Format Infinity error

Question

package pfe;
Hi,
I am asking to write a class to perturb numerical data. To do that, I used the JAMA package ( http://math.nist.gov/javanumerics/jama/ ), the BLOG engine ( http://people.csail.mit.edu/milch/blog/index.html ), the class Covariance of the package "org.apache.commons.math" ( http://commons.apache.org/math/apidocs/org/apache/commons/math/stat/correlation/Covariance.html ) and the class StatUtils ( http://math.nist.gov/javanumerics/jama/doc/ )
When executing my program, I had an Infinity error ! Here is my code:
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.text.DecimalFormat;
import Jama.Matrix;
import Jama.CholeskyDecomposition;
import blog.distrib.MultivarGaussian;
import org.apache.commons.math.stat.StatUtils;
import org.apache.commons.math.stat.correlation.Covariance;
public class GADP_Test {
//The perturbation level.
static public double d = 0;
//n_X is the number of confidential attributes.
static public int n_X() {
return nX;
//n_X is the number of non-confidential attributes.
static public int n_S() {
return nS;
//n_X is the number of all (confidential and non-confidential) attributes.
static public int n_U() {
return nX + nS;
static public Matrix U;
static public Matrix mu_U;
//The covariance matrix sigma_UU.
static public Matrix sigma_UU; // = new Matrix(_sigma_UU);
//The database rows to be perturbed C_U.
static public Matrix C_U;
//The perturbed vector Y.
static public Matrix Y;
public GADP_Test(Matrix cu,double x) {
GADP_Test.C_U = cu;
GADP_Test.d=x;
static public void GADP_perturbData(Matrix U, double d, Matrix mu, Matrix sigma) {
if (mu == null) {
mu = mu_U;
if (sigma == null) {
sigma = sigma_UU;
Matrix X = U.getMatrix(0, 0, 0, n_X() - 1);
Matrix S = U.getMatrix(0, 0, n_X(), n_U() - 1);
Matrix sigma_XX = sigma.getMatrix(0, n_X() - 1, 0, n_X() - 1);
Matrix sigma_YY = sigma_XX;
Matrix sigma_SX = sigma.getMatrix(n_X(), n_U() - 1, 0, n_X() - 1);
Matrix sigma_SY = sigma_SX;
Matrix sigma_XS = sigma_SX.transpose();
Matrix sigma_YS = sigma_XS;
Matrix sigma_YX = sigma_XX.timesEquals(d); // sigma_YX = d * sigma_XX
Matrix sigma_XY = sigma_YX.transpose();
Matrix sigma_YU = concat(sigmaYX, sigma_YS);
Matrix sigma_UY = sigma_YU.transpose();
mu = mu.transpose();     // make it a row vector
Matrix mu_X = mu.getMatrix(0, n_X() - 1, 0, 0);
// sigma_YU . (sigma_UU)^-1
Matrix sigma_YU_UU_inv = sigma_YU.times(sigma_UU.inverse());
// Loop computing the perturbations of the C_n's (eqs. 11 & 12):
for (int n = 0; n < C_U.getRowDimension(); n++) {
Matrix C_n = C_U.getMatrix(n, n, 0, C_U.getColumnDimension() - 1);
println("Original data: C_" + n + " = ");
C_n.print(_format, _width);
Matrix E_Y_C_n = mu_X.plus(sigma_YU_UU_inv.times(C_n.minus(mu_U).transpose()));
Matrix S_Y_C_n = sigma_YY.minus(sigma_YU_UU_inv.times(sigma_UY));
Matrix Y_norm = Matrix.random(n_X(), 1);
println("Y_norm" + n + " = ");
Y_norm.print(_format, _width);
Matrix chol = new CholeskyDecomposition(S_Y_C_n).getL();
println("chol" + n + " = ");
chol.print(_format, _width);
Y = E_Y_C_n.plus(chol.transpose().times(Y_norm)).transpose();
println("Perturbed data: Y" + n + " = ");
Y.print(_format, _width);
double[][] ff=Y.getArray();
System.out.println(ff[0][0]);
static private double[] muU;
static private double[][] sigmaUU;
//n_X is the number of confidential attributes (the dimension of the vector X).
static private int nX = 3;
//n_S is the number of non-confidential attributes (the dimension of the vector S).
static private int nS = 2;
//This takes two matrices with equal number of rows and returns the matrix obtained by juxtaposing the 1st and the 2nd.
