Optimal game length analysis using KataHex 20240908

I have written all the content here:

https://dev.to/hzyhhzy/analysis-of-the-length-of-optimal-games-of-hex-game-using-alphazero-like-ai-16n7

Maybe I should better use a hyperlink
Analysis of the length of optimal games of Hex game using alphazero-like AI

I add a new feature "Move number limit" to KataHex. 20240908 is the latest release. KataHex 20240908

Nice analysis, thanks!

We used to roughly estimate the optimal game to occupy 1/3 of the board, but that was excluding "useless moves", like filling in the gaps in bridges. Often those moves are forcing, so one can play them to prolong the game or gain time on the clock if increment is included. The assumption of your analysis was that one of the players prolongs as much as possible, so that seems quite right that the number is significantly higher, F(n) ~ 0.45 * n^2.

Very interesting if that function is convergent to a value larger than 0!!!