Position : Win or Draw ?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Ovyron
Posts: 3827
Joined: Tue Jul 03, 2007 2:30 am

Re: Position : Win or Draw ?

Post by Ovyron » Sun Dec 15, 2019 12:22 am

fishpov wrote:
Sat Dec 14, 2019 4:31 pm
I am pretty sure sure this is draw, because on move after and after Sting gives inconcistent evaluation even after one hour.
someone could give me autoplaying result ?
In Fritz GUI, how to enable it ?
Thanks.
Does it have shootout? If so the first thing you want to do is make it play against itself, because if it can't even beat itself perhaps it needs to switch to vaping.

peter
Posts: 1852
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Position : Win or Draw ?

Post by peter » Sun Dec 15, 2019 8:20 pm

MikeB wrote:
Sat Dec 14, 2019 8:02 pm
How about this one:

Much more difficult for SF to see the fortress. Even Crystal needs some Forward- Backward to keep the 0.00 in hash to about 10 moves from starting position after backwarding of about 20 moves. Further back than those, eval rises to +- once and again.
Might be some misevalutation of the plus e-pawn without counter- pawn, I guess.

Komodo does a little better as for eval at starting position, at least not a clear +-, yet it doesn't keep the 0.00 from 50 moves- horizon very far backward.

Started with empty hash:

3k4/1p3n2/1Pp3p1/2Pp1p2/3P1P2/4P3/4K1N1/8 w - - 0 1

Analysis by Komodo 13.2.5 64-bit:

1.Nh4 Kd7 2.Nxg6 Nd8 3.Ne5+ Ke6 4.Kf2 Kf6 5.Kg3 Kg7 6.Nf3 Nf7 7.Kh4 Kh6 8.Kh3 Kg7 9.Nh4 Kf6 10.Kg3 Nd8 11.Nf3 Kf7 12.Kh4 Kg6 13.Ne5+ Kh6 14.Kh3 Kg7 15.Kg3 Kh7 16.Nd7 Kg7 17.Kg2 Kf7 18.Kh3 Kg7 19.Kh2 Kf7 20.Kg2 Ke6 21.Ne5 Kf6 22.Kg1 Ke7 23.Kh2 Kf8 24.Kh1 Kg7 25.Nd7 Kg8 26.Kg1 Kf7 27.Kh2 Kg7 28.Kg3 Kf7 29.Nb8 Kg7 30.Kf2 Kf7 31.Kg2 Kg6 32.Kf3
+/- (1.40) Depth: 60 00:06:53 14941MN
Peter.

peter
Posts: 1852
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Position : Win or Draw ?

Post by peter » Sun Dec 15, 2019 9:11 pm

peter wrote:
Sun Dec 15, 2019 8:20 pm
Komodo does a little better
LC0 net 49921:



Analysis by Lc0 v0.23.0 2xRTX2080Ti syzygy:

1.Nh4 Ke7 2.Nxg6+ Kf6 3.Ne5 Nd8 4.Nd3 Kg7 5.Kf2 Kh6 6.Kg2 Kh5 7.Ne1 Kh6 8.Nf3 Kg7 9.Kg3 Kf6 10.Kh3 Kg7 11.Ne5 Kh7 12.Kh4 Kh6 13.e4 dxe4 14.Kg3 Kg7
+- (1.93) Depth: 19/41 00:07:25 75244kN, tb=889

13...dxe4? 13...fxe4

Fat Fritz 2x2080:

3k4/1p3n2/1Pp3p1/2Pp1p2/3P1P2/4P3/4K1N1/8 w - - 0 1

Analysis by Fat Fritz v301:

1.Nh4 Ke7 2.Nxg6+ Kf6 3.Ne5 Nd8 4.Kf2 Kg7 5.Kg2 Kh6 6.Nf3 Kg7 7.Kh3 Kf6 8.Kg3 Kg7 9.Ne5 Kh7 10.Kh4 Kh6 11.Kg3 Kg7 12.Nd3 Kg6 13.Kh4
+- (2.04) Depth: 18/36 00:05:15 24842kN, tb=140

12...Kg6? 12...Nf7
Peter.

peter
Posts: 1852
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Position : Win or Draw ?

