PyMCT

This is a simple implement for Manto Carlo Tree Search algorithm.

Install

pip install --upgrade PyMCT

Example

This is a example for searching a randomly generated tree, each transcation to a new state is given with random reward.

Import moudles:

from random import randint
from PyMCT.MCT import MCTNode, MCTS, State

create the test case:

class Test:
    _root:MCTNode
    _MCTS:MCTS
    
    def __init__(self, root_state:int, c:int=2, max_iter:int=10):       
        root_state = State(root_state)
        self._root = MCTNode(state=root_state)
        #Set the serach with max iteration and max tree heights
        self._MCTS = MCTS(self.root,c=c,max_iter=max_iter, max_height=2, debug=True)
        
    #Always return a random reward. Note that the function must take one MCTNode as argument, and return a value.
    def reward_func(slef, node:MCTNode):
        return randint(0, 10)
    
    #Randomly expand the tree with new node. Note that the function must take one MCTNode as argument, and return a list of new states.
    def discover_func(self, node:MCTNode):
        new_states = list()
        for i in range(randint(1,10)):
            new_states.append(State(i))
        return new_states

    def run(self):
        #Pass in the reward and discover function, start the algorithm!
        self.MCTS.iterate(self.reward_func, self.discover_func)
        
        #Find the optimal path.
        self.MCTS.find_optimal_path()
        
        #Display the tree, Note that if no tag is given to MCTNode, a random tag will be generated and display here.
        self.MCTS.render_tree()
        
        #Print the oprimal path. This is the list of MCTNodes.
        print(self.MCTS.optimal_path)
    
    @property
    def root(self):
        return self._root
    
    @property
    def MCTS(self):
        return self._MCTS

run the test:

if __name__ == '__main__':
    test = Test(0)
    test.run()