Schafkopf AI

About

This project is about teaching an AI to play Schafkopf.

It covers topics such as:

• low-level programming in C#
  • usage of memory-efficient structs
  • parallel / branchless techniques and SIMD
  • usage of unsafe pointer arithmetics
• DDD / TDD techniques
  • concepts are properly modeled for low coupling
  • tests allow to gradually enhance the game logic with better performing structures / algorithms
• AI training
  • custom neural network framework
  • several approaches to train agents (wip)

Quickstart

Install .NET Tools

sudo apt-get update && sudo apt-get install -y dotnet-sdk-7.0

Download the Source Code

git clone https://github.com/Bonifatius94/Schafkopf.AI
cd Schafkopf.AI

Build + Run Tests

dotnet test --configuration Release

Run Benchmarks

dotnet run  --configuration Release \
    --project Schafkopf.Lib.Benchmarks/Schafkopf.Lib.Benchmarks.csproj

Run AI Training

dotnet run  --configuration Release \
    --project Schafkopf.Training/Schafkopf.Training.csproj

DISCLAIMER: The training is currently under construction and won't deliver any results yet.

Install Docker (Optional)

sudo apt-get update && sudo apt-get install -y docker.io docker-compose
sudo usermod -aG docker $USER && reboot

Run AI Training (Docker)

docker-compose -f train-compose.yml build && \
    docker-compose -f train-compose.yml up

DISCLAIMER: The training is currently under construction and won't deliver any results yet.