Converting HARK models into Gymnasium environments
Opened this issue · 0 comments
Building on #1295 ...
Once there is a way of specifying a model with equations and a simple generic monte carlo simulation engine...
It should be possible convert HARK models directly into Gymnasium environments:
https://gymnasium.farama.org/tutorials/gymnasium_basics/environment_creation/#sphx-glr-tutorials-gymnasium-basics-environment-creation-py
Gymnasium is a fork of OpenAI's Gym which, unlike Gym, is being maintained by a neutral non-profit third party, the Farama Foundation. The Farama Foundation also has a multi-agent environment, called PettingZoo.
This would make it easier to test hand-crafted solutions against various reinforcement learning based systems, which could be useful for more complex problems, including problems where multiple agents interact through market aggregation.