Author: blass uri
Date: 05:54:46 10/06/99
Go up one level in this thread
On October 06, 1999 at 06:33:37, Dan Newman wrote: >On October 05, 1999 at 12:26:17, Robert Hyatt wrote: > >>On October 05, 1999 at 11:06:03, blass uri wrote: >> >>>On October 05, 1999 at 10:51:29, Robert Hyatt wrote: >>><snipped> >>>>On October 05, 1999 at 04:14:13, blass uri wrote: >>>> >>>>>On October 04, 1999 at 22:37:48, Robert Hyatt wrote: >>>>> >>>>>>On October 04, 1999 at 20:33:35, blass uri wrote: >>> >>>>>>>The relevant position is: >>>>>>> >>>>>>>7k/4K2p/7P/3p4/8/4Q3/1q6/8 w - - 0 1 >>>>>>> >>>>>>>The first evaluation above 0 of Fritz5 is +5.16 pawns for white >>>>>>>I do not believe that it can be explained by positional compensation. >>>>>>> >>>>>>>If it is because of a bug then Fritz3 and Fritz4 have the same bug(I do not know >>>>>>>if Fritz5.32 shows similiar behaviour) >>>>>>> >>>>>>>Uri >>>>>> >>>>>> >>>>>>are you sure this is the position? This seems to be a mate in 8 according >>>>>>to my PII/300 notebook (crafty): >>>>>> >>>>>> 9 4.89 0.48 1. Ke6 Qa1 2. Kd7 Qa4+ 3. Kd6 Qb4+ >>>>>> 4. Kxd5 Qb7+ 5. Kc4 Qa6+ 6. Kb4 Qd6+ >>>>>> 7. Kb5 Qd5+ 8. Kb4 <HT> >>>>>> 9-> 7.46 0.48 1. Ke6 Qa1 2. Kd7 Qa4+ 3. Kd6 Qb4+ >>>>>> 4. Kxd5 Qb7+ 5. Kc4 Qa6+ 6. Kb4 Qd6+ >>>>>> 7. Kb5 Qd5+ 8. Kb4 <HT> >>>>>> 10 8.27 ++ 1. Ke6!! >>>>>> 10 12.34 Mat08 1. Ke6 Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>>>> Qe5+ <HT> >>>>>> 10-> 18.59 Mat08 1. Ke6 Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>>>> Qe5+ Qg7+ 8. Qxg7# <HT> >>>>>> 11 23.23 Mat08 1. Ke6 Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>>>> Qe5+ Qg7+ 8. Qxg7# >>>>>> 11-> 36.33 Mat08 1. Ke6 Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>>>> Qe5+ Qg7+ 8. Qxg7# >>>>>> 12 50.03 Mat08 1. Ke6 Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>>>> Qe5+ Qg7+ 8. Qxg7# >>>>> >>>>>I am sure that it is the position and I believe that the reason that Fritz >>>>>cannot see the mate is that it is a null move problem for Fritz. >>>>>1.Ke6 threats nothing. >>>>> >>>>>Try the position after Ke6 with white to move. >>>>> >>>>>Uri >>>> >>>> >>>>I get this: >>>> >>>> 9 5.01 -Mat07 1. ... Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>> Qe5+ Qg7+ 8. Qxg7# <HT> >>>> 9-> 6.13 -Mat07 1. ... Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>> Qe5+ Qg7+ 8. Qxg7# <HT> >>>> 10 11.37 -Mat07 1. ... Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>> Qe5+ Qg7+ 8. Qxg7# >>>> 10-> 11.76 -Mat07 1. ... Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>> Qe5+ Qg7+ 8. Qxg7# >>>> 11 24.97 -Mat07 1. ... Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>> Qe5+ Qg7+ 8. Qxg7# >>>> 11-> 25.92 -Mat07 1. ... Qa1 2. Qb6 Qe1+ 3. Kd7 Kg8 4. >>>> Qf6 Qg3 5. Qe6+ Kh8 6. Qe8+ Qg8 7. >>>> Qe5+ Qg7+ 8. Qxg7# >>>> >>>>although it is certainly possible that small differences in the two programs >>>>can cause interesting null-move behavior... >>> >>>Fritz can also see mate against itsealf after Ke6 but the point is that ke6 >>>threats nothing so Fritz cannot see the mate before Ke6. >>> >>> >>>You