/**
* Function to compute Cholesky decomposition of the given input matrix. The input must be a real
* symmetric positive-definite matrix.
*
* @param in commons-math3 Array2DRowRealMatrix
* @return matrix block
* @throws DMLRuntimeException if DMLRuntimeException occurs
*/
private static MatrixBlock computeCholesky(Array2DRowRealMatrix in) throws DMLRuntimeException {
if (!in.isSquare())
throw new DMLRuntimeException(
"Input to cholesky() must be square matrix -- given: a "
+ in.getRowDimension()
+ "x"
+ in.getColumnDimension()
+ " matrix.");
CholeskyDecomposition cholesky = new CholeskyDecomposition(in);
RealMatrix rmL = cholesky.getL();
return DataConverter.convertToMatrixBlock(rmL.getData());
}
/**
* Returns the result of postmultiplying {@code this} by {@code m}.
*
* @param m matrix to postmultiply by
* @return {@code this * m}
* @throws DimensionMismatchException if {@code columnDimension(this) != rowDimension(m)}
*/
public Array2DRowRealMatrix multiply(final Array2DRowRealMatrix m)
throws DimensionMismatchException {
MatrixUtils.checkMultiplicationCompatible(this, m);
final int nRows = this.getRowDimension();
final int nCols = m.getColumnDimension();
final int nSum = this.getColumnDimension();
final double[][] outData = new double[nRows][nCols];
// Will hold a column of "m".
final double[] mCol = new double[nSum];
final double[][] mData = m.data;
// Multiply.
for (int col = 0; col < nCols; col++) {
// Copy all elements of column "col" of "m" so that
// will be in contiguous memory.
for (int mRow = 0; mRow < nSum; mRow++) {
mCol[mRow] = mData[mRow][col];
}
for (int row = 0; row < nRows; row++) {
final double[] dataRow = data[row];
double sum = 0;
for (int i = 0; i < nSum; i++) {
sum += dataRow[i] * mCol[i];
}
outData[row][col] = sum;
}
}
return new Array2DRowRealMatrix(outData, false);
}
/**
* Function to compute matrix inverse via matrix decomposition.
*
* @param in commons-math3 Array2DRowRealMatrix
* @return matrix block
* @throws DMLRuntimeException if DMLRuntimeException occurs
*/
private static MatrixBlock computeMatrixInverse(Array2DRowRealMatrix in)
throws DMLRuntimeException {
if (!in.isSquare())
throw new DMLRuntimeException(
"Input to inv() must be square matrix -- given: a "
+ in.getRowDimension()
+ "x"
+ in.getColumnDimension()
+ " matrix.");
QRDecomposition qrdecompose = new QRDecomposition(in);
DecompositionSolver solver = qrdecompose.getSolver();
RealMatrix inverseMatrix = solver.getInverse();
return DataConverter.convertToMatrixBlock(inverseMatrix.getData());
}
public void outputMatrix(String file) throws Exception {
int row = lpsolver.getNrows() + 1;
int col = lpsolver.getNcolumns() + 1;
ArrayRealVector f = new ArrayRealVector(col - 1);
for (int j = 1; j < col; j++) {
f.setEntry(j - 1, lpsolver.getMat(0, j));
}
Array2DRowRealMatrix Aeq = new Array2DRowRealMatrix(numOfBus, col - 1);
ArrayRealVector beq = new ArrayRealVector(numOfBus);
for (int i = 1; i < numOfBus + 1; i++) {
for (int j = 1; j < col; j++) {
Aeq.setEntry(i - 1, j - 1, lpsolver.getMat(i, j));
}
beq.setEntry(i - 1, lpsolver.getRh(i));
}
Array2DRowRealMatrix Aiq = new Array2DRowRealMatrix(row - numOfBus - 1, col - 1);
ArrayRealVector biq = new ArrayRealVector(row - numOfBus - 1);
int cnt = 0;
for (int i = numOfBus + 1; i < row; i++) {
for (int j = 1; j < col; j++) {
Aiq.setEntry(cnt, j - 1, lpsolver.getMat(i, j));
}
biq.setEntry(cnt, lpsolver.getRh(i));
cnt++;
}
ArrayRealVector lb = new ArrayRealVector(col - 1);
