mirror of https://github.com/jlizier/jidt
972 lines
39 KiB
Java
Executable File
972 lines
39 KiB
Java
Executable File
/*
|
|
* Java Information Dynamics Toolkit (JIDT)
|
|
* Copyright (C) 2012, Joseph T. Lizier
|
|
*
|
|
* This program is free software: you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License as published by
|
|
* the Free Software Foundation, either version 3 of the License, or
|
|
* (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
package infodynamics.measures.continuous.kraskov;
|
|
|
|
import java.util.Calendar;
|
|
import java.util.Random;
|
|
|
|
import infodynamics.measures.continuous.ConditionalMutualInfoCalculatorMultiVariate;
|
|
import infodynamics.measures.continuous.ConditionalMutualInfoMultiVariateCommon;
|
|
import infodynamics.utils.EuclideanUtils;
|
|
import infodynamics.utils.KdTree;
|
|
import infodynamics.utils.MathsUtils;
|
|
import infodynamics.utils.MatrixUtils;
|
|
import infodynamics.utils.NearestNeighbourSearcher;
|
|
import infodynamics.utils.UnivariateNearestNeighbourSearcher;
|
|
import infodynamics.utils.EmpiricalMeasurementDistribution;
|
|
import infodynamics.utils.NativeUtils;
|
|
|
|
/**
|
|
* <p>Computes the differential conditional mutual information of two multivariate
|
|
* <code>double[][]</code> sets of observations, conditioned on another
|
|
* (implementing {@link ConditionalMutualInfoCalculatorMultiVariate}),
|
|
* using Kraskov-Stoegbauer-Grassberger (KSG) estimation (see references below).
|
|
* The implementation is made using fast-neighbour searches with an
|
|
* underlying k-d tree algorithm.
|
|
* This is an abstract class, building on the common code base in
|
|
* {@link ConditionalMutualInfoMultiVariateCommon},
|
|
* to gather common functionality between the two
|
|
* algorithms defined by Kraskov et al.
|
|
* Two child classes {@link ConditionalMutualInfoCalculatorMultiVariateKraskov1} and
|
|
* {@link ConditionalMutualInfoCalculatorMultiVariateKraskov2} then
|
|
* actually implement the two KSG algorithms.</p>
|
|
*
|
|
* <p>Crucially, the calculation is performed by examining
|
|
* neighbours in the full joint space (as specified by Frenzel and Pompe,
|
|
* and later by Vejmelka and Paluš, and Vlachos and Kugiumtzis)
|
|
* rather than two MI calculators.</p>
|
|
*
|
|
* <p>Usage is as per the paradigm outlined for {@link ConditionalMutualInfoCalculatorMultiVariate},
|
|
* with:
|
|
* <ul>
|
|
* <li>For constructors see the child classes.</li>
|
|
* <li>Further properties are defined in {@link #setProperty(String, String)}.</li>
|
|
* <li>Computed values are in <b>nats</b>, not bits!</li>
|
|
* </ul>
|
|
* </p>
|
|
*
|
|
* <p>Finally, note that {@link Cloneable} is implemented allowing clone()
|
|
* to produce only an automatic shallow copy, which is fine
|
|
* for the statistical significance calculation it is intended for
|
|
* (none of the array
|
|
* data will be changed there).</p>
|
|
*
|
|
* <p><b>References:</b><br/>
|
|
* <ul>
|
|
* <li>Frenzel and Pompe, <a href="http://dx.doi.org/10.1103/physrevlett.99.204101">
|
|
* "Partial Mutual Information for Coupling Analysis of Multivariate Time Series"</a>,
|
|
* Physical Review Letters, <b>99</b>, p. 204101+ (2007).</li>
|
|
* <li>Vejmelka and Paluš, <a href="http://dx.doi.org/10.1103/physreve.77.026214">
|
|
* "Inferring the directionality of coupling with conditional mutual information"</a>,
|
|
* Physical Review E, <b>77</b>, 026214, (2008)</li>
|
|
* <li>I. Vlachos and D. Kugiumtzis, <a href="http://dx.doi.org/10.1103/physreve.82.016207">
|
|
* "Nonuniform state-space reconstruction and coupling detection"</a>,
|
|
* Physical Review E, <b>82</b>, 016207, (2010)</li>
|
|
* <li>Kraskov, A., Stoegbauer, H., Grassberger, P.,
|
|
* <a href="http://dx.doi.org/10.1103/PhysRevE.69.066138">"Estimating mutual information"</a>,
|
|
* Physical Review E 69, (2004) 066138.</li>
|
|
* </ul>
|
|
*
|
|
* @author Joseph Lizier (<a href="joseph.lizier at gmail.com">email</a>,
|
|
* <a href="http://lizier.me/joseph/">www</a>)
|
|
* @author Ipek Özdemir
|
|
*/
|
|
public abstract class ConditionalMutualInfoCalculatorMultiVariateKraskov
|
|
extends ConditionalMutualInfoMultiVariateCommon
|
|
implements Cloneable { // See comments on clonability above
|
|
|
|
/**
|
|
* we compute distances to the kth neighbour in the joint space
|
|
*/
|
|
protected int k = 4;
|
|
|
|
/**
|
|
* The norm type in use (see {@link #PROP_NORM_TYPE})
|
|
*/
|
|
protected int normType = EuclideanUtils.NORM_MAX_NORM;
|
|
|
|
/**
|
|
* Property name for the number of K nearest neighbours used in
|
|
* the KSG algorithm in the full joint space (default 4).
|
|
*/
|
|
public final static String PROP_K = "k";
|
|
/**
|
|
* Property name for what type of norm to use between data points
|
|
* for each marginal variable -- Options are defined by
|
|
* {@link KdTree#setNormType(String)} and the
|
|
* default is {@link EuclideanUtils#NORM_MAX_NORM}.
