Update Regression Tests about grmpy HOT 3 CLOSED

Does Criterion function mean the likelihood function, an estimation with 1 evaluation or anything else?
If one of the mentioned options is appropriate, there would be the problem that our battery generates dictionaries that are deterministic with probability 0.1. That means that for both options the function would fail to provide neither the log likelihood value nor an estimation (zero division error ...). An easy solution would be to constrain the generating process in the whole loop.
But maybe my i am wrong regarding my starting point, so please let me know what you mean by 'single evaluation of the criterion function'.

from grmpy.

Yes, just the likelihood for a single function evaluation at the starting values. You can also impose the case that we need a non-deterministic initialization file.

from grmpy.

Please only create non-deterministic regression tests so we can avoid the case distinction
if 0.00 in dict_['DIST']['coeff']: stat = np.sum(df.sum()) tests += [(stat, dict_)] else: stat = np.sum(df.sum())

Also, isn't the if statement too restrictive. How about zero covariances?

from grmpy.