can prove it by analyzing the position after Ke6 with *white* to move. >>> >>>Uri >> >> >>With white to move, it reaches a tablebase draw score instantly, meaning >>black is zugzwanged... Although I don't see why Fritz would miss the mate. >>I tried it on my quad xeon and I find the mate instantly even though I have all >>the existing tablebase files on this box, so that null move could quickly screw >>up with that 'draw' score returned after a null at ply=2. > >Here's a run of my program Skyblue on a P6/200 with null move turned on: > > 6. -7 0.44 27348 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qf7 Qh3 Kg8 > 6> -7 0.55 40364 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qf7 Qh3 Kg8 > 7. -7 0.72 58899 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Qd8+ Ke4 Kg8 > Qd4 > 7> -7 1.05 105697 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Qd8+ Ke4 Kg8 > Qd4 > 8. -7 1.42 161829 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Qd8+ Ke4 Qe8+ > Kd4 Qxe3+ Kxe3 Kg8 Kd4 > 8> -7 3.02 368776 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Qd8+ Ke4 Qe8+ > Kd4 Qxe3+ Kxe3 Kg8 Kd4 > 9. -7 4.07 518752 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Qd8+ Ke4 Qe8+ > Kd4 Qxe3+ Kxe3 Kg8 Kd4 Kh8 > 9. 0 10.38 1382701 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> > 9> 0 10.82 1459411 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >10. 0 12.31 1704858 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >10> 0 14.72 2107062 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >11. 0 18.78 2776881 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >11> 0 24.06 3677158 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >12. 0 33.39 5264064 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >12> 0 43.60 7022910 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >13. 0 1:09 11431291 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> >13> 0 1:31 15234044 Qf4 Qb7+ Ke8 Qc6+ Ke7 Qb7+ <TPR draw> > >Here it is with null move off: > > 6. -7 0.54 34526 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Kg8 Qe5 > 6> -7 0.82 68732 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Kg8 Qe5 > 7. -7 1.04 101081 Kd6 Qb4+ Kxd5 Qb7+ Ke5 Qb8+ Kd4 Qd8+ Ke4 Kg8 > Qd4 > 7. 0 1.26 126478 Ke6 Qa1 Ke7 Qb2 Ke6 <TPR draw> > 7> 0 1.75 196352 Ke6 Qa1 Ke7 Qb2 Ke6 <TPR draw> > 8. 0 2.03 235454 Ke6 Qa1 Ke7 Qb2 Ke6 <TPR draw> > 8> 0 3.51 463732 Ke6 Qa1 Ke7 Qb2 Ke6 <TPR draw> > 9. 0 4.33 585858 Ke6 Qa1 Ke7 Qb2 Ke6 <TPR draw> > 9> 0 7.79 1145268 Ke6 Qa1 Ke7 Qb2 Ke6 <TPR draw> >10+ 100 9.06 1339642 Ke6 >10> +mate8 10.87 1718339 Ke6 Qa1 Qb6 Qe1+ Kd7 Kg8 Qf6 Qg3 Qe6+ Kh8 > Qe8+ Qg8 Qe5+ Qg7+ Qxg7+ <Checkmate> > >So, it appears to be a null move problem of some sort in my program. >Perhaps if I let it run longer... > >Interesting problem, I'll have to study it. > >-Dan. This position is from a study of noam alex(some plies after the beginning of the study). The original position of the study is very hard for programs(even not null movers). 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.