Maximilian
Maximilian
Okay, I have digged deeper into this and found the source of the error: In my example, I am specifying the data as a numpy array with dtype `np.int64`. PyMC3...
So, to me, there are essentially two questions: - Why are integers converted to `int32` using the conversion map from `float64` to `int32`? (The line in `pandas_to_array()` I quoted above)...
Thank you for your reply. I could scale down my observations, I guess, at the (acceptable) cost of losing some precision in the model. However, could you explain what you...
> > However, could you explain what you mean with "we cast to int32 and not to int64 is that the latter can overflow float64 calculations for large integer values"?...
Did @HenrikMettler resolve all your comments, @jakobj ? Could you give this another check?
@HenrikMettler could you fix the tests, rebase on master and squash the commits?
Hi @HenrikMettler , shall we still get this PR merged? Sorry for my inactivity. Can you rebase on `master` and then push to trigger the CI once again?