Time varying C vector

[xugami]

Hi,
Thanks for creating this library to bring the Active inference to Python. I've tried to translate matlab script used in Smith et al. (2022): “A Step-by-Step Tutorial on Active Inference Modelling and its Application to Empirical Data”. I wonder if the C-vector can also be time-varied in pymdp?
Thank you
Xu

[conorheins]

Hi @xugami,
Thanks for your interest and this good question: yes, you can input a time dependent C vector. One can do this my simply turning the modality specific C vector (e.g. C[modality_i]) into a matrix, where each column C[modality_i][:,t] encodes the prior preferences over observations within modality_i at timestep t.

I haven't tested this feature very extensively but it should in theory work! Please let me know if you run into any issues.