Humble C++ lib to compute ICM:
- exactly
- with a Monte-Carlo evaluation for larger numbers
with a big focus on performance.
Extra features:
- evaluate the statistical guarantee along your MC computation
- Python bindings using Boost Python
This code is part of a larger freestyle experiment around valuation functions in poker, that is described in these (french) posts:
- Poker : MTT et ICM #1 - La Question
- Poker : MTT et ICM #2 - Le Calcul Brutal
- Poker : MTT et ICM #3 - Méthode de Monte-Carlo
- Poker : MTT et ICM #4 - Perf Tuning avec du Deep Learning
(hopefully more to come)
As experimented in post #4, training a simple NN on MC-generated data may provide better performance (hundreds of times faster). The related Python code will come in a separate repo a priori.
On my own I'll improve and extend this codebase only on need. If you want to contribute, feel free to submit any pull request or directly contact me via email: [email protected].
|
|- include # Public headers
|- src # Private headers and implementation
|- test # Boost tests
Explore the include folder and you'll find the properly documented few main functions.
Requires:
Some additional information is available in CMakeLists.txt.
Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Public License.
You can use the code freely for any non-commercial use as long as you propagate its license. See the license file.