Last active
May 14, 2023 11:46
-
-
Save jessegrabowski/ccda08b8a758f882f5794b8b89ace07a to your computer and use it in GitHub Desktop.
ARMA-GARCH in PyMC
They way you've written it, I don't think you need to collect updates at all in the scan function. This is because you're not actually making any random variables inside the scan, so you don't need to do anything special with the underlying random number generator (this is what collect_default_updates
does.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Alright, sorry to keep pestering you @jessegrabowski but one last question (maybe):
Here is my implementation, but I don't think things are correct for some reason (actually for two reasons: when I try to run with
JAX
backend, it is complaining about numpy generators, and if I don't run withJAX
, it is using a bunch of samplers other than NUTS for different parameters (which should all be able to be sampled by NUTS since they are continuous)).Anyway, here is the relevant part of the model (I dropped the part that computes
obs_mean
):where
obs_data
is the observed data andobs_mean
is the calculated mean from the model (same shape asobs_data
).Can you see anything amiss here?