|
|
*/
|
|
public final static String PROP_NORM_TYPE = "NORM_TYPE";
|
|
/**
|
|
* Property name for an amount of random Gaussian noise to be
|
|
* added to the data (default 1e-8 to match the noise order in MILCA toolkit.).
|
|
*/
|
|
public static final String PROP_ADD_NOISE = "NOISE_LEVEL_TO_ADD";
|
|
/**
|
|
* Property name for a dynamics exclusion time window
|
|
* otherwise known as Theiler window (see Kantz and Schreiber).
|
|
* Default is 0 which means no dynamic exclusion window.
|
|
*/
|
|
public static final String PROP_DYN_CORR_EXCL_TIME = "DYN_CORR_EXCL";
|
|
/**
|
|
* Property name for the number of parallel threads to use in the
|
|
* computation (default is to use all available)
|
|
*/
|
|
public static final String PROP_NUM_THREADS = "NUM_THREADS";
|
|
/**
|
|
* Valid property value for {@link #PROP_NUM_THREADS} to indicate
|
|
* that all available processors should be used.
|
|
*/
|
|
public static final String USE_ALL_THREADS = "USE_ALL";
|
|
/**
|
|
* Property name for the flag to enable or disable the GPU module.
|
|
*/
|
|
public static final String PROP_USE_GPU = "USE_GPU";
|
|
/**
|
|
* Property name for the path to JIDT GPU library.
|
|
*
|
|
* Path must be full and contain the library filename.
|
|
* Example: /home/johndoe/myfolder/libKraskov.so
|
|
*/
|
|
public static final String PROP_GPU_LIBRARY_PATH = "GPU_LIBRARY_PATH";
|
|
|
|
/**
|
|
* Whether to add an amount of random noise to the incoming data
|
|
*/
|
|
protected boolean addNoise = true;
|
|
/**
|
|
* Amount of random Gaussian noise to add to the incoming data
|
|
*/
|
|
protected double noiseLevel = (double) 1e-8;
|
|
/**
|
|
* Whether we use dynamic correlation exclusion
|
|
*/
|
|
protected boolean dynCorrExcl = false;
|
|
/**
|
|
* Size of dynamic correlation exclusion window.
|
|
*/
|
|
protected int dynCorrExclTime = 0;
|
|
/**
|
|
* Number of parallel threads to use in the computation;
|
|
* defaults to use all available.
|
|
*/
|
|
protected int numThreads = Runtime.getRuntime().availableProcessors();
|
|
/**
|
|
* Private variable to record which KSG algorithm number
|
|
* this instance is implementing
|
|
*/
|
|
protected boolean isAlgorithm1 = false;
|
|
/**
|
|
* Whether to enable the GPU module
|
|
*/
|
|
protected boolean useGPU = false;
|
|
/**
|
|
* Path to JIDT GPU library
|
|
*/
|
|
protected String gpuLibraryPath = "";
|
|
/**
|
|
* Check whether C native code has been loaded
|
|
*/
|
|
protected boolean cudaLibraryLoaded = false;
|
|
|
|
/**
|
|
* protected k-d tree data structure (for fast nearest neighbour searches)
|
|
* representing the joint source-dest space
|
|
*/
|
|
protected KdTree kdTreeJoint;
|
|
/**
|
|
* protected k-d tree data structure (for fast nearest neighbour searches)
|
|
* representing the (var1,conditional) space
|
|
*/
|
|
protected KdTree kdTreeVar1Conditional;
|
|
/**
|
|
* protected univariate neighbour searcher data structure (for fast nearest neighbour searches)
|
|
* representing the (var1) space; used only if var1 is univariate
|
|
*/
|
|
protected UnivariateNearestNeighbourSearcher uniNNSearcherVar1;
|
|
/**
|
|
* protected k-d tree data structure (for fast nearest neighbour searches)
|
|
* representing the (var2,conditional) space
|
|
*/
|
|
protected KdTree kdTreeVar2Conditional;
|
|
/**
|
|
* protected univariate neighbour searcher data structure (for fast nearest neighbour searches)
|
|
* representing the (var2) space; used only if var2 is univariate
|
|
*/
|
|
protected UnivariateNearestNeighbourSearcher uniNNSearcherVar2;
|
|
/**
|
|
* protected data structure (for fast nearest neighbour searches)
|
|
* representing the conditional space.
|
|
* Could be implemented by either a k-d tree or sorted array, depending
|
|
* on whether the conditional is multi-variate or not.
|
|
*/
|
|
protected NearestNeighbourSearcher nnSearcherConditional;
|
|
|
|
/**
|
|
* Constant for digamma(k), with k the number of nearest neighbours selected
|
|
*/
|
|
protected double digammaK;
|
|
|
|
/**
|
|
* Construct an instance of the KSG conditional MI calculator
|
|
*/
|
|
public ConditionalMutualInfoCalculatorMultiVariateKraskov() {
|
|
super();
|
|
}
|
|
|
|
@Override
|
|
public void initialise(int dimensions1, int dimensions2, int dimensionsCond) {
|
|
kdTreeJoint = null;
|
|
kdTreeVar1Conditional = null;
|
|
kdTreeVar2Conditional = null;
|
|
nnSearcherConditional = null;
|
|
uniNNSearcherVar1 = null;
|
|
uniNNSearcherVar2 = null;
|
|
super.initialise(dimensions1, dimensions2, dimensionsCond);
|
|
}
|
|
|
|
/**
|
|
* Sets properties for the KSG conditional MI calculator.
|
|
* New property values are not guaranteed to take effect until the next call
|
|
* to an initialise method.
