Internal API

Library-internal helpers; not part of the public API. More...

Namespaces
namespace	mptensor::internal
	Library-internal helpers and types.

Classes
struct	mptensor::internal::is_unscoped_enum< I, bool >
struct	mptensor::internal::normalized_t

Typedefs
using	mptensor::internal::UIndex = BasicIndex< std::size_t >
using	mptensor::internal::UAxes = UIndex
using	mptensor::internal::UShape = UIndex

Functions
template<typename T , typename I >
T	mptensor::internal::checked_index_cast (I value)
size_t	mptensor::internal::normalize_value (std::ptrdiff_t v, size_t n, bool end_inclusive, const char *kind, std::ptrdiff_t position, const char *position_name="position")
UAxes	mptensor::internal::identity_axes (size_t n)
std::pair< size_t, size_t >	mptensor::internal::normalize_slice_range (std::ptrdiff_t begin, std::ptrdiff_t end, size_t n, size_t axis)
void	mptensor::internal::normalize_slice_ranges (const Index &begin, const Index &end, const UShape &shape, UIndex &ubegin, UIndex &uend)
template<typename MatrixType >
Tensor< MatrixType >	mptensor::internal::transpose_impl (const Tensor< MatrixType > &a, const UAxes &axes, size_t urank_new)
template<typename MatrixType >
Tensor< MatrixType >	mptensor::internal::reshape_impl (const Tensor< MatrixType > &a, const UShape &shape_new)

Variables
template<typename I >
constexpr bool	mptensor::internal::is_index_value_v
constexpr normalized_t	mptensor::internal::normalized {}

Detailed Description

Library-internal helpers; not part of the public API.

Warning

Not part of the public API. These names may change or be removed without notice, and user code should not use them.

Helpers used inside the library. Public functions convert their arguments to the internal types (internal::UIndex, internal::UAxes, internal::UShape, all holding size_t) once at the entry point; everything after that works on size_t.

Typedef Documentation

UAxes

using mptensor::internal::UAxes = typedef UIndex

Internal axes.

UIndex

using mptensor::internal::UIndex = typedef BasicIndex<std::size_t>

Internal element index.

UShape

using mptensor::internal::UShape = typedef UIndex

Internal shape.

Function Documentation

checked_index_cast()

template<typename T , typename I >

T mptensor::internal::checked_index_cast

(

I

value

)

Convert an integer to T, or throw std::out_of_range if it does not fit.

identity_axes()

UAxes mptensor::internal::identity_axes ( size_t  n )

inline

[0, 1, ..., n-1]

normalize_slice_range()

std::pair< size_t, size_t > mptensor::internal::normalize_slice_range ( std::ptrdiff_t  begin, std::ptrdiff_t  end, size_t  n, size_t  axis  )

inline

Normalize a scalar slice [begin, end) on an axis of size n.

Exceptions

std::out_of_range

if a bound is out of range or the slice is empty.

normalize_slice_ranges()

void mptensor::internal::normalize_slice_ranges ( const Index &  begin, const Index &  end, const UShape &  shape, UIndex &  ubegin, UIndex &  uend  )

inline

Normalize Index-style slice bounds; raw begin[r] == end[r] means the full axis.

normalize_value()

size_t mptensor::internal::normalize_value ( std::ptrdiff_t  v, size_t  n, bool  end_inclusive, const char *  kind, std::ptrdiff_t  position, const char *  position_name = "position"  )

inline

Normalize one value into [0, n) (or [0, n] if end_inclusive).

Parameters

v	Value to normalize; negative values count from n.
n	Size of the range (rank or dimension).
end_inclusive	Accept n itself (for an exclusive slice end).
kind	Name of the value shown in the error message, e.g. "axis".
position	Element number (or axis) shown in the error message; negative for a scalar argument.
position_name	Word shown before position, e.g. "position" or "axis".

Exceptions

std::out_of_range

if v is out of range.

reshape_impl()

template<typename MatrixType >

Tensor< MatrixType > mptensor::internal::reshape_impl	(	const Tensor< MatrixType > &	a,
		const UShape &	shape_new
	)

Body of reshape(): shape_new is already normalized.

transpose_impl()

template<typename MatrixType >

Tensor< MatrixType > mptensor::internal::transpose_impl	(	const Tensor< MatrixType > &	a,
		const UAxes &	axes,
		size_t	urank_new
	)

Body of transpose(): axes are already normalized.

Variable Documentation

is_index_value_v

template<typename I >

constexpr bool mptensor::internal::is_index_value_v

inlineconstexpr

Initial value:

=
    std::is_integral_v<I> || is_unscoped_enum<I>::value

True for integer types and unscoped enumerations (accepted as index values).

normalized

constexpr normalized_t mptensor::internal::normalized {}

inlineconstexpr

Value of normalized_t.