Example usage for org.apache.commons.math.linear RealMatrix getSubMatrix

List of usage examples for org.apache.commons.math.linear RealMatrix getSubMatrix

  1. HOME
  2. Java
  3. org.apache.commons
  4. org.apache.commons.math.linear.*
  5. RealMatrix
  6. getSubMatrix

Introduction

In this page you can find the example usage for org.apache.commons.math.linear RealMatrix getSubMatrix.

Prototype

RealMatrix getSubMatrix(int startRow, int endRow, int startColumn, int endColumn) throws MatrixIndexException;

Source Link

Document

Gets a submatrix.

Usage

From source file:de.mpicbg.knime.hcs.base.utils.Table2Matrix.java

public static RealMatrix extractMatrix(List<DataRow> rows, List<Attribute> params) {
    double[][] matrix = new double[rows.size()][params.size()];
    int nbparams = params.size();
    int m = 0;/*from   ww  w.  ja  v  a 2  s.  com*/
    for (DataRow row : rows) {
        int n = 0;
        for (Attribute readout : params) {
            Double val = readout.getDoubleAttribute(row);
            if ((val == null) || Double.isInfinite(val) || Double.isNaN(val)) {
                break;
            }
            matrix[m][n] = val;
            n += 1;
        }
        if (n == nbparams) {
            m += 1;
        }
    }
    // remove the unused rows.
    RealMatrix rmatrix = new Array2DRowRealMatrix(matrix);
    if (m > 0) {
        rmatrix = rmatrix.getSubMatrix(0, m - 1, 0, nbparams - 1);
    }
    return rmatrix;
}

From source file:juicebox.tools.utils.common.MatrixTools.java

/**
 * @return region within matrix specified by indices
 *///w  ww .  j a  v  a 2 s  .c o  m
public static RealMatrix getSubMatrix(RealMatrix matrix, int[] indices) {
    return matrix.getSubMatrix(indices[0], indices[1], indices[2], indices[3]);
}

From source file:net.sf.jtmt.similarity.matrix.AbstractSimilarity.java

public RealMatrix transform(RealMatrix termDocumentMatrix) {
    int numDocs = termDocumentMatrix.getColumnDimension();
    RealMatrix similarityMatrix = new OpenMapRealMatrix(numDocs, numDocs);
    for (int i = 0; i < numDocs; i++) {
        RealMatrix sourceDocMatrix = termDocumentMatrix.getSubMatrix(0,
                termDocumentMatrix.getRowDimension() - 1, i, i);
        for (int j = 0; j < numDocs; j++) {
            RealMatrix targetDocMatrix = termDocumentMatrix.getSubMatrix(0,
                    termDocumentMatrix.getRowDimension() - 1, j, j);
            similarityMatrix.setEntry(i, j, computeSimilarity(sourceDocMatrix, targetDocMatrix));
        }/*w  ww .ja v  a2  s .  c om*/
    }
    return similarityMatrix;
}

From source file:net.sf.jtmt.clustering.DocumentCollection.java

/**
 * Instantiates a new document collection.
 *
 * @param tdMatrix the td matrix//  w w w  .  ja va  2 s .c  o m
 * @param docNames the doc names
 */
public DocumentCollection(RealMatrix tdMatrix, String[] docNames) {
    int position = 0;
    this.tdMatrix = tdMatrix;
    this.documentMap = new HashMap<String, RealMatrix>();
    this.documentNames = new ArrayList<String>();
    for (String documentName : docNames) {
        documentMap.put(documentName,
                tdMatrix.getSubMatrix(0, tdMatrix.getRowDimension() - 1, position, position));
        documentNames.add(documentName);
        position++;
    }
}

From source file:juicebox.tools.utils.juicer.apa.APARegionStatistics.java

public APARegionStatistics(RealMatrix data) {
    int max = data.getColumnDimension();
    int midPoint = max / 2;
    double centralVal = data.getEntry(midPoint, midPoint);

    int regionWidth = APA.regionWidth;

