marmaduke woodman

> I am not sure if pymc3 works with aesara. But you are right, we shouldn't use theano in the future, because it is not maintained. I wasn't able to...

Aesara seems to hard code Python header locations, I couldn't override with `CFLAGS`. On macOS with only the command line tools installed and not XCode, a hacky hack required: ```...

After reverting the modifications to the oscillator model (which you can also include when you actually use the model), I have test failures which occur when evaluating the simulation graph...

> Does this error appear for a specific line of the templates? No, there's no further lines in the traceback, but stepping through the unit test, it happens on `yh.eval()`...

> Alright, thank you. Could it be due to the changes of the oscillator variables of interest? Certainly, I removed those changes because we prefer not to change the historical...

@e-richter do you want to fix this up or shall we close the PR? while a autograd tensor backend for TVB is worthwhile, I have been more focused on Jax...

hi, thanks for your effort. do you intend on completing this or need some help to complete?

I'm onboard for pooch but I'd just like to get #633 done first so that it's easier to add the dependency.

> keep the data replicated in 2 places or replace Zenodo entirely? On a longer term view, it could be of interest to keep the Zenodo for the DOI but...

@liadomide do you think this PR is still relevant, given the work in #691 ? (btw do you have a link to the issue TVB-1999? I cannot find it at...