is this a fortress?

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.
zenpawn
Posts: 294
Joined: Sat Aug 06, 2016 6:31 pm
Location: United States

Re: is this a fortress?

Post by zenpawn » Tue Oct 08, 2019 9:38 am

MikeB wrote:
Mon Oct 07, 2019 11:12 pm
I don't see the fortress, it seems to me that back can force a pawn off the board, either a or b pawn or the g4 pawn and once a white pawn comes off , it's over.
Agreed, though it's less about the a-pawn than the g-pawn. Once black installs its own pawn on g4, the standard f3 fortress square is gone, and it turns out the e3 square is not a sufficient alternative (due in large part to zugzwangs that prevent the white king from just shuffling).
Erin Dame
Author of RookieMonster

jhellis3
Posts: 390
Joined: Fri Aug 16, 2013 10:36 pm

Re: is this a fortress?

Post by jhellis3 » Tue Oct 08, 2019 3:04 pm

Not sure what this 'fortress code' does, but it doesn't seem to be much good.
Yikes.

MikeB
Posts: 3391
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: is this a fortress?

Post by MikeB » Tue Oct 08, 2019 6:43 pm

jhellis3 wrote:
Tue Oct 08, 2019 3:04 pm
Not sure what this 'fortress code' does, but it doesn't seem to be much good.
Yikes.
Its working great - where there is fortress ...

Code: Select all

info depth 34 seldepth 102 multipv 1 score cp 8041 nodes 334138924 nps 19157145 hashfull 330 tbhits 0 time 17442 pv h2e2 d8g8 e2f2 d7e7 f2h2 e7e8 h2d2 g8g7 c1b2 g7g6 b2a1 g6g7 e1c1 g7g6 d2d1 e8b8 c1d2 g6g7 d1e1 g7g8 e1g1 b8e8 d2e1 g8g6 e1f1 g6g7 a1a2 e8g8 f1e2 g8e8 g1f1 g7g6 g3g1 g6g7 f1f2 g7g6 e2f1 g6g7 f2e2 e8b8 f1e1 g7g8 a2a1 b8e8 e1c1 e8b8 c1f1 b8e8 c2c4 b4c3 f1c1
info depth 35 currmove h2e2 currmovenumber 1
<snip>
info depth 35 currmove e1b4 currmovenumber 28
info depth 35 currmove g3g4 currmovenumber 29
info depth 35 seldepth 102 multipv 1 score cp 8107 lowerbound nodes 475224698 nps 20229214 hashfull 492 tbhits 0 time 23492 pv g3g4
info depth 34 currmove g3g4 currmovenumber 1
info depth 35 seldepth 102 multipv 1 score cp 8172 lowerbound nodes 484004860 nps 20277550 hashfull 498 tbhits 0 time 23869 pv g3g4
info depth 33 currmove g3g4 currmovenumber 1
info depth 35 seldepth 102 multipv 1 score cp 8261 lowerbound nodes 487817413 nps 20301194 hashfull 501 tbhits 0 time 24029 pv g3g4
info depth 32 currmove g3g4 currmovenumber 1
info depth 35 seldepth 102 multipv 1 score cp 8379 lowerbound nodes 491942345 nps 20329036 hashfull 503 tbhits 0 time 24199 pv g3g4
info depth 31 currmove g3g4 currmovenumber 1
info depth 35 seldepth 102 multipv 1 score cp 8525 lowerbound nodes 496600254 nps 20357475 hashfull 506 tbhits 0 time 24394 pv g3g4
info depth 30 currmove g3g4 currmovenumber 1
info depth 35 seldepth 102 multipv 1 score cp 8702 lowerbound nodes 500302612 nps 20378925 hashfull 510 tbhits 0 time 24550 pv g3g4
info depth 29 currmove g3g4 currmovenumber 1
info depth 35 seldepth 102 multipv 1 score cp 8904 lowerbound nodes 519741529 nps 20492115 hashfull 529 tbhits 0 time 25363 pv g3g4

