Bug when loading covariance with several observables for a single observable FastFit

[MJKirk]

Bug introduced in pull request #48 -- there is an edge case that if we using FastFit with only a single observable, but are loading a covariance with multiple observables, we get the whole covariance matrix instead of a single element. Fix by checking the size of the covariance matrix being loaded.

[coveralls]

Coverage decreased (-0.007%) to 93.784% when pulling 856918b50bdcd9ebaa389590014de46c238c1083 on MJKirk:covariance_loading_bugfix into 4d8007d5a0dd456c527e23abe87a9494b6949db4 on flav-io:master.

[DavidMStraub]

Thanks, good point! Actually this also affects the new statistics.likelihood module that will be released with the next version (today, maybe?). This is meant as a replacement for statistics.fits but will probably coexist for a couple of versions. I will fix it there, too.

[MJKirk]

Small thing I noticed while checking my fix, the error message when the loaded covariance doesn't contain all the observables is slightly opaque. https://github.com/flav-io/flavio/blob/4e0b8623624a5d7d36c8df8a4dde32a19b5d9c96/flavio/statistics/fits.py#L591-L595 The string on line 594 doesn't get printed and the code continues to the next line, and so the only error you get is that permutation is not defined.

[DavidMStraub]

Oh, which Exception does it throw then, actually?

[MJKirk]

You get UnboundLocalError The full traceback is

Traceback (most recent call last):
  File "flavio_test.py", line 23, in <module>
    f_single_obs.load_sm_covariance("test_covar.p")
  File "/home/mkirk/Documents/Software/flavio/flavio/statistics/fits.py", line 665, in load_sm_covariance
    self.load_sm_covariance_dict(d=data)
  File "/home/mkirk/Documents/Software/flavio/flavio/statistics/fits.py", line 680, in load_sm_covariance_dict
    assert len(permutation) == len(self.observables), \
UnboundLocalError: local variable 'permutation' referenced before assignment

So nothing is shown about the covariance matrix missing an observable

[DavidMStraub]

Thanks! Nice talk :wink:

[DavidMStraub]

Oh, that fix actually doesn't make sense. Shouldn't code during talks.

[MJKirk]

https://github.com/MJKirk/flavio/commit/e7cbc0441918547a3a531d332d3bb89a593bc544 is what I did in my fork

[DavidMStraub]

Oh exactly, that's what one should do. I can add it.