|
|
*
|
|
* <p>Valid property names, and what their
|
|
* values should represent, include:</p>
|
|
* <ul>
|
|
* <li>{@link #PROP_K} -- number of k nearest neighbours to use in joint kernel space
|
|
* in the KSG algorithm (default is 4).</li>
|
|
* <li>{@link #PROP_NORM_TYPE}</li> -- normalization type to apply to
|
|
* working out the norms between the points in each marginal space.
|
|
* Options are defined by {@link KdTree#setNormType(String)} -
|
|
* default is {@link EuclideanUtils#NORM_MAX_NORM_STRING}.
|
|
* <li>{@link #DYN_CORR_EXCL_TIME_NAME} -- a dynamics exclusion time window,
|
|
* also known as Theiler window (see Kantz and Schreiber);
|
|
* default is 0 which means no dynamic exclusion window.</li>
|
|
* <li>{@link #PROP_ADD_NOISE} -- a standard deviation for an amount of
|
|
* random Gaussian noise to add to
|
|
* each variable, to avoid having neighbourhoods with artificially
|
|
* large counts. (We also accept "false" to indicate "0".)
|
|
* The amount is added in after any normalisation,
|
|
* so can be considered as a number of standard deviations of the data.
|
|
* (Recommended by Kraskov. MILCA uses 1e-8; but adds in
|
|
* a random amount of noise in [0,noiseLevel) ).
|
|
* Default 1e-8 to match the noise order in MILCA toolkit..</li>
|
|
* <li>{@link #PROP_NUM_THREADS} -- the integer number of parallel threads
|
|
* to use in the computation. Can be passed as a string "USE_ALL"
|
|
* to use all available processors on the machine.
|
|
* Default is "USE_ALL".
|
|
* <li>any valid properties for {@link ConditionalMutualInfoMultiVariateCommon#setProperty(String, String)},
|
|
* notably including {@link ConditionalMutualInfoMultiVariateCommon#PROP_NORMALISE}.</li>
|
|
* </ul>
|
|
*
|
|
* <p>Unknown property values are ignored.</p>
|
|
*
|
|
* @param propertyName name of the property
|
|
* @param propertyValue value of the property
|
|
* @throws Exception for invalid property values
|
|
*/
|
|
@Override
|
|
public void setProperty(String propertyName, String propertyValue) {
|
|
if (propertyName.equalsIgnoreCase(PROP_K)) {
|
|
k = Integer.parseInt(propertyValue);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_NORM_TYPE)) {
|
|
normType = KdTree.validateNormType(propertyValue);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_DYN_CORR_EXCL_TIME)) {
|
|
dynCorrExclTime = Integer.parseInt(propertyValue);
|
|
dynCorrExcl = (dynCorrExclTime > 0);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_ADD_NOISE)) {
|
|
if (propertyValue.equals("0") ||
|
|
propertyValue.equalsIgnoreCase("false")) {
|
|
addNoise = false;
|
|
noiseLevel = 0;
|
|
} else {
|
|
addNoise = true;
|
|
noiseLevel = Double.parseDouble(propertyValue);
|
|
}
|
|
} else if (propertyName.equalsIgnoreCase(PROP_NUM_THREADS)) {
|
|
if (propertyValue.equalsIgnoreCase(USE_ALL_THREADS)) {
|
|
numThreads = Runtime.getRuntime().availableProcessors();
|
|
} else { // otherwise the user has passed in an integer:
|
|
numThreads = Integer.parseInt(propertyValue);
|
|
}
|
|
} else if (propertyName.equalsIgnoreCase(PROP_USE_GPU)) {
|
|
useGPU = Boolean.parseBoolean(propertyValue);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_GPU_LIBRARY_PATH)) {
|
|
gpuLibraryPath = propertyValue;
|
|
} else {
|
|
// Assume this is a property for the common parent class
|
|
super.setProperty(propertyName, propertyValue);
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public String getProperty(String propertyName) {
|
|
if (propertyName.equalsIgnoreCase(PROP_K)) {
|
|
return Integer.toString(k);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_NORM_TYPE)) {
|
|
return KdTree.convertNormTypeToString(normType);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_DYN_CORR_EXCL_TIME)) {
|
|
return Integer.toString(dynCorrExclTime);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_ADD_NOISE)) {
|
|
return Double.toString(noiseLevel);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_NUM_THREADS)) {
|
|
return Integer.toString(numThreads);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_USE_GPU)) {
|
|
return Boolean.toString(useGPU);
|
|
} else if (propertyName.equalsIgnoreCase(PROP_GPU_LIBRARY_PATH)) {
|
|
return gpuLibraryPath;
|
|
} else {
|
|
// try the superclass:
|
|
return super.getProperty(propertyName);
|
|
}
|
|
}
|
|
|
|
/* (non-Javadoc)
|
|
* @see infodynamics.measures.continuous.ConditionalMutualInfoMultiVariateCommon#finaliseAddObservations()
|
|
*/
|
|
@Override
|
|
public void finaliseAddObservations() throws Exception {
|
|
// Allow the parent to generate the data for us first
|
|
super.finaliseAddObservations();
|
|
|
|
if (dynCorrExcl && addedMoreThanOneObservationSet) {
|
|
// We have not properly implemented dynamic correlation exclusion for
|
|
// multiple observation sets, so throw an error
|
|
throw new RuntimeException("Addition of multiple observation sets is not currently " +
|
|
"supported with property " + PROP_DYN_CORR_EXCL_TIME + " set");
|
|
}
|
|
|
|
if (totalObservations <= k + 2*dynCorrExclTime) {
|
|
throw new Exception("There are less observations provided (" +
|
|
totalObservations +
|
|
") than required for the number of nearest neighbours parameter (" +
|
|
k + ") and any dynamic correlation exclusion (" + dynCorrExclTime + ")");
|
|
}
|
|
|
|
if (addNoise) {
|
|
Random random = new Random();
|
|
// Add Gaussian noise of std dev noiseLevel to the data
|
|
for (int r = 0; r < var1Observations.length; r++) {
|
|
for (int c = 0; c < dimensionsVar1; c++) {
|
|
var1Observations[r][c] +=
|
|
random.nextGaussian()*noiseLevel;
|
|
}
|
|
for (int c = 0; c < dimensionsVar2; c++) {
|
|
var2Observations[r][c] +=
|
|
random.nextGaussian()*noiseLevel;
|
|
}
|
|
// This next loop will only execute if dimensionsCond > 0
|
|
for (int c = 0; c < dimensionsCond; c++) {
|
|
condObservations[r][c] +=
|
|
random.nextGaussian()*noiseLevel;
|
|
}
|
|
}
|
|
}
|
|
|
|
// Set the constants:
|
|
digammaK = MathsUtils.digamma(k);
|
|
}
|
|
|
|
/**
|
|
* {@inheritDoc}
|
|
*
|
|
* @return the average conditional MI in nats (not bits!)
