Skip to main content

Browse the archive

https://github.com/jlizier/jpid
05 April 2024, 21:54:37 UTC
Raw File
Tip revision: 153ffebd8283fbe12612156b695e346379c378f4 authored by Joseph Lizier on 09 January 2013, 11:37:05 UTC
Comment about simple example in unit tests
Tip revision: 153ffeb
readme.html 
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">

<html>
<head>
	<title>JPID - Java Partial Information Decomposition toolkit</title>
	<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
	<!-- link rel="stylesheet" href="../styles.css" type="text/css" -->
</head>

<body> <!-- leftmargin="0" topmargin="0" marginwidth="0" marginheight="0" -->

<h1>Java Partial Information Decomposition</h1>

<p>
A Java software toolkit for computing partial information decomposition (PID) - both on average and on a local scale.
</p>

<p>
This document describes the usage for this toolkit. It is included in the distribution as readme.html.
</p>

<p>
Copyright (C) 2012 Joseph Lizier (joseph.lizier at gmail.com) and Benjamin Flecker (btflecker at gmail.com)<br/>
</p>

<h2>License</h2>
<p>
Licensed under <a href="http://www.gnu.org/licenses/gpl.html">GNU General Public License v3</a> or any later version - see <font face="courier">license-gplv3.txt</font> included in the distribution.
</p>

<h2>Dependencies</h2>
<p>
<ul>
	<li>JUnit 3 - main toolkit code does not require this, but you will need it to run unit tests</li>
</ul>
</p>


<h2>Referencing/Acknowledgement</h2>
<p>
You are asked to reference your use of this toolkit in any resulting publications to the following:
<ul>
 <li>Joseph T. Lizier and Benjamin Flecker, "Java Partial Information Decomposition toolkit", 2012;</li>
 <!-- We won't mention the paper we're working on here for the moment. -->
</ul>
You should also reference the underlying PID method to the original paper:
<ul>
 <li>Paul L. Williams and Randall D. Beer, "Nonnegative Decomposition of Multivariate Information", <a href="http://arxiv.org/abs/1004.2515">arXiv:1004.2515</a>, 2010.</li>
</ul>
</p>

<!--h2>Download</h2>
<p>
Download the most recent distribution of the toolkit <a href="??">here</a>
</p-->

<h2>Usage</h2>

<p>
<b>Structure</b>: The distribution is structured as follows:
<ol>
	<li><font face="courier">java/source</font> - source files for the toolkit.</li>
	<li><font face="courier">java/javadocs</font> - Javadocs for the toolkit.</li>
	<li><font face="courier">java/unittests</font> - JUnit tests for the toolkit.</li>
	<li><font face="courier">python</font> - code to run demonstration applications calling the java toolkit.</li>
	<li><font face="courier">readme.html</font> - this html file</li>
	<li><font face="courier">license-gplv3</font> - GNU General Public License v3.</li>
</ol>
</p>

<p>
<b>Using the toolkit</b>: main use of the toolkit is via the class <font face="courier">partialinfodecomp.discrete.PartialInfoDecomposer</font>. One would typically call its methods in the following order:
<ol>
	<li>Construct, e.g.: <font face="courier">PartialInfoDecomposer pid = new PartialInfoDecomposer(int, int[])</font></li>
	<li>Provide observations (these can be accumulated, i.e. added multiple times):
		<ol>
			<li><font face="courier">pid.addObservations(int[], int[][])</font></li>
			<li><font face="courier">pid.addObservation(int, int[])</font></li>
		</ol>
	</li>
	<li>Compute required Imin and PI entities (all returned in bits):
		<ol>
			<li><font face="courier">pid.Imin(NodeOfElements)</font></li>
			<li><font face="courier">pid.PI(NodeOfElements)</font></li>
		</ol>
	</li>
	<li>Re-initialise (then start adding observations again): <font face="courier">pid.initialise()</font></li>
</ol>
</p>

<p>
<b>Sample code</b>:
A sample application of the code is provided in the python directory ...
</p>

<hr/>
<p class="footer"><a href="mailto:lizier at mis dot mpg dot de">Joe Lizier</a> and<a href="mailto:btflecker at gmail dot com">Benjamin Flecker</a>, Last Updated 07/05/2012.</p>

</body>
</html>
back to top