README

Wrapper for nnls.c as implemented in Chordino / Chroma NNLS by Matthias Mauch and Chris Cannon. See http://www.isophonics.net/nnls-chroma.

Documentation

The package exposes a single function NNLS(a, b, m, n, tol).

a {Float32Array} A flat m x n matrix. m0n0, m1,n0, m2,n0 ... mN,nN.
b {Float32Array} An m-length vector.
m {Number} Number of rows.
n {Number} Number of columns.
tol {Number} Tolerance for precision. 0 produces best results, higher values (eg, 0.01) faster but more approximate.

Returns an n-length Float32Array.

Example usage.

import essyNNLS from 'essy-nnls';

const a = new Float32Array(8);
a[0] = 0.0372;  // m0, n0
a[1] = 0.6861;  // m1, n0
a[2] = 0.6233;  // m2, n0
a[3] = 0.6344;  // m3, n0
a[4] = 0.2869;  // m0, n1
a[5] = 0.7071;  // m1, n1
a[6] = 0.6245;  // m2, n1
a[7] = 0.6170;  // m3, n1

const b = new Float32Array(4);
b[0] = 0.8587;
b[1] = 0.1781;
b[2] = 0.0747;
b[3] = 0.8405;

essyNNLS.NNLS(a, b, 4, 2, 0);  // => [0, 0.629]