FSparse is an initiative to build a OOP API of sparse matrices with some basic kernels and utility functions using Modern Fortran.

This project is very much a work in progress, contributions are wellcome.

COO: COordinate Sparse format

CSR: Compressed Sparse Row format

CSC: Compressed Sparse Column format

ELL: ELLPACK

SELL-C: sliced ELLPACK format

(Where ? stands for the precision s,d,q)

Conversion subroutines follow the naming pattern _sourcetype2targettype_m ex:

call dense2coo( dense , coo )

(available) Matrix vector products are interfaced by the procedure

call matvec( Mat , vec_x, vec_y ) ! vec_y = vec_y + Mat * vec_x

use fsparse
use iso_fortran_env, only: sp=>real32
type(COO_sp) :: COO
real(sp), allocatable :: dense(:,:)

allocate( dense(4,5) )
dense = reshape([9.0,4.0,0.0,4.0,0.0, &
                 7.0,8.0,0.0,0.0,0.0, &
                 -1.0,5.0,0.0,0.0,8.0,&
                 6.0,-3.0,0.0,0.0,0.0],[4,5])

call dense2coo( dense , COO )

use fsparse
use iso_fortran_env, only: dp=>real64
type(COO_dp) :: CSR
real(dp), allocatable :: vec_x(:)
real(dp), allocatable :: vec_y(:)

call CSR%malloc(4,5,10)
CSR%data(:)   = dble([9,-3,4,7,8,-1,8,4,5,6])
CSR%col(:)    = [1,5,1,2,2,3,4,1,3,4]
CSR%rowptr(:) = [1,3,5,8,11]

allocate( vec_x(5) , source = 1._dp )
allocate( vec_y(4) , source = 0._dp )
call matvec( CSR , vec_x, vec_y )

FYPP is used to enable a generic programming framework, strongly inspired by the stdlib approach. A simple script is distributed to preprocess all .fypp files into .f90 before building.

Only simple and double preceision

python deployement.py

Adding quad precission support

python deployement.py --with_qp

The project was built using the Fortran Package Manager. A manifest file is included to build and test with FPM. For example:

fpm build --profile release
fpm test --profile release

To use fsparse within your FPM project, add the following to your fpm.toml file:

[dependencies]
fsparse = { git="https://github.com/jalvesz/FSPARSE" }

Iterative Methods for Sparse Linear Systems

Efficient Sparse Matrix-Vector Multiplication on cuda

gsl sparse matrices

Calcul Scientifique Parallèle

Implementing a Sparse Matrix Vector Product for the SELL-C/SELL-C-σ formats on NVIDIA GPUs

• José R. Alves Z.
  • Mechanical Engineer, Researcher, Scientific Software Developer
• Samuele Giuli
• Ivan Privec

Compilation of this library was possible thanks to Transvalor S.A. research activities