Detailed Description

Various algorithms that can be applied to arrays and views.

We can group the different algorithms into

Algorithms similar to the standard library: nda::all, nda::any, nda::max_element, nda::min_element.
Reduction algorithms: nda::product, nda::sum.
Generic algorithm: nda::fold.

Functions
template<Array A>
bool	nda::all (A const &a)
	Do all elements of the array evaluate to true?
template<Array A>
bool	nda::any (A const &a)
	Does any of the elements of the array evaluate to true?
template<Array A, typename F>
auto	nda::fold (F f, A const &a)
	The same as nda::fold, except that the initial value is a default constructed value type of the array.
template<Array A, typename F, typename R>
auto	nda::fold (F f, A const &a, R r)
	Perform a fold operation on the given nda::Array object.
template<Array A, Array B>
constexpr auto	nda::hadamard (A &&a, B &&b)
	Hadamard product of two nda::Array objects.
constexpr auto	nda::hadamard (nda::Scalar auto a, nda::Scalar auto b)
	Hadamard product of two arithmetic types.
template<typename T, typename U, size_t R>
constexpr auto	nda::hadamard (std::array< T, R > const &a, std::array< U, R > const &b)
	Hadamard product of two std::array objects.
template<typename T, typename U>
constexpr auto	nda::hadamard (std::vector< T > const &a, std::vector< U > const &b)
	Hadamard product of two std::vector objects.
template<Array A>
auto	nda::max_element (A const &a)
	Find the maximum element of an array.
template<Array A>
auto	nda::min_element (A const &a)
	Find the minimum element of an array.
template<Array A, typename Value = get_value_t<A>> requires (nda::Scalar<Value> or nda::Array<Value>)
auto	nda::product (A const &a)
	Multiply all the elements of an nda::Array object.
template<Array A, typename Value = get_value_t<A>> requires (nda::Scalar<Value> or nda::Array<Value>)
auto	nda::sum (A const &a)
	Sum all the elements of an nda::Array object.
template<Array A>
auto	nda::sum (A const &a, int axis)
	Sum elements of an nda::Array along a specified axis.
template<Array A, std::integral I, size_t N>
auto	nda::sum (A const &a, std::array< I, N > axes)
	Sum elements of an nda::Array along specified axes.

Function Documentation

◆ all()

template<Array A>

bool nda::all ( A const & a )

#include <nda/algorithms.hpp>

Do all elements of the array evaluate to true?

The given nda::Array object can also be some lazy expression that evaluates to a boolean. For example:

auto A = nda::array<double, 2>::rand(2, 3);
auto greater0 = nda::map([](auto x) { return x > 0.0; })(A);
auto res = nda::all(greater0);

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.

Returns: True if all elements of the array evaluate to true, false otherwise.

Definition at line 110 of file algorithms.hpp.

◆ any()

template<Array A>

bool nda::any ( A const & a )

#include <nda/algorithms.hpp>

Does any of the elements of the array evaluate to true?

The given nda::Array object can also be some lazy expression that evaluates to a boolean. For example:

auto A = nda::array<double, 2>::rand(2, 3);
auto greater05 = nda::map([](auto x) { return x > 0.5; })(A);
auto res = nda::any(greater05);

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.

Returns: True if at least one element of the array evaluates to true, false otherwise.

Definition at line 89 of file algorithms.hpp.

◆ fold()

template<Array A, typename F, typename R>

auto nda::fold	(	F	f,
		A const &	a,
		R	r )

#include <nda/algorithms.hpp>

Perform a fold operation on the given nda::Array object.

It calculates the following (where r is an initial value);

auto res = f(...f(f(f(r, a(0,...,0)), a(0,...,1)), a(0,...,2)), ...);

Note: The array is always traversed in C-order.

Template Parameters

A	nda::Array type.
F	Callable type.
R	Type of the initial value.

Parameters

f	Callable object taking two arguments compatible with the initial value and the array value type.
a	nda::Array object.
r	Initial value.

Returns: Result of the fold operation.

Definition at line 59 of file algorithms.hpp.

◆ hadamard() [1/4]

template<Array A, Array B>

auto nda::hadamard	(	A &&	a,
		B &&	b )

nodiscardconstexpr

#include <nda/algorithms.hpp>

Hadamard product of two nda::Array objects.

Template Parameters

A	nda::Array type.
B	nda::Array type.

Parameters

a	nda::Array object.
b	nda::Array object.

Returns: A lazy nda::expr_call object representing the elementwise product of the two input objects.

Definition at line 293 of file algorithms.hpp.

