Shape change

Collaboration diagram for Shape change:

template<template< typename > class Matrix, typename C >
Tensor< Matrix, C >	mptensor::transpose (Tensor< Matrix, C > T, const Axes &axes)
	Transposition of tensor with lazy evaluation.
template<template< typename > class Matrix, typename C >
Tensor< Matrix, C >	mptensor::transpose (const Tensor< Matrix, C > &T, const Axes &axes, size_t urank_new)
	Transposition of tensor without lazy evaluation.
template<template< typename > class Matrix, typename C >
Tensor< Matrix, C >	mptensor::reshape (const Tensor< Matrix, C > &T, const Shape &shape_new)
	Change the shape of tensor.
template<template< typename > class Matrix, typename C >
Tensor< Matrix, C >	mptensor::slice (const Tensor< Matrix, C > &T, size_t n_axes, size_t i_begin, size_t i_end)
	Slice a tensor.
template<template< typename > class Matrix, typename C >
Tensor< Matrix, C >	mptensor::slice (const Tensor< Matrix, C > &T, const Index &index_begin, const Index &index_end)
	Slice a tensor.
template<template< typename > class Matrix, typename C >
Tensor< Matrix, C >	mptensor::extend (const Tensor< Matrix, C > &T, const Shape &shape_new)
	Extend the size of a tensor.

Detailed Description

Operations in order to change the shape of a tensor

Function Documentation

extend()

template<template< typename > class Matrix, typename C >

Tensor< Matrix, C > mptensor::extend	(	const Tensor< Matrix, C > &	T,
		const Shape &	shape_new
	)

Extend the size of a tensor.

Parameters

[in]	T	Tensor to be extended.
[in]	shape_new	New shape.

Returns

Extended tensor.

Note

New shape should be larger than the original shape.

reshape()

template<template< typename > class Matrix, typename C >

Tensor< Matrix, C > mptensor::reshape	(	const Tensor< Matrix, C > &	T,
		const Shape &	shape_new
	)

Change the shape of tensor.

Parameters

[in]	T	Tensor to be reshaped.
[in]	shape_new	New shape.

Returns

Reshaped tensor.

Note

The new shape should be compatible with the original shape. The total size of tensor does not change.

Warning

mptensor uses the column major, unlike NumPy.

slice() [1/2]

template<template< typename > class Matrix, typename C >

Tensor< Matrix, C > mptensor::slice	(	const Tensor< Matrix, C > &	T,
		const Index &	index_begin,
		const Index &	index_end
	)

Slice a tensor.

This function mimics slicing in python such as [start:end]. If index_begin[r]==index_end[r], this rank is not sliced. For example,

slice(T, Index(1,0,3), Index(4,0,6)); 

will return T[1:4, :, 3:6].

Parameters

[in]	T	Tensor to be sliced.
[in]	index_begin	Start of indices.
[in]	index_end	End of indices.

Returns

Sliced tensor.

slice() [2/2]

template<template< typename > class Matrix, typename C >

Tensor< Matrix, C > mptensor::slice	(	const Tensor< Matrix, C > &	T,
		size_t	n_axes,
		size_t	i_begin,
		size_t	i_end
	)

Slice a tensor.

This function mimics slicing in python such as [start:end].

Parameters

[in]	T	Tensor to be sliced.
[in]	n_axes	Axes to be sliced.
[in]	i_begin	Start of index.
[in]	i_end	End of index.

Returns

Sliced tensor.

transpose() [1/2]

template<template< typename > class Matrix, typename C >

Tensor< Matrix, C > mptensor::transpose	(	const Tensor< Matrix, C > &	T,
		const Axes &	axes,
		size_t	urank_new
	)

Transposition of tensor without lazy evaluation.

When axes=[1,2,0], \( T_{ijk} \) is transformed into \( T_{jki} \).

Parameters

[in]	T	Tensor to be transposed.
[in]	axes	Order of axes.
[in]	urank_new	Upper rank of new tensor.

Returns

Transposed tensor

Note

This function may cause communications.

transpose() [2/2]

template<template< typename > class Matrix, typename C >

Tensor< Matrix, C > mptensor::transpose ( Tensor< Matrix, C >  T, const Axes &  axes  )

inline

Transposition of tensor with lazy evaluation.

When axes=[1,2,0], \( T_{ijk} \) is transformed into \( T_{jki} \).

Parameters

[in]	T	Tensor to be transposed.
[in]	axes	Order of axes

Returns

Transposed tensor