Author: Robert Hyatt
Date: 10:46:53 08/26/05
Go up one level in this thread
On August 26, 2005 at 13:39:00, Gian-Carlo Pascutto wrote:
>On August 26, 2005 at 13:34:21, Robert Hyatt wrote:
>
>>Here's what I got: Note most moves are hopeless. I only asked for the best 10,
>>and that was really too many here... Note that some of the PVs end in the usual
>><HT> but that is now shown in the annotate output, unfortunately, so any PV
>>could have a little more "stuff" on the end.
>>
>>Note that this is also the perfect place for Hsu's last incarnation of singular
>>extensions, where they extended, but less, if there are only two decent moves.
>>Qe2 and Qf1 really seem to be "it" here. I'd suspect that SE would work well
>>here in general, if it isn't the "cheapo-type" that Bruce and I worked on a few
>>years back, but rather is a "real SE algorithm as explained by Hsu"...
>>
>>
>>
>> ({16:-1.31} 49. Qe2 Qf4 50. Rf1 Qg5 51. Bc4 Ke7 52. Bf7 Re3 53.
>>Qa2 Qg2+ 54. Qxg2 hxg2+ 55. Kxg2 Rxe4 56. Bc4 Be3 57. Bd5 Rd4 58. Bc6 $17)
>> ({16:-1.63} 49. Qf1 Ke7 50. Qe2 Qf4 51. Rf1 Qg5 52. Bb1 Bc5 53.
>>Qb2 Rb3 54. Qc2 Kf8 55. Qa2 Ba3 56. Rd1 Rc3 $19)
>> ({16:-3.37} 49. Rb1 d5 50. Rf1 dxe4 51. Qxe4 Qxe4+ 52. Bxe4 Kc7
>>53. Rc1+ Kd7 54. Re1 Ra2 55. Bd5 Rf2 56. Be4 Rb2 57. Rd1 Ra2 58. Rc1 Kd6 $17)
>
>I was hoping to get to know how many plies/time are needed to see Rb1 Rc3 (or
>other winning moves, but not Rxd3 ;-)
>
>I have problems here, but I know by now from testing I'm not the only one.
>
>--
>GCP
Tell me how you'd like it run. Looks like the Rb1 d5 is pretty bad as well, but
I can't say whether that is all tactics or some tactics and some eval there...
I can just make it search Rb1 and post that log. Will that give you what you
need???
(the log would have every search for rb1, iteration by iteration, with the PV
and score...)
The problem with some of these null-move positions is that it is possible that
program a's positional eval is enough to break the null-move search so that it
doesn't hurt as much, and sometimes it hurts even worse of course...
Comparing is difficult since all the null-move search cares about is "do I fail
high or not, not why"
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.