RBLAS

Rust bindings and wrappers for BLAS (Basic Linear Algebra Subprograms).

Overview

RBLAS wraps each external call in a trait with the same name (but capitalized). This trait contains a single static method, of the same name. These traits are generic over the four main types of numbers BLAS supports: f32, f64, Complex32, and Complex64.

For example the functions cblas_saxpy, cblas_daxpy, cblas_caxypy, and cblas_zaxpy are called with the function Axpy::axpy.

Additionally, RBLAS introduces a few traits to shorten calls to these BLAS functions: Vector for types that implement vector-like characteristics and Matrix for types that implement matrix-like characteristics. The Vector trait is already implemented by Vec and [] types.

Documentation

Installation

Add this to your Cargo.toml:

[dependencies]
rblas = "0.0.10"

and this to your crate root:

extern crate rblas;

By default, the library links with blas dynamically. To link to an alternate implementation, like OpenBLAS, use the environment variable CARGO_BLAS. If you've already built the bindings, you may need to clean and build again.

export CARGO_BLAS=openblas

Example

extern crate rblas;

use rblas::Dot;

fn main() {
    let x = vec![1.0, -2.0, 3.0, 4.0];
    let y = [1.0, 1.0, 1.0, 1.0, 7.0];

    let d = Dot::dot(&x, &y[..x.len()]);
    assert_eq!(d, 6.0);
}

Sugared Example

#[macro_use]
extern crate rblas as blas;
use blas::math::Mat;
use blas::{Matrix, Vector};
use blas::math::Marker::T;

fn main() {
    let x = vec![1.0, 2.0];
    let xr = &x as &Vector<_>;
    let i = mat![1.0, 0.0; 0.0, 1.0];
    let ir = &i as &Matrix<_>;

    assert!(xr + &x == 2.0 * xr);
    assert!(ir * xr == x);

    let dot = (xr ^ T) * xr;
    assert!(dot == 5.0);
}