|
|
*/
|
|
@Override
|
|
public double computeAverageLocalOfObservations() throws Exception {
|
|
// Compute the conditional MI
|
|
double startTime = Calendar.getInstance().getTimeInMillis();
|
|
lastAverage = computeFromObservations(false, null)[0];
|
|
condMiComputed = true;
|
|
if (debug) {
|
|
Calendar rightNow2 = Calendar.getInstance();
|
|
long endTime = rightNow2.getTimeInMillis();
|
|
System.out.println("Calculation time: " + ((endTime - startTime)/1000.0) + " sec" );
|
|
}
|
|
return lastAverage;
|
|
}
|
|
|
|
/**
|
|
* {@inheritDoc}
|
|
*
|
|
* If {@code reordering} is null, it is assumed there is no reordering of
|
|
* the given variable.
|
|
*
|
|
* @return the conditional MI under the new ordering, in nats (not bits!).
|
|
*/
|
|
@Override
|
|
public double computeAverageLocalOfObservations(int variableToReorder,
|
|
int[] reordering) throws Exception {
|
|
|
|
if (reordering == null) {
|
|
return computeAverageLocalOfObservations();
|
|
}
|
|
double[][] originalData;
|
|
KdTree originalKdTreeJoint = kdTreeJoint;
|
|
kdTreeJoint = null; // So that it is rebuilt for the new ordering
|
|
KdTree originalKdTreeVar1Conditional = kdTreeVar1Conditional;
|
|
UnivariateNearestNeighbourSearcher originalUniNNSearcherVar1 = uniNNSearcherVar1;
|
|
KdTree originalKdTreeVar2Conditional = kdTreeVar2Conditional;
|
|
UnivariateNearestNeighbourSearcher originalUniNNSearcherVar2 = uniNNSearcherVar2;
|
|
if (variableToReorder == 1) {
|
|
originalData = var1Observations;
|
|
kdTreeVar1Conditional = null; // So that it is rebuilt for the new ordering if required
|
|
uniNNSearcherVar1 = null; // So that it is rebuilt for the new ordering if required
|
|
var1Observations = MatrixUtils.extractSelectedTimePointsReusingArrays(originalData, reordering);
|
|
} else {
|
|
originalData = var2Observations;
|
|
kdTreeVar2Conditional = null; // So that it is rebuilt for the new ordering
|
|
uniNNSearcherVar2 = null; // So that it is rebuilt for the new ordering if required
|
|
var2Observations = MatrixUtils.extractSelectedTimePointsReusingArrays(originalData, reordering);
|
|
}
|
|
// Compute the conditional MI
|
|
double newCondMI = computeFromObservations(false, null)[0];
|
|
// restore original data
|
|
kdTreeJoint = originalKdTreeJoint;
|
|
if (variableToReorder == 1) {
|
|
var1Observations = originalData;
|
|
kdTreeVar1Conditional = originalKdTreeVar1Conditional;
|
|
uniNNSearcherVar1 = originalUniNNSearcherVar1;
|
|
} else {
|
|
var2Observations = originalData;
|
|
kdTreeVar2Conditional = originalKdTreeVar2Conditional;
|
|
uniNNSearcherVar2 = originalUniNNSearcherVar2;
|
|
}
|
|
return newCondMI;
|
|
}
|
|
|
|
@Override
|
|
public double[] computeLocalOfPreviousObservations() throws Exception {
|
|
double[] localValues = computeFromObservations(true, null);
|
|
lastAverage = MatrixUtils.mean(localValues);
|
|
condMiComputed = true;
|
|
return localValues;
|
|
}
|
|
|
|
/**
|
|
* @returns the series of local conditional MI values in nats, not bits.
|
|
*/
|
|
@Override
|
|
public double[] computeLocalUsingPreviousObservations(double[][] states1,
|
|
double[][] states2, double[][] condStates) throws Exception {
|
|
// Do normalisation of the incoming data if required:
|
|
double[][] states1ToUse, states2ToUse, condStatesToUse;
|
|
if (normalise) {
|
|
states1ToUse = MatrixUtils.normaliseIntoNewArray(states1, var1Means, var1Stds);
|
|
states2ToUse = MatrixUtils.normaliseIntoNewArray(states2, var2Means, var2Stds);
|
|
if (dimensionsCond != 0) {
|
|
condStatesToUse = MatrixUtils.normaliseIntoNewArray(condStates, condMeans, condStds);
|
|
} else {
|
|
condStatesToUse = null;
|
|
}
|
|
} else {
|
|
states1ToUse = states1;
|
|
states2ToUse = states2;
|
|
condStatesToUse = condStates;
|
|
}
|
|
// And call the algorithm:
|
|
double[] localValues = computeFromObservations(true,
|
|
new double[][][]{states1ToUse, states2ToUse, condStatesToUse});
|
|
return localValues;
|
|
}
|
|
|
|
/**
|
|
* This protected method handles the multiple threads which
|
|
* computes either the average or local conditional MI (over parts of the total
|
|
* observations), computing the x, y and z
|
|
* distances between all tuples in time.
