SPARSE

Containers

Sparse containers are data structures that store indices and values associated with linear algebra objects, they allow storage savings all values associated with indices not explicitly stored are assumed to be zeros.

SPARSE 1 Kernels

No level 1 kernels have been implemented so far.

SPARSE 2 Kernels

Functions in this category are the equivalent of standard SPARSE-2 type functions.

Functions:

spmv Matrix Vector Multiplication
spmv-struct Matrix Vector Multiplication for structured problems
trsv Triangular Solve for A*x=b
sptrsv Parallelized Sparse Triangular Solve for L*x=b or U*x=b

SPARSE 3 Kernels

Functions in this category are the equivalent of standard SPARSE-3 type functions.

Functions:

spadd Matrix Matrix Addition - symbolic and numeric phases
spgemm Matrix Matrix Multiplication - symbolic and numeric phases

SPARSE Solvers and Preconditioners

Functions in this category are SPARSE solvers preconditioners type functions.

Functions:

gmres Iterative linear solver for positive definite systems of equations
gauss_seidel Iterative method to solve A*x=b
spiluk Incomplete LU factorization ILU(k) - symbolic and numeric phases
par_ilut A Parallel threshold incomplete LU factorization - symbolic and numeric phases

SPARSE Extra Kernels

Functions in this category are Extra SPARSE used for convenience while writing algorithms

Functions:

SortCrs Sorts and merge the entries in a Crs Matrix such that column indices are strictly increasing within each row
coo2crs Converts a coo matrix into a Crs Matrix
getDiagCopy Extract a copy of the diagonal of a Crs Matrix
findRelOffset Extract the indices of diagonal entires of a Crs Matrix