simple-linalg 😄

Simple Linear algebra function in JS

Installation 📦

npm install simple-linalg

Motivation 🚀

simple-linalg is a spin-off project from the Kalman Filter library. We created this library to provide essential linear algebra functions in a user-friendly, efficient, and accessible way.

We believe in keeping things simple and efficient, and that's what simple-linalg is all about!

Here's what makes simple-linalg stand out:

KISS: Keep It Simple and Stupid 😇 - We focus on delivering a straightforward and easy-to-understand library. Pure Functional: All functions are designed to be pure and functional, which makes it easier to reason about the code 📚. Native JS Array Manipulation: We use native JavaScript arrays for simplicity and compatibility with a wide range of projects 🌐. Small Footprint: Our only dependency is matrix-inverse, which keeps the library lightweight and easy to integrate 🎈.

So, if you're looking for a simple, efficient, and easy-to-use linear algebra library for your JavaScript projects, look no further! simple-linalg is here to help you out! 😊

Examples

add

const {add} = require('simple-linalg');
const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

add(a, b) 
// => [[9, 9], [9, 17]];

diag

const {diag} = require('simple-linalg');

diag([3, 7]) 
// => [[3, 0], [7, 0]];

diagBlock

const {diagBlock} = require('simple-linalg');
const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

const blocks = [a, b];

const result = diagBlock({blocks});

// => [[3, 7, 0, 0], [4, 9, 0, 0], [0, 0, 6, 2], [0, 0, 5, 8]];

cosSimilarity

const {cosSimilarity} = require('simple-linalg');

const a = [0, 7];
const b = [0, 2];

cosSimilarity(a, b) 
// => 1

dotProduct

const {dotProduct} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

dotProduct(a, b) 
// => [[18, 14], [20, 72]];

euclideanDist

const {euclideanDist} = require('simple-linalg');

const a = [3, 6];
const b = [6, 2];

euclideanDist(a, b) 
// => 5

elemWise

const {elemWise} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

let sum = 0;
elemWise([a,b], function(list, rowId, colId){
	const [aValue, bValue] = list;
	sum += aValue+bValue;
	// you can use rowId and colId here
}) 

// sum is 44

frobenius

const {frobenius} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

frobenius(a, b) 
// => 6

identity

const {identity} = require('simple-linalg');

identity(2) 

// [[1,0],[0,1]]

identity

const {invert} = require('simple-linalg');

identity(2) 

// [[1,0],[0,1]]

invert

const {invert} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
invert(a) 

matMul

const {matMul} = require('simple-linalg');
const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

matMul(a, b) 
// => [[53, 62], [69, 80]];

matPermutation

const {matPermutation} = require('simple-linalg');
const a = [[3, 7], [4, 9]];

matPermutation({matrix: a, outputSize: [3,3], rowIndexes: [2, 0], colIndexes: [0,2]}) 
// => [[4, 0, 9], [0, 0, 0], [3, 0, 7]];

mapMatrix

const {mapMatrix} = require('simple-linalg');
const a = [[3, 7], [4, 9]];


let sum = 0;
mapMatrix([a,b], function(value, rowId, colId){
	sum += value;
	// you can use rowId and colId here
}) 
// => sum is 23;

norm

const {norm} = require('simple-linalg');
const a = [3, 4];

norm(a)
// => norm is 5;

padWithZeroCols

const {padWithZeroCols} = require('simple-linalg');
const a = [[3, 7], [4, 9]];

padWithZeroCols(a, {columns: 4}) 
// => [[3, 7, 0, 0], [4, 9, 0, 0]];

subtract

const {subtract} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
const b = [[6, 2], [5, 8]];

sub(a, b) 
// => [[-3, 5], [-1, 1]];

subSquareMatrix

const {subSquareMatrix} = require('simple-linalg');

const a = [[3, 7, 4, 5], [4, 9, 8, 7], [8, 7, 2, -1]];

subSquareMatrix(a, [0, 2]) 
// => [[3, 4], [8, 2]];

sum

const {sum} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
sum(a)

// => 23;

trace

const {trace} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
trace(a)

// => 7;

transpose

const {transpose} = require('simple-linalg');

const a = [[3, 7], [4, 9]];
transpose(a)

// => [[3, 9], [4, 7]];

zeros

const {zeros} = require('simple-linalg');

zeros(2)

// => [[0, 0], [0, 0]];

Going further

This library is not optimized for performance or complex usage If you want to focus on performance, I would suggest moving to Native C-bindings. For example opencv4nodejs