|
|
*
|
|
* <p>The method returns:<ol>
|
|
* <li>for (returnLocals == false), an array of size 1,
|
|
* containing the average conditional MI </li>
|
|
* <li>for local conditional MIs (returnLocals == true), the array of local conditional MI values</li>
|
|
* </ol>
|
|
*
|
|
* @param returnLocals whether to return an array or local values, or else
|
|
* sums of these values
|
|
* @param newObservations set to null for computing for the observation set for the PDF, or pass in a new set
|
|
* of observations to compute the average/locals for (using the existing observations to construct the PDF)
|
|
* @return either the average conditional MI, or array of local conditional MI value, in nats not bits
|
|
* @throws Exception
|
|
*/
|
|
protected double[] computeFromObservations(boolean returnLocals, double[][][] newObservations) throws Exception {
|
|
int N = var1Observations.length; // number of observations
|
|
|
|
double[] returnValues = null;
|
|
|
|
// How many time points are we averaging over?
|
|
int numTimePointsToComputeFor = (newObservations == null) ?
|
|
N : newObservations[0].length;
|
|
|
|
if (useGPU) {
|
|
returnValues = gpuComputeFromObservations(0, N, returnLocals);
|
|
} else if (numThreads == 1) {
|
|
// Single-threaded implementation:
|
|
ensureKdTreesConstructed();
|
|
|
|
if (newObservations == null) {
|
|
returnValues = partialComputeFromObservations(0, numTimePointsToComputeFor, returnLocals);
|
|
} else {
|
|
returnValues = partialComputeFromNewObservations(0, numTimePointsToComputeFor,
|
|
newObservations[0], newObservations[1], newObservations[2], returnLocals);
|
|
}
|
|
} else {
|
|
// We're going multithreaded:
|
|
ensureKdTreesConstructed();
|
|
|
|
if (returnLocals) {
|
|
// We're computing local conditional MI
|
|
returnValues = new double[numTimePointsToComputeFor];
|
|
} else {
|
|
// We're computing average conditional MI
|
|
returnValues = new double[CondMiKraskovThreadRunner.RETURN_ARRAY_LENGTH];
|
|
}
|
|
|
|
// Distribute the observations to the threads for the parallel processing
|
|
int lTimesteps = numTimePointsToComputeFor / numThreads; // each thread gets the same amount of data
|
|
int res = numTimePointsToComputeFor % numThreads; // the first thread gets the residual data
|
|
if (debug) {
|
|
System.out.printf("Computing Kraskov conditional MI with %d threads (%d timesteps each, plus %d residual)\n",
|
|
numThreads, lTimesteps, res);
|
|
}
|
|
Thread[] tCalculators = new Thread[numThreads];
|
|
CondMiKraskovThreadRunner[] runners = new CondMiKraskovThreadRunner[numThreads];
|
|
for (int t = 0; t < numThreads; t++) {
|
|
int startTime = (t == 0) ? 0 : lTimesteps * t + res;
|
|
int numTimesteps = (t == 0) ? lTimesteps + res : lTimesteps;
|
|
if (debug) {
|
|
System.out.println(t + ".Thread: from " + startTime +
|
|
" to " + (startTime + numTimesteps)); // Trace Message
|
|
}
|
|
runners[t] = new CondMiKraskovThreadRunner(this, startTime, numTimesteps, newObservations, returnLocals);
|
|
tCalculators[t] = new Thread(runners[t]);
|
|
tCalculators[t].start();
|
|
}
|
|
|
|
// Here, we should wait for the termination of the all threads
|
|
// and collect their results
|
|
for (int t = 0; t < numThreads; t++) {
|
|
if (tCalculators[t] != null) { // TODO Ipek: can you comment on why we're checking for null here?
|
|
tCalculators[t].join();
|
|
}
|
|
// Now we add in the data from this completed thread:
|
|
if (returnLocals) {
|
|
// We're computing local MI; copy these local values
|
|
// into the full array of locals
|
|
System.arraycopy(runners[t].getReturnValues(), 0,
|
|
returnValues, runners[t].myStartTimePoint, runners[t].numberOfTimePoints);
|
|
} else {
|
|
// We're computing the average MI, keep the running sums of digammas and counts
|
|
MatrixUtils.addInPlace(returnValues, runners[t].getReturnValues());
|
|
}
|
|
}
|
|
}
|
|
|
|
// Finalise the results:
|
|
if (returnLocals) {
|
|
return returnValues;
|
|
} else {
|
|
// Average out the components for the final equation(s) and for debugging:
|
|
double averageDiGammas = returnValues[CondMiKraskovThreadRunner.INDEX_SUM_DIGAMMAS] / (double) numTimePointsToComputeFor;
|
|
double avNxz = returnValues[CondMiKraskovThreadRunner.INDEX_SUM_NXZ] / (double) numTimePointsToComputeFor;
|
|
double avNyz = returnValues[CondMiKraskovThreadRunner.INDEX_SUM_NYZ] / (double) numTimePointsToComputeFor;
|
|
double avNz = returnValues[CondMiKraskovThreadRunner.INDEX_SUM_NZ] / (double) numTimePointsToComputeFor;
|
|
if (debug) {
|
|
System.out.printf("<n_xz>=%.3f, <n_yz>=%.3f, <n_z>=%.3f\n",
|
|
avNxz, avNyz, avNz);
|
|
}
|
|
if (this.isAlgorithm1) {
|
|
// Algorithm 1:
|
|
if (debug) {
|
|
System.out.printf("Av = digamma(k)=%.3f + <digammas>=%.3f = %.3f \n",
|
|
MathsUtils.digamma(k), averageDiGammas, MathsUtils.digamma(k) + averageDiGammas);
|
|
}
|
|
double[] result = new double[1];
|
|
result[0] = MathsUtils.digamma(k) + averageDiGammas;
|
|
return result;
|
|
} else {
|
|
// Algorithm 2:
|
|
// We also retrieve the sums of inverses for debugging purposes:
|
|
double averageInverseCountInJointXZ =
