Something changed in `pytensor > 2.12.3` (and thus `pymc > 5.6.1`) that makes my `pytensor.gradient.grad` call get stuck - any ideas?

Questions v5
23

Thanks - I’ve tried to slice down to 2D… Both f_inv_x and x are (10, 2), and something like:

jac = tg.jacobian(f_inv_x[i, :], x[i, :])

throws DisconnectedInput errors. I mentioned this above… Maybe I’m slicing wrong? Would a subtensor somehow be more appropriate?

The indexing I’ve settled on for now doesn’t seem to hurt… it just looks a little clumsy