protected double calculateLogFieldLikelihood() {
if (!intervalsKnown) {
// intervalsKnown -> false when handleModelChanged event occurs in super.
wrapSetupIntervals();
setupGMRFWeights();
intervalsKnown = true;
}
double currentLike = 0;
DenseVector diagonal1 = new DenseVector(fieldLength);
DenseVector currentGamma = new DenseVector(popSizeParameter.getParameterValues());
SymmTridiagMatrix currentQ =
getScaledWeightMatrix(
precisionParameter.getParameterValue(0), lambdaParameter.getParameterValue(0));
currentQ.mult(currentGamma, diagonal1);
// currentLike += 0.5 * logGeneralizedDeterminant(currentQ) - 0.5 *
// currentGamma.dot(diagonal1);
currentLike +=
0.5 * (fieldLength - 1) * Math.log(precisionParameter.getParameterValue(0))
- 0.5 * currentGamma.dot(diagonal1);
if (lambdaParameter.getParameterValue(0) == 1) {
currentLike -= (fieldLength - 1) / 2.0 * LOG_TWO_TIMES_PI;
} else {
currentLike -= fieldLength / 2.0 * LOG_TWO_TIMES_PI;
}
return currentLike;
}
public static void main(String... args) {
Collection<Double> goldVec = VSMUtil.getGoldStandard().values();
DenseVector goldVector = new DenseVector(goldVec.size());
int idx = 0;
for (double gold : goldVec) {
goldVector.add(idx, gold);
idx++;
}
System.out.println(goldVector.size());
}
@Override
public double get(int row, int col) {
if (row == col) {
return diag.get(row);
}
return -backing.get(row, col);
}
protected void computeDegrees(Organism organism) throws ApplicationException {
Collection<Collection<Long>> groupedNetworks = getAllNetworks(organism);
degreesInOrganism.zero();
DenseVector degreesInNetwork = new DenseVector(numGenes);
for (Collection<Long> networks : groupedNetworks) {
for (long networkId : networks) {
degreesInNetwork.zero();
Network network = cache.getNetwork(Data.CORE, organism.getId(), networkId);
SymMatrix networkData = network.getData();
networkData.columnSums(degreesInNetwork.getData());
degreesInOrganism.add(degreesInNetwork);
}
}
}
@Override
// todo: update to use vector ops
public void multAdd(double alpha, double[] x, double[] y) {
int l = backing.numRows();
for (int i = 0; i < l; i++) {
y[i] = -y[i];
}
backing.multAdd(x, y);
for (int i = 0; i < l; i++) {
y[i] = diag.get(i) * x[i] - y[i];
}
}
private void computeDiag() {
diag = new DenseVector(backing.numRows());
backing.rowSums(diag.getData());
MatrixUtils.add(diag, 1d);
}
