Using the Monte Carlo Tree Search Algorithm for a Card Game AI: Simulation

<p>When the&nbsp;<code>Simulation</code>&nbsp;method is called, we use the&nbsp;<code>GameState</code>&nbsp;of the given&nbsp;<code>MCTS_Node</code>&nbsp;to create a hypothetical round of the card game. We simulate each phase in order, before finally simulating the switching of the turn.</p> <p>The&nbsp;<em>Play Phase</em>&nbsp;is the meat of the turn, where the AI is able to play cards to the gameboard. We use an infinite loop to get a random possible action from the game state, breaking the loop if that action is to end the turn. Otherwise we apply it to the game state, and get another action.</p> <p>This then continues to simulate the round, until eventually assigning it a&nbsp;<em>Score</em>. This score is how beneficial the theorised game state is for the AI.</p> <p><a href="https://medium.com/@mattgmez98/using-the-monte-carlo-tree-search-algorithm-for-a-card-game-ai-simulation-40a0218494e4"><strong>Click Here</strong></a></p> <p>&nbsp;</p>
Tags: Monte Carlo