Chess engine: Stockfish 20010919



Stockfish - UCI chess engine, compiled by Joergoster

50-moves rule improvement for transposition table

User "adentong" reported recently of a game where Stockfish blundered a game in a tournament because during a search there was an hash-table issue for positions inside the tree very close to the 50-moves draw rule. This is part of a problem which is commonly referred to as the Graph History Interaction (GHI), and is difficult to solve in computer chess because storing the 50-moves counter in the hash-table loses Elo in general.

This patch tries to address the issue in this particular game and similar reported games: it prevents that values from the transposition table are getting used when the 50-move counter is close to reaching 100 (). The idea is that in such cases values from previous searches, with a much lower 50-move count, become less and less reliable.

More precisely, the heuristic we use in this patch is that we don't take the transposition table cutoff when we have reached a 45-moves limit, but let the search continue doing its job. There is a possible slowdown involved, but it will
also help to find either a draw when it thought to be losing, or a way to avoid the draw by 50-move rule. This heuristics probably will not fix all possible cases, but seems to be working reasonably well in practice while not losing too much Elo.

Passed non-regression tests:
STC:
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 274452 W: 59700 L: 60075 D: 154677 Elo -0.47

LTC:
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 95235 W: 15297 L: 15292 D: 64646 Elo +0.02

Bench: 4586187


Rating JCER=3344

Stockfish 20010919 - download


Comments

Popular posts from this blog

We already have the new Fritz 20 program - what does it look like and how does it work?

Stockfish 250702 wins Android Chess Engines Tournament, 2025.08.03-04