matlibr is a barebones implementation of a 2D matrix datatype in C. It also includes some basic arithmetic operations that are relevant to matrices.
matlibr supports
- Double precision and integer data types.
- Basic matrix arithmetic operations like addition, subtraction and multiplication, accelerated by BLAS libraries.
- Matrix-scalar operations like scaling and scalar addition.
- Matrix-vector operations like matrix-vector addition and subtraction.
- Gather operations along rows or columns for index-based access to matrix data.
- Norms of the matrix including L1 and L2 norms.
The following example taken from matrix_addition.c displays the creation and destruction of matrices.
#include <stdio.h>
#include <matrix.h>
int main() {
size_t seed = 42;
int nrow = 10, ncol = 5;
Matrix a = mat_create(nrow, ncol);
// Fill matrix a with random values between 0.0 and 1.0
mat_fill_random(&a, seed);
Matrix b = mat_copy(&a);
mat_scale(&b, 2.34);
printf("\n************Matrix addition***************\n");
printf("A: \n");
mat_print(&a);
printf(" +\n B: \n");
mat_print(&b);
printf(" = \n");
// Addition
mat_add(&a, &b);
mat_print(&a);
printf("\n******************************************\n");
mat_destroy(&a);
mat_destroy(&b);
}The current build only supports Linux-based operating systems.
g++ is the default compiler used to build sources. If it is not installed, please check here for instructions to install gcc, or please use your favorite package manager (dnf for Fedora/RHEL, apt for Ubuntu, pacman for Arch Linux, etc.).
After cloning the repository, we need to set up the dependencies (OpenBLAS). OpenBLAS can be built by running
cd scripts
source setup_dependencies.sh <NUMBER OF THREADS>
It is recommended to provide the maximum number of threads on your machine for the second command for the fastest installation.
NB 1. If you wish, you can install OpenBLAS directly to your system by running source install_openblas.sh <NUMBER OF THREADS> instead.
Once the set up is done, to build matlibr, please run
cd scripts
source build.sh <NUMBER OF THREADS>
This builds a static library libmatlibr.a which can then be linked to your projects.
For installation,
source install.sh <NUMBER OF THREADS>
This installs matlibr to the system.
cd build
ctest -V
This will run the unit tests defined in tests.
Since matlibr has OpenBLAS as a dependency, it is required that OpenBLAS also be downloaded, built and installed alongside matlibr.
Assuming that you have followed the above steps and installed OpenBLAS and matlibr to your system, you can simply link the static libraries and includes at compile time.
g++ -I<PATH TO OPENBLAS INCLUDE DIR> -I<PATH TO MATLIBR INCLUDE DIR> -L<PATH TO OPENBLAS LIBRARY> -L<PATH TO MATLIBR LIBRARY> foo.c -o foo -lmatlibr -lopenblas