infer-actively / pymdp

A Python implementation of active inference for Markov Decision Processes
MIT License
420 stars 83 forks source link

Time-dependent preferences #52

Closed conorheins closed 2 years ago

conorheins commented 2 years ago

This PR includes functionality to allow time-dependent C vectors (temporal preferences) to be passed into the control functions (in particular, calc_expected_utility in control.py.

Also, some better docstrings and extra argument checks for observations in the utils.py function process_observation