Sukker - Linear Algebra library written in rust

Linear algebra in Rust!

Parallelized using rayon with support for many common datatypes, sukker tries to make matrix operations easier for the user, while still giving you as the user the performance you deserve.

Basic operations on sparse matrices are also supported now

Need a feature? Please let me/us know!

Why V2 already?

With added error handling and a good amount of rewriting, a major version was due to avoid any confusion.

Examples

Dens Matrices

use sukker::{LinAlgFloats, Matrix};

fn main() {
    let a = Matrix::<f32>::randomize((8, 56));
    let b = Matrix::<f32>::randomize((56, 8));

    let c = a.matmul(&b).unwrap();

    let res = c.sin().exp(3).unwrap().pow(2).add_val(4.0).abs();

    // To print this beautiful matrix:
    res.print(5);
}

Sparse Matrices

use std::collections::HashMap;
use sukker::{SparseMatrix, SparseMatrixData};

fn main() {
    let mut indexes: SparseMatrixData<f64> = HashMap::new();

    indexes.insert((0, 1), 2.0);
    indexes.insert((1, 0), 4.0);
    indexes.insert((2, 3), 6.0);
    indexes.insert((3, 3), 8.0);

    let sparse = SparseMatrix::<f64>::init(indexes, (4, 4));

    sparse.print(3);
}

More examples can be found here

Documentation

Full API documentation can be found here.

Features

• Easy to use!
• Blazingly fast
• Multiple features on dense matrices
• Serde support
• Support for all signed numeric datatypes
• Can be sent over threads
• Sparse matrices with matrix multiplications