static private Matrix _concat(Matrix A, Matrix B) {
double[][] a = A.getArray();
double[][] b = B.getArray();
int n = A.getRowDimension();
int m1 = A.getColumnDimension();
int m2 = B.getColumnDimension();
double[][] c = new double[n][m1 + m2];
for (int i = 0; i < n; i++) {
for (int j = 0; j < m1 + m2; j++) {
c[i][j] = j < m1 ? a[i][j] : b[i][j - m1];
return new Matrix(c);
//A method to initialize the printing format parameters.
static private void _initializePrinting(int width, int decimals) {
_width = width;
_decimals = decimals;
_format.setMaximumFractionDigits(decimals);
_format.setMinimumFractionDigits(decimals);
_format.setGroupingUsed(false);
//Width for printing a matrix entry as a double.
static private int _width = 6;
//Number of decimals for printing a matrix entry as a double.
static private int _decimals = 2;
//A DecimalFormat object for formatting the printing of a double.
static private DecimalFormat _format = new DecimalFormat();
//Shorten spelling of print(String).
private static void print(String s) {
System.out.print(s);
//Shorten spelling of println(String).
private static void println(String s) {
System.out.println(s);
//Shorten spelling of println().
private static void println() {
System.out.println();
public void calculateMean() {
double[][] tab = C_U.getArray();
double[] col = new double[tab.length];
double[] lig = new double[tab[0].length];
for (int j = 0; j < tab[0].length; j++) {
for (int i = 0; i < tab.length; i++) {
col[i] = tab[i][j];
lig[j] = StatUtils.mean(col);
double[][] matrix = new double[1][lig.length];
for (int i = 0; i < lig.length; i++) {
matrix[0] = lig[i];
mu_U = new Matrix(matrix);
mu_U.print(_format,_width);
public void calculateCovariance(){
Covariance cov = new Covariance(C_U.getArray());
double[][] reslt = cov.getCovarianceMatrix().getData();
sigma_UU = new Matrix(reslt);
sigma_UU.print(_format,_width);
public void result(){
     _initializePrinting(7,2);
C_U.print(_format,_width);
     GADP_perturbData(mu_U,d,mu_U,sigma_UU);
public static double getD() {
return d;
public static void setD(double d) {
GADP_Test.d = d;
I think the problem comes from the CholeskyDecomposition but I am not sure...
Thank you for your help !

Answer 1

Here is my code again, indented :-)
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.text.DecimalFormat;
import Jama.Matrix;
import Jama.CholeskyDecomposition;
import blog.distrib.MultivarGaussian;
import org.apache.commons.math.stat.StatUtils;
import org.apache.commons.math.stat.correlation.Covariance;
public class GADP_Test {
//The perturbation level.
    static public double d = 0;
//n_X is the number of confidential attributes.
    static public int n_X() {
        return _n_X;
//n_X is the number of non-confidential attributes.
    static public int n_S() {
        return _n_S;
//n_X is the number of all (confidential and non-confidential) attributes.
    static public int n_U() {
        return _n_X + _n_S;
//U is the vector of all (confidential and non-confidential) attributes. It is the concatenation of X, the vector of confidential attributes, and S, the vector of non-confidential attributes
    static public Matrix U;
//The mean vector mu_U.
    static public Matrix mu_U;    // = new Matrix(_mu_U);
//The covariance matrix sigma_UU.
    static public Matrix sigma_UU; // = new Matrix(_sigma_UU);
//The database rows to be perturbed C_U.
    static public Matrix C_U;
//The perturbed vector Y.