|
|
returnValues[CondMiKraskovThreadRunner.INDEX_SUM_INV_NXZ] / (double) numTimePointsToComputeFor;
|
|
double averageInverseCountInJointYZ =
|
|
returnValues[CondMiKraskovThreadRunner.INDEX_SUM_INV_NYZ] / (double) numTimePointsToComputeFor;
|
|
double inverseKTerm;
|
|
if (dimensionsCond > 0) {
|
|
// We will add in the 2/K term as usual
|
|
inverseKTerm = 2.0 / (double) k;
|
|
} else {
|
|
// We will only add in 1/K term since without a conditional there is
|
|
// technically one less variable in the full joint space
|
|
inverseKTerm = 1.0 / (double) k;
|
|
}
|
|
double averageMeasure = MathsUtils.digamma(k) - inverseKTerm + averageDiGammas +
|
|
averageInverseCountInJointXZ + averageInverseCountInJointYZ;
|
|
if (debug) {
|
|
System.out.printf("Av = digamma(k)=%.3f + <digammas>=%.3f +<inverses>=%.3f - $d/k=%.3f = %.3f" +
|
|
" (<1/n_yz>=%.3f, <1/n_xz>=%.3f)\n",
|
|
MathsUtils.digamma(k), averageDiGammas,
|
|
averageInverseCountInJointXZ + averageInverseCountInJointYZ,
|
|
dimensionsCond > 0 ? 2 : 1,
|
|
inverseKTerm, averageMeasure,
|
|
averageInverseCountInJointYZ, averageInverseCountInJointXZ);
|
|
}
|
|
double[] result = new double[1];
|
|
result[0] = averageMeasure;
|
|
return result;
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Protected method to be used internally for threaded implementations.
|
|
* This method implements the guts of each Kraskov algorithm, computing the number of
|
|
* nearest neighbours in each dimension for a sub-set of the data points.
|
|
* It is intended to be called by one thread to work on that specific
|
|
* sub-set of the data.
|
|
*
|
|
* <p>The method returns:<ol>
|
|
* <li>for average conditional MIs (returnLocals == false), the relevant sums of digamma(n_{xz}), digamma(n_{yz})
|
|
* and digamma(n_z)
|
|
* for a partial set of the observations</li>
|
|
* <li>for local conditional MIs (returnLocals == true), the array of local conditional MI values</li>
|
|
* </ol>
|
|
*
|
|
* @param startTimePoint start time for the partial set we examine
|
|
* @param numTimePoints number of time points (including startTimePoint to examine)
|
|
* @param returnLocals whether to return an array or local values, or else
|
|
* sums of these values
|
|
* @return an array of the relevant sum of digamma(n_xz+1) and digamma(n_yz+1) and digamma(n_z), then
|
|
* sum of n_xz, n_yz, n_z and for algorithm 2 only, sum of 1/n_xz and 1/n_yz
|
|
* (these latter five are only for debugging purposes).
|
|
* @throws Exception
|
|
*/
|
|
protected abstract double[] partialComputeFromObservations(
|
|
int startTimePoint, int numTimePoints, boolean returnLocals) throws Exception;
|
|
|
|
/**
|
|
* Protected method to be used internally for GPU implementations.
|
|
* This method serves the same purpose as partialComputeFromObservations,
|
|
* but for GPU computation. Each algorithm must override this method
|
|
* and implement a GPU routine to calculate all values in a single call
|
|
* to the GPU code.
|
|
*/
|
|
protected double[] gpuComputeFromObservations(int startTimePoint,
|
|
int numTimePoints, boolean returnLocals, int nb_surrogates,
|
|
int[][] newOrderings) throws Exception {
|
|
|
|
if (debug) {
|
|
System.out.println("Start GPU calculation");
|
|
}
|
|
|
|
ensureCudaLibraryLoaded();
|
|
|
|
boolean useMaxNorm;
|
|
|
|
if ( normType == EuclideanUtils.NORM_MAX_NORM) {
|
|
useMaxNorm = true;
|
|
} else if ( normType == EuclideanUtils.NORM_EUCLIDEAN || normType == EuclideanUtils.NORM_EUCLIDEAN_SQUARED) {
|
|
useMaxNorm = false;
|
|
} else {
|
|
throw new Exception("Only max and square norms are implemented. Abort.");
|
|
}
|
|
|
|
double[] res;
|
|
|
|
try {
|
|
if (debug) {
|
|
System.out.printf("Calling GPU calculation with returnLocals=%b and nb_surrogates=%d\n", returnLocals, nb_surrogates);
|
|
}
|
|
res = CMIKraskov(totalObservations, var1Observations, dimensionsVar1,
|
|
var2Observations, dimensionsVar2, condObservations, dimensionsCond,
|
|
k, dynCorrExclTime, returnLocals, useMaxNorm,
|
|
isAlgorithm1, nb_surrogates, null!=newOrderings, newOrderings);
|
|
if (debug) {
|
|
System.out.println("GPU calculation finished successfully. Returning results");
|
|
}
|
|
} catch (Throwable e) {
|
|
System.out.println("WARNING. Error in GPU code. Reverting back to CPU.");
|
|
e.printStackTrace();
|
|
res = partialComputeFromObservations(0, totalObservations, returnLocals);
|
|
}
|
|
|
|
return res;
|
|
}
|
|
|
|
/**
|
|
* FIXME
|
|
*/
|
|
protected double[] gpuComputeFromObservations(int startTimePoint,
|
|
int numTimePoints, boolean returnLocals) throws Exception {
|
|
return gpuComputeFromObservations(startTimePoint, numTimePoints, returnLocals, 0, null);
|
|
}
|
|
|
|
/**
|
|
* FIXME
|
|
*/
|
|
protected double[] gpuComputeFromObservations(int startTimePoint,
|
|
int numTimePoints, boolean returnLocals, int[][] newOrderings) throws Exception {
|
|
return gpuComputeFromObservations(startTimePoint, numTimePoints,
|
|
returnLocals, newOrderings.length, newOrderings);
|
|
}
|
|
|
|
/**
|
|
* Native method to calculate CMI in GPU.
