Compute Y += A*X for COO matrix A and dense vectors X,Y
Compute B = A for COO matrix A, CSR matrix B
Compute B += A for COO matrix A, dense matrix B
Compute Y += A*X for CSC matrix A and dense vectors X,Y
Compute Y += A*X for CSC matrix A and dense block vectors X,Y
Compute C = A (binary_op) B for CSR matrices A,B where the column
indices with the rows of A and B are known to be sorted.
Compute C = A (binary_op) B for CSR matrices that are in the
canonical CSR format. Specifically, this method requires that
the rows of the input matrices are free of duplicate column indices
and that the column indices are in sorted order.
Compute C = A (binary_op) B for CSR matrices that are not
necessarily canonical CSR format. Specifically, this method
works even when the input matrices have duplicate and/or
unsorted column indices within a given row.
Slice columns given as an array of indices (pass 1).
This pass counts idx entries and computes a new indptr.
Slice columns given as an array of indices (pass 2).
This pass populates indices/data entries for selected columns.
Count the number of occupied diagonals in CSR matrix A
Extract k-th diagonal of CSR matrix A
Eliminate zero entries from CSR matrix A
Determine whether the matrix structure is canonical CSR.
Canonical CSR implies that column indices within each row
are (1) sorted and (2) unique. Matrices that meet these
conditions facilitate faster matrix computations.
Determine whether the CSR column indices are in sorted order.
Compute CSR entries for matrix C = A*B.
Compute C = A*B for CSR matrices A,B
Compute Y += A*X for CSR matrix A and dense vectors X,Y
Compute Y += A*X for CSR matrix A and dense block vectors X,Y
Slice rows given as an array of indices.
Slice rows given as a (start, stop, step) tuple.
Scale the columns of a CSR matrix in place
Scale the rows of a CSR matrix in place
Sort CSR column indices inplace
Sum together duplicate column entries in each row of CSR matrix A
Compute B = A for CSR matrix A, CSC matrix B
Compute B += A for CSR matrix A, C-contiguous dense matrix B
Expand a compressed row pointer into a row array