    static public Matrix Y;
    public GADP_Test(Matrix cu,double x) {
        GADP_Test.C_U = cu;
        GADP_Test.d=x;
    static public void GADP_perturbData(Matrix U, double d, Matrix mu, Matrix sigma) {
        if (mu == null) {
            mu = mu_U;
        if (sigma == null) {
            sigma = sigma_UU;
        Matrix X = U.getMatrix(0, 0, 0, n_X() - 1);
        Matrix S = U.getMatrix(0, 0, n_X(), n_U() - 1);
        Matrix sigma_YY = sigma_XX;
        Matrix sigma_SX = sigma.getMatrix(n_X(), n_U() - 1, 0, n_X() - 1);
        Matrix sigma_SY = sigma_SX;
        Matrix sigma_XS = sigma_SX.transpose();
        Matrix sigma_YS = sigma_XS;
        Matrix sigma_YX = sigma_XX.timesEquals(d); // sigma_YX = d * sigma_XX
        Matrix sigma_XY = sigma_YX.transpose();
        Matrix sigma_YU = _concat(sigma_YX, sigma_YS);
        Matrix sigma_UY = sigma_YU.transpose();
        mu = mu.transpose();     // make it a row vector
        Matrix mu_X = mu.getMatrix(0, n_X() - 1, 0, 0);
        Matrix sigma_YU_UU_inv = sigma_YU.times(sigma_UU.inverse());
        // Loop computing the perturbations of the C_n's (eqs. 11 & 12):
        for (int n = 0; n < C_U.getRowDimension(); n++) {
            Matrix C_n = C_U.getMatrix(n, n, 0, C_U.getColumnDimension() - 1);
            println("Original data: C_" + n + " = ");
            C_n.print(_format, _width);
            Matrix E_Y_C_n = mu_X.plus(sigma_YU_UU_inv.times(C_n.minus(mu_U).transpose()));
            Matrix S_Y_C_n = sigma_YY.minus(sigma_YU_UU_inv.times(sigma_UY));
            Matrix Y_norm = Matrix.random(n_X(), 1);
            println("Y_norm" + n + " = ");
            Y_norm.print(_format, _width);
            Matrix chol = new CholeskyDecomposition(S_Y_C_n).getL();
            println("chol" + n + " = ");
            chol.print(_format, _width);
            Y = E_Y_C_n.plus(chol.transpose().times(Y_norm)).transpose();
            println("Perturbed data: Y" + n + " = ");
            Y.print(_format, _width);
    static private double[] _mu_U;
    static private double[][] _sigma_UU;
//n_X is the number of confidential attributes (the dimension of the vector X).
    static private int _n_X = 3;
//n_S is the number of non-confidential attributes (the dimension of the vector S).
    static private int _n_S = 2;
//This takes two matrices with equal number of rows and returns the matrix obtained by juxtaposing the 1st and the 2nd.
    static private Matrix _concat(Matrix A, Matrix B) {
        double[][] a = A.getArray();
        double[][] b = B.getArray();
        int n = A.getRowDimension();
        int m1 = A.getColumnDimension();
        int m2 = B.getColumnDimension();
        double[][] c = new double[n][m1 + m2];
        for (int i = 0; i < n; i++) {
            for (int j = 0; j < m1 + m2; j++) {
                c[i][j] = j < m1 ? a[i][j] : b[i][j - m1];
        return new Matrix(c);
//A method to initialize the printing format parameters.
    static private void _initializePrinting(int width, int decimals) {
        _width = width;
        _decimals = decimals;
        _format.setMaximumFractionDigits(decimals);
        _format.setMinimumFractionDigits(decimals);
        _format.setGroupingUsed(false);
//Width for printing a matrix entry as a double.
    static private int _width = 6;
//Number of decimals for printing a matrix entry as a double.
    static private int _decimals = 2;
//A DecimalFormat object for formatting the printing of a double.
    static private DecimalFormat _format = new DecimalFormat();
//Shorten spelling of print(String).
    private static void print(String s) {
        System.out.print(s);
    static private void printMatrix (Matrix M)
      double[][] array = M.getArray();
      for (int i=0; i<array.length; i++)
          println();
          for (int j=0; j<array[0].length; j++)
          print("\t"+array[i][j]);
//Shorten spelling of println(String).
    private static void println(String s) {
        System.out.println(s);
//Shorten spelling of println().
    private static void println() {
        System.out.println();
    public void calculateMean() {
        double[][] tab = C_U.getArray();
        double[] col = new double[tab.length];
        //tab[0] pour connaître le nombre de lignes
        double[] lig = new double[tab[0].length];
        for (int j = 0; j < tab[0].length; j++) {
            for (int i = 0; i < tab.length; i++) {
                col[i] = tab[i][j];
           // lig[j] = StatUtils.geometricMean(col);
            lig[j] = StatUtils.mean(col);
        double[][] matrix = new double[1][lig.length];
        for (int i = 0; i < lig.length; i++) {
            matrix[0] = lig[i];
mu_U = new Matrix(matrix);
mu_U.print(_format,_width);
public void calculateCovariance(){
Covariance cov = new Covariance(C_U.getArray());
double[][] reslt = cov.getCovarianceMatrix().getData();
sigma_UU = new Matrix(reslt);
sigma_UU.print(_format,_width);
public void result(){
     _initializePrinting(7,2);
C_U.print(_format,_width);
     GADP_perturbData(mu_U,d,mu_U,sigma_UU);
public static double getD() {
return d;
public static void setD(double d) {
GADP_Test.d = d;