50 likes | 188 Views
This paper explores advanced strategies for playing Modified Chinese Checkers, examining the limitations of traditional tree search algorithms like BFS, DFS, and Backtracking due to the static nature of trees and the branching factor. It discusses the application of the A* Algorithm for heuristic search and introduces improvements through IDA* and SMA* to optimize memory usage and reduce computation time. Additionally, it considers the role of the grey marble as a potential game-changer and proposes innovative heuristic enhancements for efficient decision-making during gameplay.
E N D
Modified Chinese Checkers Sabrina Wang, Ben Perlmutter, Melinda Lim
Why Trees are Bad • Trees are static. • BFS – too many nodes/branches, not enough processing power • DFS – The depth isn’t always known. • Backtracking – Intermediate steps depend on opposition moves, too many nodes/branches to explore • Iterative Deepening – depends on the oppositions moves which are variable and can’t be predicted
Generic Best First Search • A* is a great option. • Challenge: finding an accurate g(n)/h(n) heuristic. • g(n) = combination of: y distance to reach node n and number of pieces in path • h(n) = combination of: y distance to reach the end from node n and number of pieces in path (based on expected value)
Improvements to A* • IDA* will help us reduce the number of h(n) values to be calculated. • What is the cut off? • Should it be different at different points of the game? • SMA* - do we need to consider memory issues?
Other Considerations • Grey Marble • Is it just another piece? • Or does it have other implications? • Do we treat it as a “special” case in the algorithm? • Other heuristic improvements to g(n) h(n) • h(n) – determine where the grey ball may be placed • “Blocking” • How to act at different stages in the game.