Closed mgrub closed 6 months ago
When using the GUM_iDFT function with a full covariance matrix, the covariance of the output is not symmetric.
GUM_iDFT
from PyDynamic.uncertainty.propagate_DFT import GUM_iDFT import scipy.linalg as scl NX = 6 X = np.arange(NX) + 2 X_cov = scl.toeplitz(0.01 * np.arange(NX)[::-1]) x, x_cov = GUM_iDFT(X, X_cov) print(scl.issymmetric(x_cov, atol=1e-17))
The covariance is expected to be (1) symmetric and (2) should correspond roughly to the Monte Carlo uncertainty evaluation.
When using the
GUM_iDFT
function with a full covariance matrix, the covariance of the output is not symmetric.The covariance is expected to be (1) symmetric and (2) should correspond roughly to the Monte Carlo uncertainty evaluation.