    /** NOTE - indices are inclusive in java, but in python the second index is not inclusive */

    peak2mean = centralVal / ((sum(data.getData()) - centralVal) / (data.getColumnDimension() - 1));

    double avgUL = mean(data.getSubMatrix(0, regionWidth - 1, 0, regionWidth - 1).getData());
    peak2UL = centralVal / avgUL;//from w  ww.j a v  a  2s  . co  m

    double avgUR = mean(data.getSubMatrix(0, regionWidth - 1, max - regionWidth, max - 1).getData());
    peak2UR = centralVal / avgUR;

    double avgLL = mean(data.getSubMatrix(max - regionWidth, max - 1, 0, regionWidth - 1).getData());
    peak2LL = centralVal / avgLL;

    double avgLR = mean(data.getSubMatrix(max - regionWidth, max - 1, max - regionWidth, max - 1).getData());
    peak2LR = centralVal / avgLR;

    DescriptiveStatistics yStats = statistics(
            data.getSubMatrix(max - regionWidth, max - 1, 0, regionWidth - 1).getData());
    ZscoreLL = (centralVal - yStats.getMean()) / yStats.getStandardDeviation();
}

From source file:lib.regressions.MultipleRegression.java

/**
 * Perform the regression computations/*from  w  ww. j  a  v  a  2  s.co m*/
 */
private void compute() {
    // Set everything to 0.
    for (int i = 0; i < (myNumVar + 1); i++) {
        myCoef[i] = 0.0;
        myStdErr[i] = 0.0;
        myTStat[i] = 0.0;
    }
    myChiSq = 0.0;
    myRSq = 0.0;
    myAdjustedRSq = 0.0;

    // Set coefficients, t-stat, etc. if there has been enough data added.
    if (myCount >= (myNumVar + 1)) {
        RealMatrix dataMatrix = new RealMatrixImpl(mySums.getSumXX());
        RealMatrix xxMatrix = dataMatrix.getSubMatrix(1, myNumVar + 1, 1, myNumVar + 1);
        RealMatrix xyMatrix = dataMatrix.getSubMatrix(1, myNumVar + 1, 0, 0);

        computeOkX(); // Determine which X components to use.
        int[] listX = getListX();
        int[] listY = { 0 };
        int numX = listX.length;
        RealMatrix xxSubMatrix = xxMatrix.getSubMatrix(listX, listX);
        RealMatrix xySubMatrix = xyMatrix.getSubMatrix(listX, listY);

        double sumY = mySums.getSumXX()[0][1];
        double sumYY = mySums.getSumXX()[0][0];

        if (!xxSubMatrix.isSingular()) {
            RealMatrix xxInverse = xxSubMatrix.inverse();
            RealMatrix coefMatrix = xxInverse.multiply(xySubMatrix);
            double[] coef = coefMatrix.getColumn(0);

            // Compute chi-squared
            myChiSq = sumYY - 2 * coefMatrix.transpose().multiply(xySubMatrix).getEntry(0, 0)
                    + +coefMatrix.transpose().multiply(xxSubMatrix).multiply(coefMatrix).getEntry(0, 0);

            // Compute R^2 and adjusted R^2
            int offset = getUseIntercept() ? 1 : 0;
            myRSq = 1.0 - myChiSq / (sumYY - sumY * sumY / myCount);
            myAdjustedRSq = 1 - ((1 - myRSq) * (myCount - 1) + 1 - offset) / (myCount - numX);

            // Compute standard errors and t-stats
            int j = 0;
            for (int i = 0; i < (myNumVar + 1); i++) {
                if (myOkX[i]) {
                    j++;
                    myCoef[i] = coef[j - 1];
                    myStdErr[i] = Math.sqrt(myChiSq * xxInverse.getEntry(j - 1, j - 1) / (myCount - numX));
                    myTStat[i] = myCoef[i] / myStdErr[i];
                }
            }
        }

