RE: [computer-go] 19x19 results (so far)
I do something similar to this in Many Faces. -Original Message- From: computer-go-boun...@computer-go.org [mailto:computer-go-boun...@computer-go.org] On Behalf Of Don Dailey Sent: Wednesday, December 24, 2008 5:47 AM To: computer-go Subject: [computer-go] 19x19 results (so far) 19x19 results of 3,4,5 rank rule: Rank Name Elo+- games score oppo. draws 1 d2p 2050 21 21 273 57% 20000% 2 base 2000 21 21 273 43% 20500% ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] 19x19 results (so far)
So if I understand this correctly, you only allow moves on the 3rd, 4th, or 5th lines to be considered (in both the tree and the playouts) unless there is another stone within manhattan distance of two? What would be really interesting is if one of the stronger open source engines was modified to have this rule on 19 by 19. It seems like it should surely help even a stronger engine if it only was applied in the tree up to some fixed depth and with distance 3. If the rule never eliminates the moves that the engine would have picked without the rule it seems really unlikely to hurt performance. So I think the results would be more and more interesting the stronger the base version is. Is there a rule like this that fits pro play? - George On Wed, Dec 24, 2008 at 8:46 AM, Don Dailey dailey@gmail.com wrote: 19x19 results of 3,4,5 rank rule: Rank Name Elo+- games score oppo. draws 1 d2p 2050 21 21 273 57% 20000% 2 base 2000 21 21 273 43% 20500% ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] 19x19 results (so far)
On Wed, 2008-12-24 at 06:25 -0800, David Fotland wrote: I do something similar to this in Many Faces. Do you have a better formulation of the rule that I could try on the reference bot? Basically, in the playouts or move choice I veto any move to the 3rd 4th or 5th line unless there is a stone of either color within distance 2 manhattan distance. Another version does 3 manhattan distance. I am prepared to try any simply formulated variations of this - as long as it doesn't involve complex pattern analysis. - Don -Original Message- From: computer-go-boun...@computer-go.org [mailto:computer-go-boun...@computer-go.org] On Behalf Of Don Dailey Sent: Wednesday, December 24, 2008 5:47 AM To: computer-go Subject: [computer-go] 19x19 results (so far) 19x19 results of 3,4,5 rank rule: Rank Name Elo+- games score oppo. draws 1 d2p 2050 21 21 273 57% 20000% 2 base 2000 21 21 273 43% 20500% ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] 19x19 results (so far)
On Wed, 2008-12-24 at 11:01 -0500, George Dahl wrote: So if I understand this correctly, you only allow moves on the 3rd, 4th, or 5th lines to be considered (in both the tree and the playouts) unless there is another stone within manhattan distance of two? Correct. What would be really interesting is if one of the stronger open source engines was modified to have this rule on 19 by 19. It seems like it should surely help even a stronger engine if it only was applied in the tree up to some fixed depth and with distance 3. If the rule never eliminates the moves that the engine would have picked without the rule it seems really unlikely to hurt performance. In 9x9 the distance 3 did not work as well. Intuitively, it seems to me (being a very weak go player) that 3 would be almost perfect and safe. But I noticed that with 3, very little is actually cut, perhaps just 2 or 3 moves really early in the opening. I have not yet tried distance 3 with 19x19. - Don So I think the results would be more and more interesting the stronger the base version is. Is there a rule like this that fits pro play? - George On Wed, Dec 24, 2008 at 8:46 AM, Don Dailey dailey@gmail.com wrote: 19x19 results of 3,4,5 rank rule: Rank Name Elo+- games score oppo. draws 1 d2p 2050 21 21 273 57% 20000% 2 base 2000 21 21 273 43% 20500% ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] 19x19 results (so far)
On Wed, Dec 24, 2008 at 01:27:46PM -0500, Don Dailey wrote: Basically, in the playouts or move choice I veto any move to the 3rd 4th or 5th line unless there is a stone of either color within distance 2 manhattan distance. Another version does 3 manhattan distance. I hope you mean that you veto any move *not* on the 3rd, 4th, or 5th line, unless there is a stone within the limit distance. Or, to phrase it in other words: - Any move on 3-4-5 line is OK - Any move close than 2 or 3 points from any stone is OK - All other moves are not OK -H -- Heikki Levanto In Murphy We Turst heikki (at) lsd (dot) dk ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] 19x19 results (so far)
On Wed, 2008-12-24 at 19:43 +0100, Heikki Levanto wrote: On Wed, Dec 24, 2008 at 01:27:46PM -0500, Don Dailey wrote: Basically, in the playouts or move choice I veto any move to the 3rd 4th or 5th line unless there is a stone of either color within distance 2 manhattan distance. Another version does 3 manhattan distance. I hope you mean that you veto any move *not* on the 3rd, 4th, or 5th line, unless there is a stone within the limit distance. Or, to phrase it in other words: - Any move on 3-4-5 line is OK - Any move close than 2 or 3 points from any stone is OK - All other moves are not OK Yes, I mis-worded it. Your way of wording it makes it clear. - Don -H ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/