tic-tac-toe-RL

Problem Statement

To simulate a tic-tac-toe game between a reinforcement-learning agent and a random agent with varying numbers of training games, with different policies for learning. And, to see the trade-off between the exploration and expoitation policy by varying the number of grid-size.

How to Run

Installation

• Clone the directory from github,and navigate to the directory tic-tac-toe-rl

        git clone https://github.com/abhisaini/tic-tac-toe-rl
  

• Create Directories to store your outputs by running following command

        bash prepare.sh
  

Running the game

• If u have python 2.7 installed and have C++11,Run following command (in tic-tac-toe-rl directory) :

       g++ filename.cpp -std=c++11 -I/usr/include/python2.7 -lpython2.7
  

• For python 3.6

        g++ filename.cpp -I/usr/include/python3.6 -lpython3.6
  

• For python 4

        g++ example.cpp -I/usr/local/include/fancy-python4 -L/usr/local/lib -lfancy-python4            
  

About Files

• Here main.cpp contains a simple tic-tac-toe learner
• unordered_map/try_map.cpp is the new updated file which stores states in a optimal way and consumes a lesser amount of time as compared to main.cpp
• It it recommended to use that inorder to train .