    }
    myIsComputed = true;
}

From source file:name.mjw.cytospade.fcsFile.java

/**
 * getCompensatedEventList ---//from  w ww.  j  av a  2  s . c  o  m
 * <p>
 * Returns the event list compensated by the SPILL matrix.
 * <p>
 *
 * @return array of double arrays containing the events.
 */
public double[][] getCompensatedEventList() {
    double[][] events = this.getEventList();
    if (events.length != this.getNumChannels())
        return events; // Unable to extract the underlying events

    // Convert the SPILL string to a compensation matrix
    String compString = this.getSpillString();
    if (compString == null)
        return events; // No compensation, just return the events

    // Split the compensation string into its values
    //
    // The basic structure for SPILL* is:
    // $SPILLOVER/n,string1,string2,...,f1,f2,f3,f4,.../

    String[] compValues = compString.split(",");
    String[] compNames = null;
    String[] compData = null;
    int compDataStart = 0;

    int n = 0;
    try {
        // Try to parse the number of acquisition parameters
        n = Integer.parseInt(compValues[0]);
        if (n <= 0 || n > this.parameters)
            throw new NumberFormatException();
    } catch (NumberFormatException nfe) {
        CyLogger.getLogger().error("Failed to parse parameter count in spill string", nfe);
        return events;
    }

    compNames = Arrays.copyOfRange(compValues, 1, n + 1);

    // Match names in spill string to columns in parameter lists
    compDataStart = Arrays.asList(this.channelShortname).indexOf(compNames[0]);
    if (compDataStart < 0) {
        CyLogger.getLogger().error("Failed to match channel " + compNames[0] + " to parameter in file");
        return events; // Failure match spill string names to channels
    }
    for (int i = 0; i < n; i++) {
        if (!compNames[i].equals(this.channelShortname[compDataStart + i])) {
            CyLogger.getLogger().error("Spill channel are not continguous parameters in file");
            return events; // Spill string columns not in order
        }
    }

    // Extract actual compensation data
    compData = Arrays.copyOfRange(compValues, n + 1, compValues.length);
    if (compData.length != (n * n))
        return events;

    /**
     * Populate the compensation matrix --- The values are stored in
     * row-major order, i.e., the elements in the first row appear
     * first.
     */
    double[][] matrix = new double[n][n];

    // Loop through the array of compensation values
    for (int i = 0; i < n; i++) {
        for (int j = 0; j < n; j++) {
            try {
                matrix[i][j] = Double.parseDouble(compData[i * n + j]);
            } catch (NumberFormatException nfe) {
                // Set default value If a NumberFormatException occurred
                matrix[i][j] = 0.0d;
            }
        }
    }

    // Compute the inverse of the compensation data and then apply
    // to data matrix (which is column major). Specifically compute
    // transpose(inverse(<SPILL MATRIX>)) * data
    RealMatrix comp = (new LUDecompositionImpl(new Array2DRowRealMatrix(matrix))).getSolver().getInverse();
    RealMatrix data = new BlockRealMatrix(events);
    data.setSubMatrix( // Update compensated portion of data matrix
            comp.transpose()
                    .multiply(data.getSubMatrix(compDataStart, compDataStart + n - 1, 0,
                            this.getEventCount() - 1))
                    .getData(),

            compDataStart, 0);
    return data.getData();
}

From source file:gephi.spade.panel.fcsFile.java

/**
 * getCompensatedEventList ---/*from www. j a v a2 s.c o m*/
 * <p>
 * Returns the event list compensated by the SPILL matrix.
 * <p>
 *
 * @return array of double arrays containing the events.
 */
public double[][] getCompensatedEventList() {
    double[][] events = this.getEventList();
    if (events.length != this.getNumChannels())
        return events; // Unable to extract the underlying events

    // Convert the SPILL string to a compensation matrix
    String compString = this.getSpillString();
    if (compString == null)
        return events; // No compensation, just return the events

    // Split the compensation string into its values
    //
    // The basic structure for SPILL* is:
    // $SPILLOVER/n,string1,string2,...,f1,f2,f3,f4,.../

