Andrew Lee

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2023-195

July 21, 2023

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2023/EECS-2023-195.pdf

This report explores the development of the programmatic solver for a traditional two-player, perfect information, finite-length strategy game from South Korea called Five Field Kono. The first goal of this research paper is to describe the application programming interfaces (API) available in the GAMESMAN software infrastructure used in the GameCrafters Computational Game Theory research group. Utilizing their APIs, I will identify some of the bottlenecks in the current infrastructure and suggest appropriate solutions to them.

As the title suggests, the second goal is to show how simple solutions can be brought to complex problems. I will demonstrate how one huge problem can be broken down into small individual parts using game-solving theories I learned in the research group, from canonical position calculation for symmetry discrimination to efficient hashing. In terms of reachable position count, Five Field Kono is the largest game that cannot be non-trivially separated into tiers that has ever been strongly solved in GamesCrafters. We hope that our work paves the way for additional optimizations that can be incorporated into the current GameCrafters infrastructure and support games with similar traits.

---

BibTeX citation:

@techreport{Lee:EECS-2023-195,
    Author= {Lee, Andrew},
    Editor= {Garcia, Dan},
    Title= {Five Field Kono Solver: Simplicity Brought to Solving Complexity},
    Year= {2023},
    Month= {Jul},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2023/EECS-2023-195.html},
    Number= {UCB/EECS-2023-195},
    Abstract= {This report explores the development of the programmatic solver for a traditional two-player, perfect information, finite-length strategy game from South Korea called Five Field Kono. The first goal of this research paper is to describe the application programming interfaces (API) available in the GAMESMAN software infrastructure used in the GameCrafters Computational Game Theory research group. Utilizing their APIs, I will identify some of the bottlenecks in the current infrastructure and suggest appropriate solutions to them.

As the title suggests, the second goal is to show how simple solutions can be brought to complex problems. I will demonstrate how one huge problem can be broken down into small individual parts using game-solving theories I learned in the research group, from canonical position calculation for symmetry discrimination to efficient hashing. In terms of reachable position count, Five Field Kono is the largest game that cannot be non-trivially separated into tiers that has ever been strongly solved in GamesCrafters. We hope that our work paves the way for additional optimizations that can be incorporated into the current GameCrafters infrastructure and support games with similar traits.},
}

EndNote citation:

%0 Report
%A Lee, Andrew 
%E Garcia, Dan 
%T Five Field Kono Solver: Simplicity Brought to Solving Complexity
%I EECS Department, University of California, Berkeley
%D 2023
%8 July 21
%@ UCB/EECS-2023-195
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2023/EECS-2023-195.html
%F Lee:EECS-2023-195