private void PrecomputeAuralProperties(Matrix <double> K, Matrix <double> M)
{
// Call to the numpy server to compute evd
Numpy.EvdResult evd = Numpy.LinAlg.Eigh(K);
var lambdas = evd.EigenValues;
var mgains = evd.EigenVectors;
// Initialize output structurs
List <zdouble> wp = new List <zdouble>();
List <zdouble> wm = new List <zdouble>();
List <double> mass = new List <double>();
List <Vector <double> > aggregatedColumns
= new List <Vector <double> >();
for (int i = 0; i < lambdas.Count;)
{
zdouble probe = GetOmega(lambdas[i], 1);
double _freq = GetFreqFrom(probe);
if (_freq < MIN_FREQ || _freq > MAX_FREQ)
{
i++;
continue;
}
wp.Add(probe);
wm.Add(GetOmega(lambdas[i], -1));
mass.Add(M[i, i]);
i = SumColumnsUntilDlc(i, _freq, mgains, lambdas, aggregatedColumns);
}
_omegaPlus = VB.DenseOfEnumerable(wp);
_omegaMinus = VB.DenseOfEnumerable(wm);
_massMatrix = MB.DenseOfDiagonalArray(mass.ToArray());
_gainMatrix = MB.DenseOfColumns(aggregatedColumns);
// call numpy server to compute pseudo-inverse
_gainMatrixT = Numpy.LinAlg.Pinv(_gainMatrix);
}