Reward terms in Pathmind Simulations

[slinlee]

@maxpumperla - when creating Pathmind Simulations, you brought up some good points about the developer experience of reward variables. (learning before/after etc). I like the idea of a dictionary of reward terms that users can select/unselect in the webapp for experimentation.

I just thought of one question though. In AnyLogic, we also use the reward terms to track metrics in the simulation that a user cares about. We would lose most of this functionality with the current implementation of Pathmind Simulation reward terms, right?

It might work out for some stats, esp if the user is trying to maximize them like "widgets produced". But a reward term a user wants to reduce like "accidents occurred" would be negative, right?

If that's the case, should we have a separate function that provides stats directly from the simulation?

[slinlee]

Wait, nvm, I answered my own question by looking again at the code https://github.com/PathmindAI/pathmind-api/blob/c4e7b804fe192a21e5d70ec44ec77aa11af8b7b4/pathmind/simulation.py#L90

This function is meant for what I described above for collecting stats to show in the UI, right?