https://github.com/jakutis/mathopt

Mathematical optimization methods.
https://github.com/jakutis/mathopt

Last synced: 6 days ago
JSON representation

Mathematical optimization methods.

• Host: GitHub
• URL: https://github.com/jakutis/mathopt
• Owner: jakutis
• License: mit
• Created: 2013-12-01T23:14:15.000Z (about 11 years ago)
• Default Branch: master
• Last Pushed: 2014-02-18T19:28:57.000Z (almost 11 years ago)
• Last Synced: 2024-12-14T23:28:36.388Z (about 1 month ago)
• Language: JavaScript
• Homepage: https://www.npmjs.org/package/mathopt
• Size: 172 KB
• Stars: 1
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.markdown
  • License: LICENSE

Awesome Lists containing this project

README

        
# mathopt

A JavaScript library of mathematical optimization methods.

- [Overview](#overview)

- [Installation](#installation)

- [API](#api)

- [Development](#development)

## Overview

* Works on browsers and NodeJS.

* Tested on these browsers:

  * TODO

## Installation

  Install manually by adding to your HTML file:

    

  Install with [npm](https://www.npmjs.org/package/mathopt):

    $ npm install --save mathopt

  Install with [component](http://component.io/jakutis/mathopt):

    $ component install jakutis/mathopt

  Install with [bower](http://bower.io):

    $ bower install --save mathopt

## API

See [a demonstration of examples](https://jakut.is/mathopt/examples/) (source code in "examples" subfolder).

### .basicPSO

Basic Particle Swarm Optimization method finds the global minimum of a given numerical function using particle swarm paradigm.

Implements the algorithm that is described in an article ["Particle swarm optimization" by James Kennedy and Russel Eberhart that is published in proceedings of IEEE International Conference on Neural Networks, 1995](http://dx.doi.org/10.1109/ICNN.1995.488968).

The default parameters are taken from an article ["The Particle Swarm - Explosion, Stability, and Convergence in a Multidimensional Complex Space" by Maurice Clerck and James Kennedy that is published in IEEE Transactions on Evolutionary Computation (2002, issue #1 of volume #6)](http://dx.doi.org/10.1109/4235.985692).

All particles are initialized with the position specified in `initialPosition` option.

Initial velocities are sampled from `U(0, 2)`.

```javascript

var cornfield = function(x, y) {

    return Math.abs(x - 100) + Math.abs(y - 100);

};

// basic usage

// prints "Minimum found at TODO"

console.log('Minimum found at ', mathopt.basicPSO(cornfield));

// demonstration of all the options

mathopt.basicPSO(function(x) {

    return cornfield(x[0], x[1]);

}, {

    // default: 21

    particles: 50,

    // default: inferred from the given function; when specified - the given function must accept a vector

    dimensions: 2,

    // default: [0, 0]

    initialPosition: [5, 5],

    // default: 0.01

    idleSpeed: 0.1,

    // default: 0.7298

    inertia: 0.7,

    // default: 2.9922/2

    localAcceleration: 1.5,

    // default: 2.9922/2

    globalAcceleration: 1.5,

    // default: `function(i, p, v, pbest, best, cb) { cb && setTimeout(cb, 0); }`

    // when onstop is null, cb argument is not passed

    oniteration: function(iteration, positions, velocities, bestpositions, bestparticle, cb) {

        console.log(iteration, bestpositions);

        cb && setTimeout(cb, 0);

    },

    // default: null

    // when set, switches to asynchronous behavior:

    // - .basicPSO does not return anything

    // - oniteration receives a `cb` callback

    onstop: function(min) {

        // prints "Minimum found at x=TODO, y=TODO, after TODO iterations"

        console.log('Minimum found at x=' + min[0] + ', y=' + min[1] + ' after ' + min.iterations + ' iterations');

    }

});

```

## Development

    TODO