Gym: How to support Texas hold'em

[SmartAI]

I want to do some research in playing Texas hold'em, how can I use gym?

[wenkesj]

@SmartAI Create a game engine that emulates Texas Hold'em, extend the gym.Env class and maintain the standard gym environment API. It may sound simple at first but multi-agent environments are _non-trivial_ to implement and standardize. You may notice that there is a sparsity of such cases, but as a work around you can encode the multi- state- and action- space as spaces.Tuple of states/actions corresponding to players.

Good luck!

[wenkesj]

@SmartAI fyi I threw something together that demonstrates how a Texas Hold'em gym environment can be done, this is just one of many possibilities

[SmartAI]

@wenkesj Thanks for your suggestion.

[christopherhesse]

Thanks for answering this @wenkesj!

[dickreuter]

Put an implementation together here:
https://github.com/dickreuter/neuron_poker

It also demonstrates some implementations with reinforcement learning.