Fix and test `gw.fishermatrix.analyze_and_save_to_txt`

[bhuyan-g]

While calculating the fisher parameters for a population of BBH events using gw.fishermatrix.analyze_and_save_to_txt I am encountering the following errors:

matrix_norm = matrix / normalizer
^M 3%|▎ | 7600/300000 [52:03<33:22:58, 2.43it/s]
Traceback (most recent call last):
File "/var/lib/condor/execute/dir_121839/BBH_fisher_posterior_multicat1.py", line 143, in <module> gw.fishermatrix.analyze_and_save_to_txt(network = network,
File "/home/gautam.bhuyan/.conda/envs/igwn_gwfish/lib/python3.10/site-packages/GWFish/modules/fishermatrix.py", line 520, in analyze_and_save_to_txt
detected, network_snr, errors, sky_localization = compute_network_errors(
File "/home/gautam.bhuyan/.conda/envs/igwn_gwfish/lib/python3.10/site-packages/GWFish/modules/fishermatrix.py", line 407, in compute_network_errors
network_fisher_inverse, _ = invertSVD(network_fisher_matrix)
File "/home/gautam.bhuyan/.conda/envs/igwn_gwfish/lib/python3.10/site-packages/GWFish/modules/fishermatrix.py", line 23, in invertSVD
[U, S, Vh] = np.linalg.svd(matrix_norm)
File "/home/gautam.bhuyan/.conda/envs/igwn_gwfish/lib/python3.10/site-packages/numpy/linalg/linalg.py", line 1681, in svd
u, s, vh = gufunc(a, signature=signature, extobj=extobj)
File "/home/gautam.bhuyan/.conda/envs/igwn_gwfish/lib/python3.10/site-packages/numpy/linalg/linalg.py", line 121, in _raise_linalgerror_svd_nonconvergence
raise LinAlgError("SVD did not converge") numpy.linalg.LinAlgError: SVD did not converge

I suspect it has something to do with the InvertSVD function in the source code

[jacopok]

Hi, could you share the parameters of the event for which this fails (or the full script you're running) so we can reproduce it?

This is often due to some "bad" values (e.g. NaN) making it into the Fisher matrix due to the waveform approximant being called outside of its expected range.
The SVD is well-defined even for singular matrices, so I don't expect that to be the issue.