I would like to attempt to write an approximate HUNLHE solver for short stacks (25bb) for self-study purposes as I am a recreational player. At the moment I am familiar with Johanson's MSc thesis where he describes the basic concepts like abstractions and the vanilla CFRM. BTW I used his count_nl_infosets tool to count how large is a 25bb game with 10/20 blinds.
https://gist.github.com/dabd/5892668I still need to read the more recent papers which I have only briefly skimmed. MCCFRM seems to be the state of the art and there are some recent interesting action abstraction developments.
Browsing the forum archives I found some sample code written by amax which seems to be of excellent quality. Particularly the Rhode Island solver here:
http://poker-ai.org/archive/www.pokerai ... 664#p42664Unfortunately the code does not include the bucket LUT generation. How did he generate the flop and turn LUTs?
Is there a similar public project for NLHE? If people already have these tools or more information on how to build them in the private forums what steps should I take to join one of these forums?
Thanks.