|
|
*/
|
|
private native double[] CMIKraskov(
|
|
int N, double[][] source, int dimx, double[][] dest, int dimy,
|
|
double[][] cond, int dimz,
|
|
int k, int theiler, boolean returnLocals, boolean useMaxNorm,
|
|
boolean isAlgorithm1, int nb_surrogates, boolean orderingsGiven,
|
|
int[][] newOrderings);
|
|
|
|
/**
|
|
* Internal method to ensure that the Cuda native library has been correctly
|
|
* loaded.
|
|
*
|
|
* @throws Exception if library not found or unable to load
|
|
*/
|
|
protected void ensureCudaLibraryLoaded() throws Exception {
|
|
|
|
if (!cudaLibraryLoaded) {
|
|
|
|
try {
|
|
if (gpuLibraryPath.length() < 1) {
|
|
NativeUtils.loadLibraryFromJar("/cuda/libKraskov.so");
|
|
} else {
|
|
System.load(gpuLibraryPath);
|
|
}
|
|
} catch (Throwable e) {
|
|
String errmsg = "GPU library not found. To compile GPU code set the enablegpu flag to true in build.xml";
|
|
if (gpuLibraryPath.length() > 0) {
|
|
errmsg += "\nGPU library was not found in the path provided. Provide full path including library file name.";
|
|
errmsg += "\nExample: /home/johndoe/myfolder/libKraskov.so";
|
|
}
|
|
throw new Exception(errmsg);
|
|
}
|
|
|
|
cudaLibraryLoaded = true;
|
|
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Internal method to ensure that the Kd-tree data structures to represent the
|
|
* observational data have been constructed (should be called prior to attempting
|
|
* to use these data structures)
|
|
*/
|
|
protected void ensureKdTreesConstructed() throws Exception {
|
|
|
|
// We need to construct the k-d trees for use by the child
|
|
// classes. We check each tree for existence separately
|
|
// some can be used across original and surrogate data
|
|
// TODO can parallelise these -- best done within the kdTree --
|
|
// though it's unclear if there's much point given that
|
|
// the tree construction itself afterwards can't really be well parallelised.
|
|
if (kdTreeJoint == null) {
|
|
kdTreeJoint = new KdTree(
|
|
new int[] {dimensionsVar1, dimensionsVar2, dimensionsCond},
|
|
new double[][][] {var1Observations, var2Observations, condObservations});
|
|
kdTreeJoint.setNormType(normType);
|
|
}
|
|
if (dimensionsVar1 > 1) {
|
|
if (kdTreeVar1Conditional == null) {
|
|
kdTreeVar1Conditional = new KdTree(
|
|
new int[] {dimensionsVar1, dimensionsCond},
|
|
new double[][][] {var1Observations, condObservations});
|
|
kdTreeVar1Conditional.setNormType(normType);
|
|
}
|
|
} else { // Univariate variable 1, so we'll search its space alone as this is faster
|
|
if (uniNNSearcherVar1 == null) {
|
|
uniNNSearcherVar1 = new UnivariateNearestNeighbourSearcher(var1Observations);
|
|
}
|
|
}
|
|
if (dimensionsVar2 > 1) {
|
|
if (kdTreeVar2Conditional == null) {
|
|
kdTreeVar2Conditional = new KdTree(
|
|
new int[] {dimensionsVar2, dimensionsCond},
|
|
new double[][][] {var2Observations, condObservations});
|
|
kdTreeVar2Conditional.setNormType(normType);
|
|
}
|
|
} else { // Univariate variable 2, so we'll search its space alone as this is faster
|
|
if (uniNNSearcherVar2 == null) {
|
|
uniNNSearcherVar2 = new UnivariateNearestNeighbourSearcher(var2Observations);
|
|
}
|
|
}
|
|
if ((nnSearcherConditional == null) && (dimensionsCond > 0)) {
|
|
nnSearcherConditional = NearestNeighbourSearcher.create(condObservations);
|
|
nnSearcherConditional.setNormType(normType);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* {@inheritDoc}
|
|
*/
|
|
@Override
|
|
public EmpiricalMeasurementDistribution computeSignificance(int numPermutationsToCheck)
|
|
throws Exception {
|
|
if (useGPU) {
|
|
double[] res = gpuComputeFromObservations(0, totalObservations, false, numPermutationsToCheck, null);
|
|
return new EmpiricalMeasurementDistribution(
|
|
MatrixUtils.select(res, 1, res.length - 1), res[0]);
|
|
} else {
|
|
return super.computeSignificance(numPermutationsToCheck);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* {@inheritDoc}
|
|
*/
|
|
@Override
|
|
public EmpiricalMeasurementDistribution computeSignificance(int[][] newOrderings)
|
|
throws Exception {
|
|
if (useGPU) {
|
|
double[] res = gpuComputeFromObservations(0, totalObservations, false, newOrderings.length, newOrderings);
|
|
return new EmpiricalMeasurementDistribution(
|
|
MatrixUtils.select(res, 1, res.length - 1), res[0]);
|
|
} else {
|
|
return super.computeSignificance(newOrderings);
|
|
}
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
* Protected method to be used internally for threaded implementations.
