Hello,
I was trying to implement a bayesian inference fitting process but my fitting model use quite a lot of parameters. Not fitting parameters, in general, one has to take care of many operations as well as integrals. Long story short, I am using a python package (plasmapy) to deal with this because it’s pretty robust but it uses numpy and astropy units and I am not sure to be able to rewrite the whole model using aesara, and I am not familiar with it anyway…
So, for me, the best would be to work with numbers and not symbolic tensors, to evaluate the model at each step. Is it something like this doable with pymc? Does it make sense?
Thank you for your consideration.
Cheers