portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-19 00:07:05 +01:00
Code Releases Activity

Writing of Eigen::Tensor of grid objects now works (for Hdf5)

Browse Source
This commit is contained in:
Michael Marshall
2019-02-11 23:26:18 +00:00
parent 9a225235b6
commit fb2cb3015e
3 changed files with 133 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

118
Grid/serialisation/BaseIO.h
View File

@ -65,12 +65,20 @@ namespace Grid {
template <typename ETensor>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value && (std::is_arithmetic<typename ETensor::Scalar>::value || Grid::is_complex<typename ETensor::Scalar>::value), void>::type
write(const std::string &s, const ETensor &output);
template<typename U, int NumIndices_, int Options_, typename IndexType_>
void write(const std::string &s, const Eigen::Tensor<iScalar<U>, NumIndices_, Options_, IndexType_> &output);
template<typename U, int N, int NumIndices_, int Options_, typename IndexType_>
void write(const std::string &s, const Eigen::Tensor<iVector<U, N>, NumIndices_, Options_, IndexType_> &output);
template<typename U, int N, int NumIndices_, int Options_, typename IndexType_>
void write(const std::string &s, const Eigen::Tensor<iMatrix<U, N>, NumIndices_, Options_, IndexType_> &output);
template <typename ETensor/*, typename U, int N*/>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value && !(std::is_arithmetic<typename ETensor::Scalar>::value || Grid::is_complex<typename ETensor::Scalar>::value)
/*&& ( std::is_base_of<typename ETensor::Scalar, iScalar<U> >::value
|| std::is_base_of<typename ETensor::Scalar, iVector<U, N>>::value
|| std::is_base_of<typename ETensor::Scalar, iMatrix<U, N>>::value )*/, void>::type
write(const std::string &s, const ETensor &output);
/*template <typename ETensor, typename U, int N>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value
&& std::is_base_of<typename ETensor::Scalar, iVector<U, N>>::value, void>::type
write(const std::string &s, const ETensor &output);
template <typename ETensor, typename U, int N>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value
&& std::is_base_of<typename ETensor::Scalar, iMatrix<U, N>>::value, void>::type
write(const std::string &s, const ETensor &output);*/
void scientificFormat(const bool set);
bool isScientific(void);
@ -182,13 +190,12 @@ namespace Grid {
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value && (std::is_arithmetic<typename ETensor::Scalar>::value || Grid::is_complex<typename ETensor::Scalar>::value), void>::type
Writer<T>::write(const std::string &s, const ETensor &output)
{
std::cout << "Eigen::Tensors of arithmetic/complex base type" << std::endl;
const typename ETensor::Index NumElements{output.size()};
assert( NumElements > 0 );
if( NumElements == 1 )
upcast->writeDefault(s, * output.data());
else {
// Create a single, flat vector to hold all the data
std::vector<typename ETensor::Scalar> flat(NumElements);
// We're not interested in trivial dimensions, i.e. dimensions = 1
unsigned int TrivialDimCount{0};
std::vector<size_t> ReducedDims;
@ -203,8 +210,9 @@ namespace Grid {
ReducedDims.push_back(s);
}
}
const unsigned int ReducedDimCount{output.NumDimensions - TrivialDimCount};
assert( ReducedDimCount > 0 ); // NB: NumElements > 1 implies this is not a scalar
assert( output.NumDimensions > TrivialDimCount > 0 ); // NB: NumElements > 1 implies this is not a scalar, so some dims should be left
// Create a single, flat vector to hold all the data
std::vector<typename ETensor::Scalar> flat(NumElements);
// Now copy all the data to my flat vector
// Regardless of the Eigen::Tensor storage order, the copy will be Row Major
std::array<typename ETensor::Index, ETensor::NumIndices> MyIndex;
@ -221,17 +229,71 @@ namespace Grid {
// Eigen::Tensors of iScalar<U>
template <typename T>
template<typename U, int NumIndices_, int Options_, typename IndexType_>
void Writer<T>::write(const std::string &s, const Eigen::Tensor<iScalar<U>, NumIndices_, Options_, IndexType_> &output)
template <typename ETensor/*, typename U, int N*/>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value && !(std::is_arithmetic<typename ETensor::Scalar>::value || Grid::is_complex<typename ETensor::Scalar>::value)
/*&& ( std::is_base_of<typename ETensor::Scalar, iScalar<U> >::value
|| std::is_base_of<typename ETensor::Scalar, iVector<U, N>>::value
|| std::is_base_of<typename ETensor::Scalar, iMatrix<U, N>>::value )*/, void>::type
Writer<T>::write(const std::string &s, const ETensor &output)
{
//upcast->writeDefault(s, tensorToVec(output));
std::cout << "I really should add code to write Eigen::Tensor (iScalar) ..." << std::endl;
std::cout << "Eigen::Tensors of iScalar<U>" << std::endl;
