pystra.dependence.correlation.compute_modified_correlation_matrix#
- compute_modified_correlation_matrix(stochastic_model)[source]#
Compute the modified (Nataf) correlation matrix.
For each pair of non-normal marginals, the physical-space correlation \(\rho_{ij}\) is mapped to the standard-normal-space correlation \(\rho_{0,ij}\) by numerically solving the bi-folded integral equation. For jointly normal pairs the mapping is the identity.
The number of quadrature points is increased adaptively for correlations close to \(\pm 1\) to maintain accuracy.
- Parameters:
stochastic_model (
StochasticModel) – The stochastic model containing marginal distributions and the physical-space correlation matrix.- Returns:
The symmetric modified correlation matrix
Roof shape(n, n)in standard normal space.- Return type:
ndarray