    String[] compValues = compString.split(",");
    String[] compNames = null;
    String[] compData = null;
    int compDataStart = 0;

    int n = 0;
    try {
        // Try to parse the number of acquisition parameters
        n = Integer.parseInt(compValues[0]);
        if (n <= 0 || n > this.parameters)
            throw new NumberFormatException();
    } catch (NumberFormatException nfe) {
        //CyLogger.getLogger().error("Failed to parse parameter count in spill string",nfe);
        return events;
    }

    compNames = Arrays.copyOfRange(compValues, 1, n + 1);

    // Match names in spill string to columns in parameter lists
    compDataStart = Arrays.asList(this.channelShortname).indexOf(compNames[0]);
    if (compDataStart < 0) {
        //CyLogger.getLogger().error("Failed to match channel "+compNames[0]+" to parameter in file");
        return events; // Failure match spill string names to channels
    }
    for (int i = 0; i < n; i++) {
        if (!compNames[i].equals(this.channelShortname[compDataStart + i])) {
            //CyLogger.getLogger().error("Spill channel are not continguous parameters in file");
            return events; // Spill string columns not in order
        }
    }

    // Extract actual compensation data
    compData = Arrays.copyOfRange(compValues, n + 1, compValues.length);
    if (compData.length != (n * n))
        return events;

    /**
     * Populate the compensation matrix --- The values are stored in
     * row-major order, i.e., the elements in the first row appear
     * first.
     */
    double[][] matrix = new double[n][n];

    // Loop through the array of compensation values
    for (int i = 0; i < n; i++) {
        for (int j = 0; j < n; j++) {
            try {
                matrix[i][j] = Double.parseDouble(compData[i * n + j]);
            } catch (NumberFormatException nfe) {
                // Set default value If a NumberFormatException occurred
                matrix[i][j] = 0.0d;
            }
        }
    }

    // Compute the inverse of the compensation data and then apply
    // to data matrix (which is column major). Specifically compute
    // transpose(inverse(<SPILL MATRIX>)) * data
    RealMatrix comp = (new LUDecompositionImpl(new Array2DRowRealMatrix(matrix))).getSolver().getInverse();
    RealMatrix data = new BlockRealMatrix(events);
    data.setSubMatrix( // Update compensated portion of data matrix
            comp.transpose()
                    .multiply(data.getSubMatrix(compDataStart, compDataStart + n - 1, 0,
                            this.getEventCount() - 1))
                    .getData(),

            compDataStart, 0);
    return data.getData();
}

From source file:org.mitre.math.linear.RealMatrixUtils.java

/**
 * Normalizes a matrix into its Z-scores.
 *///from www. j a v a2  s.c  o  m
public void normalizeMatrix(RealMatrix matrix) {
    //int features = matrix.getRowDimension();
    //int samples = matrix.getColumnDimension();
    int m = matrix.getRowDimension();
    int n = matrix.getColumnDimension();

    // Normalize each row
    for (int i = 0; i < m; i++) {
        // would be easier/quicker if we can get the whole row as an array (or vector)
        RealMatrix subMatrix = matrix.getSubMatrix(i, i, 0, n - 1); // n - 1 for 0 based indexing
        double sum = norm1(subMatrix) / n;
        // minusEquals (subtractEquals) subMatrix
        for (int j = 0; j < n; j++) {
            subMatrix.addToEntry(0, j, -1.0 * sum);
        }
        double std = norm1(arrayTimes(subMatrix, subMatrix)) / n;
        timesEquals(subMatrix, 1.0 / std);

        //setSubMatrix(i,i, 0, samples -1, m)
        for (int j = 0; j < n; j++) {
            matrix.setEntry(i, j, subMatrix.getEntry(0, j));
        }
    }

    // Normalize each feature
    //for (int i = 0; i < features; i++) {
    //Matrix m = matrix.getMatrix(i, i, 0, samples - 1);
    //double sum = m.norm1() / samples;
    //double sum = norm1(m) / samples;
    //m.minusEquals(new Matrix(1, samples, sum));
    //double std = m.arrayTimes(m).norm1() / samples;
    //m.times(1.0 / std);
    //matrix.setMatrix(i, i, 0, samples - 1, m);

    //}
}