chandler37 / gobackgammon

golang backgammon player

gobackgammon

golang is a perfect language for a program that does Monte Carlo simulations of backgammon positions to determine the expected value of points gained for a player. That's because (1) it can easily utilize all available CPU hardware for the computation and (2) the generated code is not as fast as C/C++ but significantly faster than Python/Ruby.

See also the webapp built on top of this, https://github.com/chandler37/gobackgammond

Copyright

Copyright 2018 David L. Chandler

See the LICENSE file in this directory.

FAQ

Q: Is it any good? A: Oh yes.

What's next

• git grep TODO will show you some future directions, especially the use of Monte-Carlo simulations to improve the AI.

• I'd like to use https://github.com/jamiealquiza/tachymeter to measure how long it takes to have ai.MakePlayerConservative(0, nil) play a game against itself. We have make bench right now to give us an idea of performance, but it seems like the variance is surprisingly high.

What do I type?

Don't type anything; instead, start with https://github.com/chandler37/gobackgammond which has a webserver to help visualize the code in this module.

But if you really want to modify this module:

• read ./Makefile
• install Go
  • MacOS? Run brew install go (see https://brew.sh/) to install golang. If your golang is out of date, brew upgrade go.
• run 'make check bench`
• publish a new git tag for a new version, say v1.m.p
• change https://github.com/chandler37/gobackgammond files go.mod and go.sum to reference v1.m.p and run make clean srv