Author: Omid David Tabibi
Date: 12:06:41 12/12/03
Go up one level in this thread
On December 12, 2003 at 13:29:36, Robert Hyatt wrote: >On December 12, 2003 at 12:38:26, Omid David Tabibi wrote: > >>On December 12, 2003 at 09:28:00, Robert Hyatt wrote: >> >>>On December 12, 2003 at 03:36:47, Omid David Tabibi wrote: >>> >>> >>>> >>>>I had similar problems in Falcon. I fixed the problem by storing the checkmate >>>>score as a bound only. Not an efficient solution of course, but haven't >>>>encountered any such bugs ever since. >>> >>> >>>There is no need to cripple the search by not storing mate scores at all. >>>Adjusting mate scores to not have such problems is very simple, as the >>>following output shows: >>> >>> >>> >>> depth time score variation (1) >>> 8 0.12 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 >>> 8-> 0.12 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 >>> 9 0.22 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 >>> 9-> 0.22 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 >>> 10 0.39 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 6. Qf8 >>> 10-> 0.39 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 6. Qf8 >>> 11 0.76 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 6. Qf8 Ka1 >>> 11-> 0.76 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 6. Qf8 Ka1 >>> 12 1.34 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 6. Qf8 Ka1 7. Kf6 >>> 12-> 1.34 10.19 1. ... Kb6 2. Qg8 Ka5 3. Qf8 Ka4 4. >>> Qh8 Ka3 5. Qg8 Kb2 6. Qf8 Ka1 7. Kf6 >>> 13 1.87 -- 1. ... Kb6 >>> 13 2.75 Mat09 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qd5 Ka7 7. Kc6 >>> Kb8 8. Qd8+ <HT> >>> 13-> 2.83 Mat09 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qd5 Ka7 7. Kc6 >>> Kb8 8. Qd8+ <HT> >>> 14 3.63 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 14-> 3.64 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 15 4.87 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 15-> 5.08 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 16 6.77 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 16-> 7.10 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 17 9.22 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 17-> 9.79 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 18 12.45 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 18-> 13.21 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 19 15.82 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 19-> 16.49 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 20 19.52 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 20-> 20.40 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 21 23.41 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 21-> 24.31 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 22 27.33 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 22-> 28.67 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 23 31.75 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 23-> 32.87 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 24 36.13 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 24-> 37.41 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 25 40.49 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 25-> 41.71 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 26 44.83 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 26-> 46.10 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 27 49.29 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 27-> 50.60 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 28 53.73 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 28-> 55.01 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 29 58.11 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 29-> 59.60 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 30 1:02 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 30-> 1:04 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 31 1:07 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 31-> 1:08 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 32 1:11 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 32-> 1:13 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> 33 1:16 Mat08 1. ... Kb6 2. Qh5 Ka6 3. Kf6 Kb6 4. >>> Ke6 Ka6 5. Kd6 Kb6 6. Qc5+ Kb7 7. Qb5+ >>> Ka7 8. Kc7 Ka8 9. Qb8# >>> <etc> >>> >>>Shredder has a _serious_ bug in hashing somewhere. >> >>There are two ways to solve a problem: find a good solution, or erase the >>problem :) > >I agree. But not storing exact mate scores is not (a) find a good solution. >I don't know where the urban legend that you can't safely store mate scores >came from, but it is false. I've been storing mate scores in my trans/ref >table since the middle 1970's. Everyone else I knew back then did so as >well. Why it would suddenly cause problems is a mystery, but why those >problems are not solved is a _real_ mystery. IE as you can see from my >output above, I don't see any funny-business with mate scores. Never have, >in fact... It is of course possible to store exact mate scores without bugs. But as we see once and again, bugs in hashing mate scores are one of the most frequent problems (don't ask me why). Many believe that they have a bug-free implementation, only to discover some years later that there is some strange bug somewhere. I don't want the headache!
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.