Terry McCracken wrote:
Here's a short line.
1. Rb6 Rxa5 2. Rb3 Be6 3. Rc3 Rb5+ 4. Ka1 Rxc3 5. Rc7+ ( Kxc7=) Kd8 6. Rxc3=
That might work too, the problem is, no engine would play 2... Be6 seeing the following leading to immediat exchange of rooks with at least complete 5 MOB on disk or in main memory.
After I managed one remis against DF10 with help of DF 12
with nullmove off,
http://www.talkchess.com/forum/viewtopi ... 60&t=32823
I went back with Firebird, just because it's fast keeping all the way to 50 moves boundary in hash.
Here is its analyzing output afterwards:
2k5/3b2R1/7R/P3r3/2r5/8/1K6/8 w - - 0 1
Analysis by FireBird 1.1 w32:
1.a6 Tb5+ 2.Ka3 Te4 3.Td6 Le6 4.Tb7
= (-0.18) Tiefe: 6 00:00:00 9kN, tb=49
...
1.a6 Tb5+ 2.Ka3 Te4 3.Tg3 Ta5+ 4.Kb2 Te2+ 5.Kb3 Le6+ 6.Kb4 Txa6 7.Tg7 Tb2+ 8.Kc3 Tb3+ 9.Kd4 Ta4+ 10.Kc5 Tc3+ 11.Kb6
=/+ (-0.49) Tiefe: 10 00:00:00 246kN, tb=3130
1.Kb3 Tf4 2.Th8+ Kc7 3.Th3 Tb5+ 4.Kc3 Tc5+ 5.Kb2 Td5 6.a6 Tf2+ 7.Kc3 Kb6 8.Td3 Tc5+ 9.Kd4 Tf4+ 10.Ke3
=/+ (-0.55) Tiefe: 11 00:00:00 377kN, tb=7004
...
1.Kb3 Tc7 2.Tg8+ Kb7 3.Tb6+ Ka7 4.Tgg6 Txa5 5.Tgd6 Tc1 6.Kb2 Tac5 7.Ka2 Lc6 8.Tb1 T1c2+ 9.Ka3 T2c3+ 10.Kb2
=/+ (-0.64) Tiefe: 14 00:00:01 2264kN, tb=65056
1.Th8+ Kc7 2.Th3 Kc6 3.Tc3 Tb5+ 4.Tb3 Txa5 5.Tgg3 Th4 6.Tg6+ Kc7 7.Tc3+ Kb7 8.Tb3+ Ka7 9.Td6 Th2+ 10.Kc3 Tc5+ 11.Kd4 Tc7 12.Td5 Ka6 13.Te5 Td2+ 14.Ke3 Tg2 15.Kd4 Lc6 16.Ta3+ Kb7 17.Tb3+ Kc8 18.Te6 Td7+ 19.Kc5 Tc2+ 20.Kb4
=/+ (-0.59) Tiefe: 14 00:00:01 3725kN, tb=126364
...
1.Th8+ Kc7 2.Th3 Kc6 3.Tc3 Tb5+ 4.Tb3 Txa5 5.Tgg3 Th4 6.Tg6+ Kc7 7.Tc3+ Kb7 8.Tb3+ Ka7 9.Td6 Th2+ 10.Kc3 Tc5+ 11.Kd4 Tc7 12.Td5 Ta2 13.Ke5 Lc6 14.Tdd3 Te2+ 15.Te3 Td2 16.Ted3 Tf2 17.Tb1 Te7+ 18.Kd4 Ta2 19.Kc4 Tc2+ 20.Tc3
=/+ (-0.67) Tiefe: 27 00:07:11 1366mN, tb=84393658
You see, eval is starting with -0.18!, climbing up to the value of empty hash slowly, the more he "forgets" what he has learned.
You were right about zugzwang, you may have been right about remis.
Peter.