Have the generalized score functions for GES been tested?

[shahashka]

I want to used the "local_score_CV_general" score function with GES, but I'm running into an issue with a random test case:

import numpy as np
from causallearn.search.ScoreBased.GES import ges
X = np.random.rand(10,10)
result = ges(X, score_func="local_score_CV_general")

With the following error message

line 360, in local_score_cv_general
dists = Q + R - 2 * X * X.T
ValueError: operands could not be broadcast together with shapes (1,100) (10,10) 

I think this issue is data independent (hence the use of a random matrix as a test), but please correct me if I'm wrong. I noticed that in tests/TestGES.py there are no unit tests for the generalized score functions. Is this functionality not yet validated?

Thank you in advance!

[kunwuz]

Thanks for reporting this! We are actively working on this and will let you know as soon as possible. The test for the score function works well, so it's a little bit weird for me why GES with the generalized score fails.

[kunwuz]

While we are working on this, perhaps you could try using version 0.1.3.8 for now, which should work fine.

[kunwuz]

@zhi-yi-huang has fixed this in #223 , could you please try again and let us know if there is any issue?

[shahashka]

Yes, it's working now. Thank you!