const typename ETensor::Index NumElements{output.size()};
assert( NumElements > 0 );
if( NumElements == 1 )
upcast->writeDefault(s, tensorToVec(* output.data()));
else {
// We're not interested in trivial dimensions, i.e. dimensions = 1
unsigned int TrivialDimCount{0};
std::vector<size_t> ReducedDims;
for(auto i = 0; i < output.NumDimensions; i++ ) {
auto dim = output.dimension(i);
if( dim <= 1 ) {
TrivialDimCount++;
assert( dim == 1 ); // Not expecting dimension to be <= 0
} else {
size_t s = static_cast<size_t>(dim);
assert( s == dim ); // check we didn't lose anything in the conversion
ReducedDims.push_back(s);
}
}
assert( output.NumDimensions > TrivialDimCount > 0 ); // NB: NumElements > 1 implies this is not a scalar, so some dims should be left
// Now add the extra dimensions, based on object zero
typename TensorToVec<typename ETensor::Scalar>::type ttv = tensorToVec(* output.data());
Flatten<typename TensorToVec<typename ETensor::Scalar>::type> f(ttv);
const std::vector<size_t> & ExtraDims{f.getDim()};
size_t ExtraCount{1};
for( auto i : ExtraDims ) {
assert( i > 0 );
ExtraCount *= i;
ReducedDims.push_back(i);
}
typedef typename ETensor::Scalar::scalar_type Scalar;
assert( sizeof( typename ETensor::Scalar ) == ExtraCount * sizeof( Scalar ) );
// Create a single, flat vector to hold all the data
const typename ETensor::Index TotalNumElements = NumElements * ExtraCount;
std::vector<Scalar> flat(TotalNumElements);
// Now copy all the data to my flat vector
// Regardless of the Eigen::Tensor storage order, the copy will be Row Major
std::array<typename ETensor::Index, ETensor::NumIndices> MyIndex;
for( int i = 0 ; i < output.NumDimensions ; i++ ) MyIndex[i] = 0;
for( typename ETensor::Index n = 0; n < TotalNumElements; ) {
const Scalar * p = reinterpret_cast<const Scalar *>( &output( MyIndex ));
for( auto j = 0; j < ExtraCount ; j++ )
flat[n++] = * p++;
// Now increment the index
for( int i = output.NumDimensions - 1; i >= 0 && ++MyIndex[i] == output.dimension(i); i-- )
MyIndex[i] = 0;
}
upcast->template writeMultiDim<Scalar>(s, ReducedDims, flat);
}
}
// Eigen::Tensors of iVector<U, N>
template <typename T>
template<typename U, int N, int NumIndices_, int Options_, typename IndexType_>
void Writer<T>::write(const std::string &s, const Eigen::Tensor<iVector<U, N>, NumIndices_, Options_, IndexType_> &output)
/*template <typename T>
template <typename ETensor, typename U, int N>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value
&& std::is_base_of<typename ETensor::Scalar, iVector<U, N>>::value, void>::type
Writer<T>::write(const std::string &s, const ETensor &output)
{
//upcast->writeDefault(s, tensorToVec(output));
std::cout << "I really should add code to write Eigen::Tensor (iVector) ..." << std::endl;
@ -239,12 +301,14 @@ namespace Grid {
// Eigen::Tensors of iMatrix<U, N>
template <typename T>
template<typename U, int N, int NumIndices_, int Options_, typename IndexType_>
void Writer<T>::write(const std::string &s, const Eigen::Tensor<iMatrix<U, N>, NumIndices_, Options_, IndexType_> &output)
template <typename ETensor, typename U, int N>
typename std::enable_if<std::is_base_of<Eigen::TensorBase<ETensor, Eigen::ReadOnlyAccessors>, ETensor>::value
&& std::is_base_of<typename ETensor::Scalar, iMatrix<U, N>>::value, void>::type
Writer<T>::write(const std::string &s, const ETensor &output)
{
//upcast->writeDefault(s, tensorToVec(output));
std::cout << "I really should add code to write Eigen::Tensor (iMatrix) ..." << std::endl;
}
}*/
template <typename T>
void Writer<T>::scientificFormat(const bool set)
@ -398,6 +462,18 @@ namespace Grid {
}
return bResult;
}
template <typename T>
static inline typename std::enable_if<!std::is_base_of<Eigen::TensorBase<T, Eigen::ReadOnlyAccessors>, T>::value, void>::type
WriteMember(std::ostream &os, const T &object) {
os << object;
}
template <typename T>
static inline typename std::enable_if<std::is_base_of<Eigen::TensorBase<T, Eigen::ReadOnlyAccessors>, T>::value, void>::type
WriteMember(std::ostream &os, const T &object) {
os << "Eigen::Tensor";
}
};
// Generic writer interface //////////////////////////////////////////////////

2
Grid/serialisation/MacroMagic.h
View File

@ -110,7 +110,7 @@ THE SOFTWARE.
#define GRID_MACRO_MEMBER(A,B) A B;
#define GRID_MACRO_COMP_MEMBER(A,B) result = (result and CompareMember(lhs. B, rhs. B));
#define GRID_MACRO_OS_WRITE_MEMBER(A,B) os<< #A <<" " #B << " = " << obj. B << " ; " <<std::endl;
#define GRID_MACRO_OS_WRITE_MEMBER(A,B) os<< #A <<" " #B << " = "; WriteMember( os, obj. B ); os << " ; " <<std::endl;
#define GRID_MACRO_READ_MEMBER(A,B) Grid::read(RD,#B,obj. B);
#define GRID_MACRO_WRITE_MEMBER(A,B) Grid::write(WR,#B,obj. B);