Author: Uri Blass
Date: 07:22:28 08/30/03
Go up one level in this thread
On August 30, 2003 at 09:04:58, Geoff wrote: >On August 30, 2003 at 08:40:19, Uri Blass wrote: > >>On August 30, 2003 at 08:18:41, Geoff wrote: >> >>>Hi >>> >>>My program just "excelled" itself with my favourite worst ever move. >>>This shambles wasn't due to an obvious bug as such, more just to lack of search >>>depth. Here is the position >>> >>>[D]2r3k1/1b2n1bp/p2N4/1p6/6Q1/1P6/1PqN1PPP/R5K1 b - - 0 30 >>> >>>My program was playing black, my opponent offered me 2 tempting hung pieces >>>adjacent to my queen. My program avoided the knight capture, but couldn't resist >>>gobbling up the white pawn on b2. >>>Can you see the outcome of this move just be looking at the position? I couldnt >>>really see it, but most computer analysis will show it very quickly. >>>Hint: I got smothered, very elegantly !! >> >>Yes >>This idea is a known idea for chess player and I guess that even most players >>with rating 1600 will have no problem to see it. >> >> >>> >>>The other reason I found this quite amusing was the opponent that trapped me >>>into this blunder. Was it Shredder?, Junior?, Fritz? , nope just humble old TSCP >>>:-) >>> >>>Anyone else got a favorite clanger that wasn't due to just an obvious bug ? >>> >>> Regards Geoff >> >>I can only tell you that if my program does not find something simple like that >>in less than 0.1 second then it is a clear proof that it has bugs or problems >>that I need to fix. >> >>Uri > >Hi Uri > >I would be interested to see how long it takes and how many nodes before your >program avoid either b2 or b4 ? Could you post some output please ? > >Yes I know its not very good, but here is my current version's output > >Ply Eval Time Nodes Principle Variation > 1: 183 0 171 c2b2 > 2: 178 0 823 c2b2 a1d1 > 3: 172 0 3620 c2b2 g4e6 g8h8 > 4: 167 1 12755 c2b2 g4e6 g8h8 a1d1 > 5: 155 6 46465 c2b2 g4e6 g8h8 a1e1 c8c1 d6b7 b2d2 > 6: 155 12 114778 c2b2 > 7: 20 65 603463 c2d2 g4e6 g8h8 d6f7 h8g8 f7d6 > 8: 20 106 963192 c2d2 g4e6 g8h8 d6f7 h8g8 f7d6 > 9: 20 218 2071857 c2d2 g4e6 g8h8 d6f7 h8g8 f7d6 >10: 20 435 4045469 c2d2 g4e6 g8h8 d6f7 h8g8 f7d6 >11: 20 2267 21624492 c8c7 g4e6 g8h8 d6f7 h8g8 f7d6 Here is my output(4547 nodes to see mate score against itself, 6528 to fail high on a different move,9128 nodes to solve the fail high,31316 nodes to finish the iteration with a better move) The time of 0.05 seconds to see the mate is not correct because another application ran at the same time and I also used a slow p850. I guess that extending 1 ply for checks is better for your program. You are not going to find it as fast as me but you are going to find it faster than you find it today. 3 0 5 3578 c2b2 g4e6 g8h8 d6f7 h8g8 f7d6 4 -9990 5 4547 c2b2 g4e6 g8h8 d6f7 h8g8 f7h6 g8h8 e6g8 e7g8 4 -9989 5 6528 e7f5 4 -9960 5 6734 e7f5 4 -165 11 9128 e7f5 d6b7 c2d3 a1d1 4 -164 16 11042 b7d5 4 -135 16 12396 b7d5 4 -6 22 15093 b7d5 d6c8 c2c8 g4c8 e7c8 a1a6 g7b2 4 -5 22 19079 c8c7 4 0 33 20955 c8c7 g4e6 g8h8 d6f7 h8g8 f7d6 4 0 38 31316 c8c7 g4e6 g8h8 d6f7 h8g8 f7d6 Uri
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.