Construction, observation and knowledge abstraction for go endgames on small boards

Chia-Ming Hsu,Hung-Cheng Lin,Yueh-Ting Chen,Chih-Wen Hsueh,Tsan-sheng Hsu

DOI: https://doi.org/10.1038/s41598-024-57338-x

IF: 4.6

2024-03-24

Scientific Reports

Abstract:A Go endgame database consists of optimal game values and moves for every legal arrangement of no more than S pieces on an N by N board. This paper describes methods for constructing such databases when and . When cycles of plies with lengths greater than 4 are encountered, two rules, one allowing cycles and the other disallowing them, are implemented. Observations and knowledge are obtained for these endgames, which may elucidate the fundamental properties of the popular game Go. First, the optimal game values are different when N is even and odd, regardless of whether the repetition of positions is allowed. When N is odd, the first player can occupy the whole board, while this is not the case when N is even. Second, allowing cycles makes the first and second players equal in strength when N is even, whereas the first player always dominates when N is odd. Using the state-of-the-art open-source deep learning Go engine KataGo to correctly solve a given position as an indicator, factors affecting level of difficulty are found, including the distributions of the optimal game values among all legal plies and the cardinality and values of the true optimal plies. A simple formula is designed that works on more than 10% of the positions so that positions with a given level of difficulty can be found with a high probability.

multidisciplinary sciences

What problem does this paper attempt to address?

### Problems Addressed by the Paper

The paper primarily explores the construction, observation, and knowledge abstraction of Go endgame databases, particularly on small boards (N×N, where 1<N≤5). Specifically:

1. **Construction of Go Endgame Databases**:
- The authors use classical retrograde analysis algorithms to construct Go endgame databases, which contain the optimal game values and best moves for all legal positions.
- By constructing these databases, the authors aim to reveal some fundamental properties of Go.

2. **Handling Cycles**:
- In Go, cycles longer than 4 moves can occur. The paper proposes two rules to handle these cycles: one that allows cycles and one that does not.
- The authors studied the outcomes of games under different rules (allowing or disallowing cycles) and discovered some interesting properties, such as when N is odd, the first player can occupy the entire board; whereas when N is even, this is not possible.

3. **Difficulty Assessment**:
- Using the state-of-the-art open-source deep learning Go engine KataGo to solve specific positions as a metric, the authors studied factors affecting the difficulty of positions, including the distribution of optimal game values among all legal moves and the number and value of truly optimal moves.
- A simple formula is proposed that can find positions with a given difficulty with high probability in over 10% of cases.

In summary, the paper delves into the fundamental properties of Go and its performance under different rules by constructing Go endgame databases and attempts to propose a method for assessing the difficulty of Go positions.