Taung Paw tar Shan Koe Mee

About Taung Paw tar Shan Koe Mee

Among the width × height cells on the field, n_mines cells contain a mine. The goal of the agent is to identify these mines.

The goal state is to explore all cells that contain no mine, and to flag all cells contain a mine.

On every turn, the player selects a cell and either explores it or flags it as mine. When a cell that contains a mine is explored, the game is lost. When the cell contains no mine, the agent learns how many adjacent cells (0 ≤ N ≤ 8) contain mines. (When N = 0, the adjacent cells are explored immediately.) When a cell is known to contain a mine, it can be flagged as such, as a mental note not to consider it for future exploration. Explored cells are visualized as their number N, or a mine symbol when they contain a mine (in which case the game is lost), and flagged cells are marked by a red flag.

The game is won when all non-mine cells are explored and all mine cells are flagged. The game is lost when the agent explores a mine cell.