ezp::sparse_csr_mat< DT, IT > Struct Template Reference

Collaboration diagram for ezp::sparse_csr_mat< DT, IT >:

Public Member Functions
	sparse_csr_mat (const sparse_csr_mat &other)
	sparse_csr_mat (sparse_csr_mat &&)=delete
sparse_csr_mat &	operator= (const sparse_csr_mat &)=delete
sparse_csr_mat &	operator= (sparse_csr_mat &&)=default
	sparse_csr_mat (const IT n, const IT nnz, IT *const row_ptr, IT *const col_idx, DT *const data)
template<data_t DT2, index_t IT2>
	sparse_csr_mat (const sparse_coo_mat< DT2, IT2 > &coo, const bool one_based=false, const bool full=false)
template<data_t DT2, index_t IT2>
auto	condense (const sparse_coo_mat< DT2, IT2 > &coo, const std::vector< IT2 > &index, const bool one_based, const bool full)
auto	insert_missing_diagonals (const IT shift)
auto	is_valid ()

Public Attributes
IT	n {}
IT	nnz {}
IT *	row_ptr {nullptr}
IT *	col_idx {nullptr}
DT *	data {nullptr}
std::vector< IT >	row_storage
std::vector< IT >	col_storage
std::vector< DT >	data_storage

Constructor & Destructor Documentation

sparse_csr_mat()

template<data_t DT, index_t IT>

template<data_t DT2, index_t IT2>

ezp::sparse_csr_mat< DT, IT >::sparse_csr_mat ( const sparse_coo_mat< DT2, IT2 > &  coo, const bool  one_based = false, const bool  full = false  )

inlineexplicit

Construct a CSR matrix from a COO matrix. The input matrix will be sorted, duplicated entries will be summed.

Parameters

coo	The input matrix in the COO format.
one_based	Indicate if the input matrix uses one based indexing (if true) or zero based indexing (if false). This flag will be used to generate correct row indexing.
full	Some solvers require diagonal entries to be explicitly present even if they are trivial, if true, all diagonal entries will be explicitly set regardless of the actual values.

Here is the call graph for this function:

Member Function Documentation

condense()

template<data_t DT, index_t IT>

template<data_t DT2, index_t IT2>

auto ezp::sparse_csr_mat< DT, IT >::condense ( const sparse_coo_mat< DT2, IT2 > &  coo, const std::vector< IT2 > &  index, const bool  one_based, const bool  full  )

inline

Sum duplicated entries, drop trivial entries and build the row pointer in a single pass.

Parameters

coo	The input matrix in the COO format.
index	The permutation that sorts the COO entries by row then by column.
one_based	Indicate if the input matrix uses one based indexing.
full	If true, explicitly present diagonal entries are kept even if they are trivial.

Here is the call graph for this function:

Here is the caller graph for this function:

insert_missing_diagonals()

template<data_t DT, index_t IT>

auto ezp::sparse_csr_mat< DT, IT >::insert_missing_diagonals ( const IT  shift )

inline

Ensure every diagonal entry is explicitly present. The main pass above can only keep a diagonal entry that already appears (with any value, including a trivial one) in the input COO matrix. If a diagonal entry is entirely absent from the input, it must be synthesized here with a value of zero, as documented for the full flag.

Parameters

shift

The indexing shift (0 or 1) used for row/col pointers.

Here is the caller graph for this function:

---

The documentation for this struct was generated from the following file:

• ezp/abstract/sparse_solver.hpp