Interoperability

Introduction

JavaPlex can be called from any Java-based system. Below we show some examples.

Matlab

As in the load_javaplex.m file in the matlab examples, one may use the following:

javaaddpath('/path/to/javaplex.jar');
import edu.stanford.math.plex4.*;

Jython

Similarly, from Jython one may use the following:

Invocation:

CLASSPATH=$CLASSPATH:/path/to/javaplex.jar jython

Loading JavaPlex:

from edu.stanford.math.plex4 import *

Java

The following example shows how to run from Java.

In a folder of your choice, download the file javaplex.jar (included in matlab_examples/lib or in javaplex-processing-lib/library from the release distributions https://github.com/appliedtopology/javaplex/releases/ ). In the same folder, create a file named WitnessComplexDemo.java with the following contents:

import edu.stanford.math.plex4.api.Plex4;
import edu.stanford.math.plex4.examples.PointCloudExamples;
import edu.stanford.math.plex4.homology.barcodes.BarcodeCollection;
import edu.stanford.math.plex4.homology.chain_basis.Simplex;
import edu.stanford.math.plex4.homology.interfaces.AbstractPersistenceAlgorithm;
import edu.stanford.math.plex4.metric.impl.EuclideanMetricSpace;
import edu.stanford.math.plex4.metric.landmark.LandmarkSelector;
import edu.stanford.math.plex4.metric.landmark.MaxMinLandmarkSelector;
import edu.stanford.math.plex4.streams.impl.LazyWitnessStream;

public class WitnessComplexDemo {

	public static void main(String[] args) {
		
		int n = 10000;
		int d = 2;
		int numLandmarkPoints = 50;
		double maxDistance = 0.1;
		
		double[][] points = PointCloudExamples.getRandomSpherePoints(n, d);
		EuclideanMetricSpace metricSpace = new EuclideanMetricSpace(points);
		
		LandmarkSelector<double[]> landmarkSelector = new MaxMinLandmarkSelector<double[]>(metricSpace, numLandmarkPoints);
		
		LazyWitnessStream<double[]> stream = new LazyWitnessStream<double[]>(metricSpace, landmarkSelector, d + 1, maxDistance, 20);
		stream.finalizeStream();
		
		System.out.println("Number of simpleces in complex: " + stream.getSize());
		
		AbstractPersistenceAlgorithm<Simplex> algorithm = Plex4.getDefaultSimplicialAlgorithm(d + 1);
		
		BarcodeCollection<Double> intervals = algorithm.computeIntervals(stream);
		
		System.out.println("\nBarcodes for " + d + "-sphere:");
		System.out.println(intervals);
	}
}

To compile the program, execute:

javac -classpath javaplex.jar WitnessComplexDemo.java

To run it, execute:

java -classpath .:javaplex.jar WitnessComplexDemo

Alternatively, one may use an IDE such as Eclipse and import the necessary jar files.

Processing

The Processing IDE and language, focused for producing programmatic art and graphics, works on a Java basis and can access arbitrary Java libraries. With a thin wrapper with some metadata, Java libraries can be packaged as plugins for Processing, making them even easier to use.

As of version 4.1.0, JavaPlex includes a Processing library plugin. Install it where Processing expects its plugin files, and you can import and use JavaPlex primitives - as demonstrated in the Java section above - from your Processing sketch. We also include an example sketch in the JavaPlex distribution demonstrating how to use JavaPlex form Processing.

Mathematica

JavaPlex can be used from the Mathematica computer algebra system by using the J/Link connection layer. Here is an example computation in Mathematica:

Needs["JLink`"];
AddToClassPath["/<path-to-javaplex>/javaplex.jar"];
Plex4 = LoadJavaClass["edu.stanford.math.plex4.api.Plex4"]
sq = Plex4`createEuclideanMetricSpace[{{1, 0}, {1, 1}, {0, 0}, {0, 1}}]
vr = Plex4`createVietorisRipsStream[sq, 2, 2.0]
pers = Plex4`getDefaultSimplicialAlgorithm[2]
ints = pers@computeIntervals[vr]
intit = ints@iterator[]
For[ intit = ints@iterator[],
         intit@hasNext[],
         int = intit@next[],
 Print[{int@getValue[]@get[0]@getFirst[]@getStart[], 
   int@getValue[]@get[0]@getFirst[]@getEnd[]}]]