I just wondered why the computer analysis for this game (link below) gives a very different Advantage Rating for the last 10 or so Moves, compared to the Advantage Rating given by Local Analysis, regardless of the Parameters I set for Local Analysis. Is this a bug?
The server analysis agrees with the tablebases (0.00 when the game was a theoretical draw, highly negative scores when black got a winning advantage). The only difference I see with local analysis is that it starts to report mates earlier (and the mates it finds are shorter than the #17 in the server analysis) but I see no fundamental difference. The server just didn't give it enough time to calculate the fastest mate.
I see, so the server analysis really uses a much shallower recursion depth to save computation time, makes sense, it's gotta analyze alot of games and the service is free anyway. There really is no fundamental difference in the calculated Ratings, but for the last move when white resigns, there the server states 5.4 Advantage, while the Local Analysis yields mate in #10. So the server's analysis recursion depth must be less than 10?
It's limited by the number of nodes, 2.25 million to be specific.
(Also, depth is usually measured in half-moves, so it'd take depth 20 to see that white is getting mated in 10).
(Also, depth is usually measured in half-moves, so it'd take depth 20 to see that white is getting mated in 10).
This topic has been archived and can no longer be replied to.