for (int i = 1; i < col - 1; i++) {
lb.setEntry(i - 1, lpsolver.getLowbo(i));
}
ArrayRealVector ub = new ArrayRealVector(col - 1);
for (int i = 1; i < col; i++) {
ub.setEntry(i - 1, lpsolver.getUpbo(i));
}
writeMatlabInputFile(file, f, Aeq, beq, Aiq, biq, ub, lb);
}
/* 133: */
/* 134: */ public void updateHighOrderDerivativesPhase2(
double[] start, double[] end, Array2DRowRealMatrix highOrder)
/* 135: */ {
/* 136:330 */ double[][] data = highOrder.getDataRef();
/* 137:331 */ for (int i = 0; i < data.length; i++)
/* 138: */ {
/* 139:332 */ double[] dataI = data[i];
/* 140:333 */ double c1I = this.c1[i];
/* 141:334 */ for (int j = 0; j < dataI.length; j++) {
/* 142:335 */ dataI[j] += c1I * (start[j] - end[j]);
/* 143: */ }
/* 144: */ }
/* 145: */ }
@Override
public void process() {
// Get image dimensions, store ICA dimensions (number of frames).
int[] dim = ip.getDimensions();
dimensions = dim[4];
IJ.showStatus("ICA: reformatting data matrix...");
// Compute number of real voxels to be used
int n = 0;
for (@SuppressWarnings("unused") Voxel v : ip) {
n++;
}
// Create new array and fill it with the image data
double[][] image_data = new double[n][dimensions];
int i = 0;
for (Voxel v : ip) {
image_data[i++] = v.tac;
}
// Transpose the data matrix
Array2DRowRealMatrix temp = new Array2DRowRealMatrix(image_data, false);
image_data = temp.transpose().getData();
temp = null;
System.gc();
// Perform the ICA computation
IJ.showStatus("ICA: performing source separation " + "(may take some time)...");
FastICA fi = null;
try {
fi = new FastICA(image_data, ican);
} catch (FastICAException e) {
System.out.println(e.getLocalizedMessage());
}
IJ.showStatus("ICA computed, reformatting results...");
// Get projections on each dimension
double[][] vectors = fi.getICVectors();
// Get independent signals (to be offered as additional information)
// It is transposed to print the columns.
double[][] sep = fi.getSeparatingMatrix();
RealMatrix sources = new Array2DRowRealMatrix(sep);
sources = sources.transpose();
Array2DRowRealMatrix result = new Array2DRowRealMatrix(vectors, false);
// If the ICA image is to be shown, create a new image with
// result.getRowDimension() frames and the original number of
// x, y, z dimensions
if (showICA) {
ImagePlus ICA_image = RealMatrix2IJ(result, dim, this.ip, "ICA image");
ICA_image.show();
}
// Please note: this is somehow incorrect. As the clustering model
// that we are following needs one voxel -> one cluster, this step
// below assigns each voxel to the principal component with the
// maximum weight for its particular kinetics. In "real life", the
// resulting images would contain the contribution of that component
// in all voxels, but for segmentation purposes this approach is
// chosen.
int column_index = 0;
for (Voxel v : ip) {
double[] projection = result.getColumn(column_index++);
// Every Voxel belongs to the maximum index of its projected TAC
int max = getMaxIndex(projection) + 1;
addTACtoCluster(v, max);
}
// Fill in the additionalInfo array.
additionalInfo = new String[2];
additionalInfo[0] = "ica_sources";
StringBuilder sb = new StringBuilder();
int rows = sources.getRowDimension();
for (int j = 0; j < rows; j++) {
double[] row = sources.getRow(j);
sb.append(Arrays.toString(row));
sb.append("\n");
}
// Remove brackets
String tempstr = sb.toString();
tempstr = tempstr.replace("[", "");
tempstr = tempstr.replace("]", "");
additionalInfo[1] = tempstr;
}