Post by peter » Sun Dec 15, 2019 10:21 pm

peter wrote:
Sun Dec 15, 2019 8:20 pm
MikeB wrote:
Sat Dec 14, 2019 8:02 pm
How about this one:

Komodo does a little better
Brainlearn with empty hash:

3k4/1p3n2/1Pp3p1/2Pp1p2/3P1P2/4P3/4K1N1/8 w - - 0 1

Analysis by BrainLearn 6.0 64 POPCNT:

1.Nh4 g5 2.Nxf5 gxf4 3.exf4 Kd7 4.Kd3 Ke6
+/= (0.39) Depth: 7/11 00:00:16 29kN
...
1.Nh4 Kd7 2.Nxg6 Ke6 3.Ne5 Nd8 4.Kd3 Kf6 5.Kc3 Ke6 6.Kb4 Kf6 7.Nf3 Ke6 8.Nh4 Kf6 9.Kc3 Ke6 10.Kd2 Kf6 11.Kd3 Ne6 12.Kc3 Nd8 13.Kb4 Ne6 14.Nf3 Ke7 15.Ne5 Nd8 16.Ka5 Ke6 17.Nd3 Kf6 18.Nb4 Ke6 19.Ka4 Kf6 20.Kb3 Ke6 21.Nc2 Nf7 22.Kc3 Kf6 23.Kd3 Nd8 24.Ne1 Nf7 25.Nf3 Nd8 26.Kc2 Nf7 27.Ne5 Nd8 28.Kb2 Ke6 29.Kc3 Kf6 30.Nd3 Nf7 31.Kb3 Nd8 32.Ne1 Ne6 33.Kb4 Nd8 34.Kc3 Ke6 35.Kc2 Nf7 36.Nd3 Kf6 37.Nb4 Nd8 38.Kc3 Ke6 39.Kd3 Ke7 40.Kd2 Kf6 41.Ke2 Ke6 42.Kf3 Kf6 43.Kf2 Kg6
+/= (0.41) Depth: 82/86 00:03:52 7029MN
1.Nh4
+/= (0.44 ++) Depth: 83/90 00:05:12 9251MN
...
1.Nh4
+/- (1.10 ++) Depth: 83/104 00:16:12 29198MN

And after a 50 moves Backward:

3k4/1p3n2/1Pp3p1/2Pp1p2/3P1P2/4P3/4K1N1/8 w - - 0 1

Analysis by BrainLearn 6.0 64 POPCNT:

1.Nh4 Kd7 2.Nxg6 Ke6 3.Ne5 Nd8 4.Nf3 Kf6 5.Kd2 Ke7 6.Ne5 Kf6 7.Kc2 Ke7 8.Nd3 Nf7 9.Kd2 Ke6 10.Nc1 Kd7 11.Kd3 Ke8 12.Kc3 Nh6 13.Nd3 Nf7 14.Ne1 Kd7 15.Kb4 Ke6 16.Nf3 Kf6 17.Ne5 Nd8 18.Ka4 Kg7 19.Kb3 Kh6 20.Nf3 Kg7 21.Kc2 Nf7 22.Ne5 Nd8 23.Kd3 Kh6 24.Ke2 Kh7 25.Kf3 Kg7 26.Kf2 Kf6 27.Kg2 Kg7 28.Nf3 Kg6 29.Nh4+ Kf6 30.Kh3 Ne6 31.Nf3 Nd8 32.Nd2 Nf7 33.Kh4 Kg6 34.Nf3 Kh6 35.Kh3 Kg7 36.Kg2 Kf6 37.Nh2 Ke6 38.Kf3 Kf6 39.Nf1 Nh6 40.Nd2 Nf7 41.Ke2 Kg6 42.Nf3 Kg7 43.Nh4 Kf6 44.Kd3 Nd8 45.Kd2 Ne6 46.Kc3 Nd8
+/- (1.10) Depth: 85/94 00:05:20 9834MN
Peter.

fishpov
Posts: 44
Joined: Sat Mar 07, 2015 5:05 pm

Re: Position : Win or Draw ?

Post by fishpov » Sun Dec 15, 2019 11:09 pm

This initial position is not a black knight but a black bishop



So win or draw ?

peter
Posts: 1852
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Position : Win or Draw ?

Post by peter » Sun Dec 15, 2019 11:40 pm

peter wrote:
Sun Dec 15, 2019 8:20 pm
MikeB wrote:
Sat Dec 14, 2019 8:02 pm
How about this one:

Komodo does a little better
Best one so far Fritz17 (Ginkgo).