◆ hadamard() [2/4]

auto nda::hadamard	(	nda::Scalar auto	a,
		nda::Scalar auto	b )

constexpr

#include <nda/algorithms.hpp>

Hadamard product of two arithmetic types.

Template Parameters

T	nda::Scalar type of the first input.
U	nda::Scalar type of the second input.

Parameters

a	First input.
b	Second input.

Returns: Product of the two inputs.

Definition at line 340 of file algorithms.hpp.

◆ hadamard() [3/4]

template<typename T, typename U, size_t R>

auto nda::hadamard	(	std::array< T, R > const &	a,
		std::array< U, R > const &	b )

nodiscardconstexpr

#include <nda/algorithms.hpp>

Hadamard product of two std::array objects.

Template Parameters

T	Value type of the first array.
U	Value type of the second array.
R	Size of the arrays.

Parameters

a	std::array object.
b	std::array object.

Returns: std::array containing the elementwise product of the two input arrays.

Definition at line 308 of file algorithms.hpp.

◆ hadamard() [4/4]

template<typename T, typename U>

auto nda::hadamard	(	std::vector< T > const &	a,
		std::vector< U > const &	b )

nodiscardconstexpr

#include <nda/algorithms.hpp>

Hadamard product of two std::vector objects.

Template Parameters

T	Value type of the first input vector.
U	Value type of the second input vector.

Parameters

a	std::vector object.
b	std::vector object.

Returns: std::vector containing the elementwise product of the two input vectors.

Definition at line 322 of file algorithms.hpp.

◆ max_element()

template<Array A>

auto nda::max_element ( A const & a )

#include <nda/algorithms.hpp>

Find the maximum element of an array.

It uses nda::fold and std::max.

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.

Returns: Maximum element of the array.

Definition at line 125 of file algorithms.hpp.

◆ min_element()

template<Array A>

auto nda::min_element ( A const & a )

#include <nda/algorithms.hpp>

Find the minimum element of an array.

It uses nda::fold and std::min.

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.

Returns: Minimum element of the array.

Definition at line 144 of file algorithms.hpp.

◆ product()

template<Array A, typename Value = get_value_t<A>>
requires (nda::Scalar<Value> or nda::Array<Value>)

auto nda::product ( A const & a )

#include <nda/algorithms.hpp>

Multiply all the elements of an nda::Array object.

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.

Returns: Product of all elements.

Definition at line 272 of file algorithms.hpp.

◆ sum() [1/3]

template<Array A, typename Value = get_value_t<A>>
requires (nda::Scalar<Value> or nda::Array<Value>)

auto nda::sum ( A const & a )

#include <nda/algorithms.hpp>

Sum all the elements of an nda::Array object.

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.

Returns: Sum of all elements.

Definition at line 179 of file algorithms.hpp.

◆ sum() [2/3]

template<Array A>

auto nda::sum	(	A const &	a,
		int	axis )

#include <nda/algorithms.hpp>

Sum elements of an nda::Array along a specified axis.

It simply calls nda::sum(A const &, std::array<I, N>).

Template Parameters

A	nda::Array type.

Parameters

a	nda::Array object.
axis	The axis along which to sum.

Returns: An nda::array with rank reduced by 1 or a scalar if the original array has rank 1.

Definition at line 260 of file algorithms.hpp.

◆ sum() [3/3]

template<Array A, std::integral I, size_t N>

auto nda::sum	(	A const &	a,
		std::array< I, N >	axes )

#include <nda/algorithms.hpp>

Sum elements of an nda::Array along specified axes.

This function behaves similar to numpy.sum. The result is an array with rank reduced by the number of axes summed over, i.e. if the original array has rank \( R \) and we sum over \( N \) axes, the resulting array has rank \( R - N \).

If all axes are summed over, \( R = N \), the call is dispatched to nda::sum.

If no axes are specified, \( N = 0 \), a copy of the input array is returned.

The given axes are expected to be unique and in the range \( [0, R-1] \).

Template Parameters

A	nda::Array type.
N	Number of axes to sum over.

Parameters

a	nda::Array object.
axes	Array of axis indices to sum over.

Returns: An nda::basic_array with rank reduced by \( N \), or a scalar if all axes are summed.

Definition at line 210 of file algorithms.hpp.

Detailed Description

Functions

Function Documentation

◆ all()

◆ any()

◆ fold()

◆ hadamard() [1/4]

◆ hadamard() [2/4]

◆ hadamard() [3/4]

◆ hadamard() [4/4]

◆ max_element()

◆ min_element()

◆ product()

◆ sum() [1/3]

◆ sum() [2/3]

◆ sum() [3/3]