LSSTDESC/TJPCov

CovarianceCalculator.get_covariance fails

Closed this issue · 0 comments

#62 introduced a bug. Now CovarianceCalculator fails to compute the covariance because covariance_from_name does not find the classes.

A dirty hack is to import the needed class together with the CovarianceCalculator.

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[2], line 2
      1 # You can get the total covariance:
----> 2 cov = cc.get_covariance();
      4 plt.imshow(np.log10(np.abs(cov)))
      5 plt.colorbar()

File ~/.local/lib/python3.10/site-packages/tjpcov/covariance_calculator.py:107, in CovarianceCalculator.get_covariance(self)
    100 """Return the covariance with all the contributions added up.
    101 
    102 Returns:
    103     array: Final covariance with all the requested contributions added
    104     up.
    105 """
    106 if self.cov_total is None:
--> 107     cov_terms = self.get_covariance_terms()
    109     # No need to do it only for rank == 0 since all Builder processes
    110     # have self.cov well defined and this is quite unexpensive.
    111     self.cov_total = sum(cov_terms.values())

File ~/.local/lib/python3.10/site-packages/tjpcov/covariance_calculator.py:128, in CovarianceCalculator.get_covariance_terms(self)
    116 """Return a dictionary with the covariance contributions.
    117 
    118 The dictionary has keys the covariace types and values their covariance
   (...)
    125     covariance contributions.
    126 """
    127 if self.cov_terms is None:
--> 128     cov_classes = self.get_covariance_classes()
    130     cov_terms = {}
    131     for ctype, cov_dict in cov_classes.items():

File ~/.local/lib/python3.10/site-packages/tjpcov/covariance_calculator.py:64, in CovarianceCalculator.get_covariance_classes(self)
     62 space_types = []
     63 for covi in cov_tbc:
---> 64     covi = covariance_from_name(covi)
     65     # Check the cov_type has not been already requested for the
     66     # same tracer types (e.g. two Gaussian contributions for ClxCl)
     67     if (covi.cov_type in cov_classes) and (
     68         covi._tracer_types in cov_classes[covi.cov_type]
     69     ):

File ~/.local/lib/python3.10/site-packages/tjpcov/__init__.py:27, in covariance_from_name(name)
     25     return mappers[name]
     26 else:
---> 27     raise ValueError(f"Unknown covariance {name}")

ValueError: Unknown covariance FourierGaussianNmt