With empty hash:

3k4/1p3n2/1Pp3p1/2Pp1p2/3P1P2/4P3/4K1N1/8 w - - 0 1

Analysis by Fritz 17 Popcnt0:

1.Nh4 Ke7 2.Nxg6+ Kf6 3.Nh4 Nd8 4.Nf3 Kf7 5.Kd2 Kg6
+/- (1.23) Depth: 51/77 00:10:35 17396MN

After a 50 move Backward:

3k4/1p3n2/1Pp3p1/2Pp1p2/3P1P2/4P3/4K1N1/8 w - - 0 1

Analysis by Fritz 17 Popcnt0:

1.Nh4
+/= (0.32 --) Depth: 7/11 00:00:00 43kN
...
1.Nh4 Nh8 2.Nf3 Nf7 3.Ne5 Nxe5 4.dxe5 Ke7 5.Kf3 Kf7 6.Kg3
= (0.00 --) Depth: 10/13 00:00:00 156kN
...
= (0.00) Depth: 60/76 00:04:43 6937MN
Last edited by peter on Sun Dec 15, 2019 11:48 pm, edited 1 time in total.
Peter.

peter
Posts: 1852
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Position : Win or Draw ?

Post by peter » Sun Dec 15, 2019 11:43 pm

fishpov wrote:
Sun Dec 15, 2019 11:09 pm
This initial position is not a black knight but a black bishop



So win or draw ?
Draw of course, I thought that had been clear from the very start. Just let it play out Sting and it soon finds out, its eval at the starting position is simply way too high.

Michael Byrne gave the changed position which is drawn too but way more difficult for the engines to evaluate correctly with empty hash.
Peter.

User avatar
Ovyron
Posts: 3827
Joined: Tue Jul 03, 2007 2:30 am

Re: Position : Win or Draw ?

Post by Ovyron » Mon Dec 16, 2019 3:23 am

So what about this one?



Can we just keep adding useless white pawns forever but it's still always a draw?

peter
Posts: 1852
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Position : Win or Draw ?

Post by peter » Mon Dec 16, 2019 2:07 pm

Ovyron wrote:
Mon Dec 16, 2019 3:23 am
So what about this one?



Can we just keep adding useless white pawns forever but it's still always a draw?
Don't know, if you'll find more useless pawns to add, but the one you added this time, already isn't useless at all anymore of course, guess you know so, don't you?
:)

Just in case you yet aren't sure about it and you weren't kidding only:



Analysis by MateFinder 101219 64:

1.Nh4 Nh8 2.Nf3 Nf7 3.e4 fxe4 4.Nh4 Nh8 5.f5 Ke7 6.fxg6 Kf8 7.Ke3 Kg8 8.Nf5 Nxg6 9.Nd6 Kf8 10.Nxb7 Ke7 11.Nd6 Kd7 12.Nxe4 Nf8 13.Nd6 Ng6 14.Nf7 Kc8 15.Ne5 Ne7 16.g6 Kb8 17.Ke2 Kb7 18.g7 Ng8 19.Nf7 Kc8 20.Ke3 Ne7 21.Nh6 Kd7 22.g8Q Nxg8 23.Nxg8 Kc8 24.Ne7+ Kb7 25.Kd3 Kb8 26.Nxc6+ Kc8 27.Na5 Kd7 28.c6+ Kd6 29.Ke3 Ke6 30.b7 Kf5 31.b8Q Kg5 32.c7 Kh4 33.c8Q Kg5 34.Qg8+ Kf6 35.Qe5#
+- (#35) Depth: 72/96 00:50:02 125955MN
Peter.

lech
Posts: 1065
Joined: Sun Feb 14, 2010 9:02 pm

Re: Position : Win or Draw ?

Post by lech » Mon Dec 16, 2019 2:52 pm

fishpov wrote:
Fri Dec 13, 2019 9:16 pm


Win or draw
I created this postion and Sting says Ok this is win but any other engine have 0 as eval even after 2 hours

Is a Sting bug or there is really a way to win ?
I really need to know the way to win, if this is true. thanks.


Analysis by Sting 19 (based on Stockfish 2.1.1) 64bit:
...
1.Ne1
+- (2.67 ++) Depth: 52 01:04:59 50651MN
1.Ne1
+- (4.04 ++) Depth: 52 01:30:52 69036MN
1.Ne1
+- (6.09 ++) Depth: 52 01:31:43 69507MN
1.Kf2
+- (9.16 ++) Depth: 52 01:33:10 70206MN
1.Kf2
+- (13.77 ++) Depth: 52 01:36:36 72183MN
1.Ke3
+- (20.69 ++) Depth: 52 01:53:17 81565MN
I found why it happens (only in some very rare draw positions).
Probably, it will be corrected . . .
Maybe, I can't be friendly, but let me be useful.

Post Reply