
||||
void* operator new(size_t size, const std::nothrow_t&) throw() { \
|
||||
try { return Eigen::internal::conditional_aligned_malloc<NeedsToAlign>(size); } \
|
||||
catch (...) { return 0; } \
|
||||
return 0; \
|
||||
EIGEN_TRY { return Eigen::internal::conditional_aligned_malloc<NeedsToAlign>(size); } \
|
||||
EIGEN_CATCH (...) { return 0; } \
|
||||
}
|
||||
#else
|
||||
#define EIGEN_MAKE_ALIGNED_OPERATOR_NEW_NOTHROW(NeedsToAlign) \
|
||||
void* operator new(size_t size, const std::nothrow_t&) throw() { \
|
||||
return Eigen::internal::conditional_aligned_malloc<NeedsToAlign>(size); \
|
||||
}
|
||||
#endif
|
||||
|
||||
#define EIGEN_MAKE_ALIGNED_OPERATOR_NEW_IF(NeedsToAlign) \
|
||||
void *operator new(size_t size) { \
|
||||
return Eigen::internal::conditional_aligned_malloc<NeedsToAlign>(size); \
|
||||
@@ -660,6 +703,8 @@ template<typename T> class aligned_stack_memory_handler
|
||||
} \
|
||||
void operator delete(void * ptr) throw() { Eigen::internal::conditional_aligned_free<NeedsToAlign>(ptr); } \
|
||||
void operator delete[](void * ptr) throw() { Eigen::internal::conditional_aligned_free<NeedsToAlign>(ptr); } \
|
||||
void operator delete(void * ptr, std::size_t /* sz */) throw() { Eigen::internal::conditional_aligned_free<NeedsToAlign>(ptr); } \
|
||||
void operator delete[](void * ptr, std::size_t /* sz */) throw() { Eigen::internal::conditional_aligned_free<NeedsToAlign>(ptr); } \
|
||||
/* in-place new and delete. since (at least afaik) there is no actual */ \
|
||||
/* memory allocated we can safely let the default implementation handle */ \
|
||||
/* this particular case. */ \
|
||||
|
||||
@@ -80,6 +80,34 @@ template<typename T> struct add_const_on_value_type<T*> { typedef T const
|
||||
template<typename T> struct add_const_on_value_type<T* const> { typedef T const* const type; };
|
||||
template<typename T> struct add_const_on_value_type<T const* const> { typedef T const* const type; };
|
||||
|
||||
|
||||
template<typename From, typename To>
|
||||
struct is_convertible_impl
|
||||
{
|
||||
private:
|
||||
struct any_conversion
|
||||
{
|
||||
template <typename T> any_conversion(const volatile T&);
|
||||
template <typename T> any_conversion(T&);
|
||||
};
|
||||
struct yes {int a[1];};
|
||||
struct no {int a[2];};
|
||||
|
||||
static yes test(const To&, int);
|
||||
static no test(any_conversion, ...);
|
||||
|
||||
public:
|
||||
static From ms_from;
|
||||
enum { value = sizeof(test(ms_from, 0))==sizeof(yes) };
|
||||
};
|
||||
|
||||
template<typename From, typename To>
|
||||
struct is_convertible
|
||||
{
|
||||
enum { value = is_convertible_impl<typename remove_all<From>::type,
|
||||
typename remove_all<To >::type>::value };
|
||||
};
|
||||
|
||||
/** \internal Allows to enable/disable an overload
|
||||
* according to a compile time condition.
|
||||
*/
|
||||
|
||||
@@ -107,9 +107,9 @@
|
||||
{Eigen::internal::static_assertion<bool(CONDITION)>::MSG;}
|
||||
|
||||
#else
|
||||
|
||||
// In some cases clang interprets bool(CONDITION) as function declaration
|
||||
#define EIGEN_STATIC_ASSERT(CONDITION,MSG) \
|
||||
if (Eigen::internal::static_assertion<bool(CONDITION)>::MSG) {}
|
||||
if (Eigen::internal::static_assertion<static_cast<bool>(CONDITION)>::MSG) {}
|
||||
|
||||
#endif
|
||||
|
||||
@@ -168,13 +168,8 @@
|
||||
) \
|
||||
)
|
||||
|
||||
#ifdef EIGEN2_SUPPORT
|
||||
#define EIGEN_STATIC_ASSERT_NON_INTEGER(TYPE) \
|
||||
eigen_assert(!NumTraits<Scalar>::IsInteger);
|
||||
#else
|
||||
#define EIGEN_STATIC_ASSERT_NON_INTEGER(TYPE) \
|
||||
#define EIGEN_STATIC_ASSERT_NON_INTEGER(TYPE) \
|
||||
EIGEN_STATIC_ASSERT(!NumTraits<TYPE>::IsInteger, THIS_FUNCTION_IS_NOT_FOR_INTEGER_NUMERIC_TYPES)
|
||||
#endif
|
||||
|
||||
|
||||
// static assertion failing if it is guaranteed at compile-time that the two matrix expression types have different sizes
|
||||
|
||||
Reference in New Issue
Block a user