gko::matrix::Permutation#

Row/column permutation operator. Stores a permutation array; apply(b, x) permutes the rows of b into x. Used to apply reorderings (RCM, AMD, MC64, nested dissection) and as the permutation half of ScaledPermutation.

template<typename IndexType = int32>
class Permutation #

Inherits from

Permutation is a matrix format that represents a permutation matrix, i.e. a matrix where each row and column has exactly one entry. The matrix can only be applied to Dense inputs, where it represents a row permutation: \(A' = PA\) means \(A'(i, j) = A(p[i], j)\).

Template Parameters:

IndexType – precision of permutation array indices.

Public Functions

inline index_type *get_permutation() noexcept#

Returns a pointer to the array of permutation.

Returns:

the pointer to the row permutation array.

inline const index_type *get_const_permutation() const noexcept#

Returns a pointer to the array of permutation.

Note

This is the constant version of the function, which can be significantly more memory efficient than the non-constant version, so always prefer this version.

Returns:

the pointer to the row permutation array.

size_type get_permutation_size() const noexcept#

Returns the number of elements explicitly stored in the permutation array.

Returns:

the number of elements explicitly stored in the permutation array.

std::unique_ptr<Permutation> compute_inverse() const#

Returns the inverse permutation.

Returns:

a newly created Permutation object storing the inverse permutation of this Permutation.

std::unique_ptr<Permutation> compose(
ptr_param<const Permutation> other,
) const#

Composes this permutation with another permutation. The resulting permutation fulfills result[i] = this[other[i]] or result = other * this from the matrix perspective, which is equivalent to first permuting by this and then by other: Combining permutations \(P_1\) and \(P_2\) with P = P_1.combine(P_2) performs the operation permute(A, P) = permute(permute(A, P_1), P_2).

Parameters:

other – the other permutation

Returns:

the combined permutation

Public Static Functions

static std::unique_ptr<Permutation> create(
std::shared_ptr<const Executor> exec,
size_type size = 0,
)#

Creates an uninitialized Permutation arrays on the specified executor.

Parameters:

exec – Executor associated to the LinOp

Returns:

A smart pointer to the newly created matrix.

static std::unique_ptr<Permutation> create(
std::shared_ptr<const Executor> exec,
array<IndexType> permutation_indices,
)#

Creates a Permutation matrix from an already allocated (and initialized) row and column permutation arrays.

Note

If permutation_indices is not an rvalue, not an array of IndexType, or is on the wrong executor, an internal copy will be created, and the original array data will not be used in the matrix.

Parameters:

  • exec – Executor associated to the matrix

  • size – size of the permutation array.

  • permutation_indices – array of permutation array

  • enabled_permute – mask for the type of permutation to apply.

Returns:

A smart pointer to the newly created matrix.

static std::unique_ptr<const Permutation> create_const(
std::shared_ptr<const Executor> exec,
size_type size,
gko::detail::const_array_view<IndexType> &&perm_idxs,
mask_type enabled_permute = row_permute,
)#

Creates a constant (immutable) Permutation matrix from a constant array.

Parameters:

  • exec – the executor to create the matrix on

  • size – the size of the square matrix

  • perm_idxs – the permutation index array of the matrix

  • enabled_permute – the mask describing the type of permutation

Returns:

A smart pointer to the constant matrix wrapping the input array (if it resides on the same executor as the matrix) or a copy of the array on the correct executor.

static std::unique_ptr<const Permutation> create_const(
std::shared_ptr<const Executor> exec,
gko::detail::const_array_view<IndexType> &&perm_idxs,
)#

Creates a constant (immutable) Permutation matrix from a constant array.

Parameters:

  • exec – the executor to create the matrix on

  • size – the size of the square matrix

  • perm_idxs – the permutation index array of the matrix

  • enabled_permute – the mask describing the type of permutation

Returns:

A smart pointer to the constant matrix wrapping the input array (if it resides on the same executor as the matrix) or a copy of the array on the correct executor.