nvthauclone002 wrote:
White: King at e1, Bishop at f1, Knight at g1. Black: King at e8.
I open chess.com's application, setting computer lv10 vs computer lv10 (white to move), and they play over 1000 moves, more and more, not result.
According to a tablebase, that takes 28 moves with best play. We don't have access to any info about the chess.com computers. Do other engines solve that problem without tablebases?
Why would you think that the computer should be able to solve it?
Any good engine should be able to handle this position. On my computer, Fritz 15 reports a forced mate after reaching a depth of 37 plies after about 6 minutes of thinking time.
Thanks. I find website which help me solve this problem. (shredderchess.com).
When I use it to play with computer lv10, sometime I see computer choose the move that help me win with less moves.
For a human it is more important to win with 'a good system' than with 'less moves'. It is impossible to play as efficient as the computer does. No player, grandmaster or worldchampion can match it. Be happy to know one method of winning each one of many different endgames and you'll have the best result. Just keep the 50-move limitation in mind.
This post was originally about the OP expressing surprise that the chess.com computer set at level 10 couldn't solve the mate, despite his letting it run for 1000 moves.
Someone said any modern chess engine could solve it. It wasn't clear if that meant with or without access to tablebases. Also, level 10 isn't allowed to think for more than a few seconds, not 6-7 minutes.
I wonder if modern engines can do the knight bishop and king versus king mate without tablebases and with a 10 second or less per move limit.
Someone said any modern chess engine could solve it. It wasn't clear if that meant with or without access to tablebases.
Actually, it was clear. If the engine had tablebase access, the result would have been immediate, not after six minutes thinking time.
Someone said any modern chess engine could solve it. It wasn't clear if that meant with or without access to tablebases.
Actually, it was clear. If the engine had tablebase access, the result would have been immediate, not after six minutes thinking time.
What about the only a few seconds per move part? How does it do then at the problem?
What an engine might do with only 10 seconds per move is unimportant, but it would be interesting to know what might happen in a blitz game at a time control of 5 minutes with no time delay. I let Fritz 15 try that, and it mated me in 33 moves with two minutes to spare. (Neither the engine nor I had tablebase access.)
I try to make black king at white corner, but can not, hundreds moves and more...
But when I lookup endgame databases, I win easily and quickly.
https://vi.lichess.org/zAGY0xXG/black (stockfish 2500 ELO can not checkmate me)
On my home computer Stockfish beats me easily from that board given 60 seconds [total, not per move]. It moves its king to the centre then finds mate in 25 after about 20 seconds of calculating.
I did that on an old chess master program, back in the 90's, it easily solved it. I used to practice that mate quite a bit. Because I once had it in a real game.
I practice it, now I can checkmate with knight and bishop. (mate in ~40 moves and ~10 times Takeback)
notmtwain asked how a computer program would do with only a few seconds per move.
I can answer that one. I was stuck in the Admin id on my PC waiting for a virus check to finish recently, so I thought I'd just bring up Rybka and practice a bishop and knight ending.
I played the lone king and was a bit shocked to find Rybka started playing ring-a-ring-a-roses in the wrong corner.
I thought there was something seriously wrong with my system, but it turned out that the last time I'd used Rybka in the Admin id months earlier I'd set the clock time rediculously low and forgotten to set it back.
Rybka had the last laugh, because I'd set both clocks rediculously low and I lost on time before the fifty move limit anyway.
I think that was two minutes for 40 moves with no extensions.
Whenever I setup positions, like the knight-bishop, basic mate. I always set the time at 5 minutes. It never had a problem, mating the lone king.
I generally have mine set to 20 seconds a move which is about the point below which Rybka shows a noticible deterioration in two knights v pawn ending and of course it has no problem with bishop and knight on that setting, in fact I would say it plays better as black than the Nalimov EGTB.
On the other hand it is very flaky on two knights v pawn in positions that need more than about 65 moves and your post just prompted me to think that the problem could well be my time setting. Thanks for that.
White: King at e1, Bishop at f1, Knight at g1. Black: King at e8.
I open chess.com's application, setting computer lv10 vs computer lv10 (white to move), and they play over 1000 moves, more and more, not result.