|
|
* This method implements the guts of each Kraskov algorithm, computing the number of
|
|
* nearest neighbours in each dimension for a sub-set of the data points.
|
|
* It is intended to be called by one thread to work on that specific
|
|
* sub-set of the data.
|
|
* In particular, this method differs from {@link #partialComputeFromObservations(int, int, boolean)}
|
|
* because it operates on a new set of observations (using the old set of observations for
|
|
* constructing the search spaces and PDFs)
|
|
*
|
|
* <p>The method returns:<ol>
|
|
* <li>for average conditional MIs (returnLocals == false), the relevant sums of digamma(n_{xz}), digamma(n_{yz})
|
|
* and digamma(n_z)
|
|
* for a partial set of the observations</li>
|
|
* <li>for local conditional MIs (returnLocals == true), the array of local conditional MI values</li>
|
|
* </ol>
|
|
*
|
|
* @param startTimePoint start time for the partial set we examine
|
|
* @param numTimePoints number of time points (including startTimePoint to examine)
|
|
* @param newVar1Observations new time series of observations for variable 1
|
|
* @param newVar2Observations new time series of observations for variable 1
|
|
* @param newCondObservations new time series of observations for variable 1
|
|
* @param returnLocals whether to return an array or local values, or else
|
|
* sums of these values
|
|
* @return an array of the relevant sum of digamma(n_xz+1) and digamma(n_yz+1) and digamma(n_z), then
|
|
* sum of n_xz, n_yz, n_z and for algorithm 2 only, sum of 1/n_xz and 1/n_yz
|
|
* (these latter five are only for debugging purposes).
|
|
* @throws Exception
|
|
*/
|
|
protected abstract double[] partialComputeFromNewObservations(
|
|
int startTimePoint, int numTimePoints,
|
|
double[][] newVar1Observations, double[][] newVar2Observations, double[][] newCondObservations,
|
|
boolean returnLocals) throws Exception;
|
|
|
|
/**
|
|
* Private class to handle multi-threading of the Kraskov algorithms.
|
|
* Each instance calls partialComputeFromObservations()
|
|
* to compute nearest neighbours for a part of the data.
|
|
*
|
|
*
|
|
* @author Joseph Lizier (<a href="joseph.lizier at gmail.com">email</a>,
|
|
* <a href="http://lizier.me/joseph/">www</a>)
|
|
* @author Ipek Özdemir
|
|
*/
|
|
private class CondMiKraskovThreadRunner implements Runnable {
|
|
protected ConditionalMutualInfoCalculatorMultiVariateKraskov condMiCalc;
|
|
protected int myStartTimePoint;
|
|
protected int numberOfTimePoints;
|
|
protected double[][][] newObservations;
|
|
protected boolean computeLocals;
|
|
|
|
protected double[] returnValues = null;
|
|
protected Exception problem = null;
|
|
|
|
public static final int INDEX_SUM_DIGAMMAS = 0;
|
|
public static final int INDEX_SUM_NXZ = 1;
|
|
public static final int INDEX_SUM_NYZ = 2;
|
|
public static final int INDEX_SUM_NZ = 3;
|
|
public static final int INDEX_SUM_INV_NXZ = 4; // Only used for algorithm 2
|
|
public static final int INDEX_SUM_INV_NYZ = 5; // Only used for algorithm 2
|
|
public static final int RETURN_ARRAY_LENGTH = 6;
|
|
|
|
public CondMiKraskovThreadRunner(
|
|
ConditionalMutualInfoCalculatorMultiVariateKraskov condMiCalc,
|
|
int myStartTimePoint, int numberOfTimePoints,
|
|
double[][][] newObservations,
|
|
boolean computeLocals) {
|
|
this.condMiCalc = condMiCalc;
|
|
this.myStartTimePoint = myStartTimePoint;
|
|
this.numberOfTimePoints = numberOfTimePoints;
|
|
this.computeLocals = computeLocals;
|
|
this.newObservations = newObservations;
|
|
}
|
|
|
|
/**
|
|
* Return the values from this part of the data,
|
|
* or throw any exception that was encountered by the
|
|
* thread.
|
|
*
|
|
* @return an exception previously encountered by this thread.
|
|
* @throws Exception
|
|
*/
|
|
public double[] getReturnValues() throws Exception {
|
|
if (problem != null) {
|
|
throw problem;
|
|
}
|
|
return returnValues;
|
|
}
|
|
|
|
/**
|
|
* Start the thread for the given parameters
|
|
*/
|
|
public void run() {
|
|
try {
|
|
if (newObservations == null) {
|
|
// Computing on existing observations
|
|
returnValues = condMiCalc.partialComputeFromObservations(myStartTimePoint, numberOfTimePoints, computeLocals);
|
|
} else {
|
|
// Computing on new observations
|
|
returnValues = condMiCalc.partialComputeFromNewObservations(
|
|
myStartTimePoint, numberOfTimePoints,
|
|
newObservations[0], newObservations[1],
|
|
newObservations[2], computeLocals);
|
|
}
|
|
} catch (Exception e) {
|
|
// Store the exception for later retrieval
|
|
problem = e;
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
// end class MiKraskovThreadRunner
|
|
|
|
// Note: no extra implementation of clone provided; we're simply
|
|
// allowing clone() to produce a shallow copy, which is fine
|
|
// for the statistical significance calculation (none of the array
|
|
// data will be changed there.
|
|
//
|
|
// public ConditionalMutualInfoCalculatorMultiVariateKraskov clone() {
|
|
// return this;
|
|
// }
|
|
}
|