info depth 35 seldepth 102 multipv 1 score cp 8804 nodes 574309454 nps 20742179 hashfull 580 tbhits 0 time 27688 pv g3g4 h5g4 h4h5 d8h8 e1h4 d7f7 h4g5 e6d7 g5g6 d7c7 h5h6 c7b6 h6h7 f7a7 h2h6 a7c7 c1d2 g4g3 d2e1 c7e7 e1f1 g3g2 f1g1 e7c7 g6h5 c7e7 h5h3 e7g7 h3f5 g7f7 f5h5 h8h7 h6h7 f7f4 h5d1 f4f1 d1f1 g2f1r g1f1 d5g8 h7h6 c5c4 d3c4 d4d3
info depth 36 currmove g3g4 currmovenumber 1

info depth 36 seldepth 102 multipv 1 score cp 8804 nodes 614643276 nps 20956843 hashfull 612 tbhits 0 time 29329 pv g3g4 h5g4 h4h5 d8h8 e1h4 d7f7 h4g5 e6d7 g5g6 d7c7 h5h6 c7b6 h6h7 f7a7 h2h6 a7c7 c1d2 g4g3 d2e1 c7e7 e1f1 g3g2 f1g1 e7c7 g6h5 c7e7 h5h3 e7g7 h3f5 g7f7 f5h5 h8h7 h6h7 f7f4 h5d1 f4f1 d1f1 g2f1r g1f1 d5g8 h7h6 c5c4 d3c4 d4d3
bestmove g3g4 ponder h5g4

peter
Posts: 1778
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: is this a fortress?

Post by peter » Tue Oct 08, 2019 11:27 pm

MikeB wrote:
Mon Oct 07, 2019 2:26 pm
11 Cores each with pondering on, 8192M hash , game 5 min with 5 sec inc.
Lc0 scoring - 70% is under two pawns, 100% is two queens advantage. nice queen sac at the end to secure the win
I don't see how Black has any more progress, if White gives up the a-pawn only, keeping the g-pawn.
Here I had another game for Crystal- Honey with Black but this time against Cfish.
12 threads of a 12x3GHz CPU, 8Gb hash each, ponder on:

Peter.

jhellis3
Posts: 390
Joined: Fri Aug 16, 2013 10:36 pm

Re: is this a fortress?

Post by jhellis3 » Wed Oct 09, 2019 9:32 pm

While I don't know the correct result of the position, in my own analysis with the latest revision of Crystal, I found it (and the SF mod with FD) deviates from the line posted above at move 37, preferring ...Kb7 instead (or a transposition to the same position).

My own human intuition tells me the position should likely be a draw with the establishment of a g-pawn & rook fortress vs queen.

However, through analysis, I was able to achieve the following positions:











The final position being Mate in 25 (or maybe less).

Obviously, the ending position being lost is certain. And I am fairly confident that black can win both the queenside pawns (though not 100% certain). I'm less confident of the positions between that and the final one though.

jhellis3
Posts: 390
Joined: Fri Aug 16, 2013 10:36 pm

Re: is this a fortress?

Post by jhellis3 » Wed Oct 09, 2019 9:42 pm

I'm now quite confident that the penultimate position is also won, and that it is quite achievable from the antecedent position. Therefore, the main question comes down to whether white can hold on to the b pawn from here:


If White can't hold the b pawn, I suspect it likely doomed.

peter
Posts: 1778
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: is this a fortress?

Post by peter » Wed Oct 09, 2019 10:10 pm

jhellis3 wrote:
Wed Oct 09, 2019 9:42 pm
If White can't hold the b pawn, I suspect it likely doomed.
I think you're wrong, Joseph.

Here I have another playout down to a drawn 7 men position. Even without White's a and b- pawn the fortress did not fall:


Position after 84...Kxg4 is Losomosov draw.

Started with 90'+30" again, lowering then to 75'+15" from 50th onwards, 8G hash, 24 threads of 12x3GHz.
Peter.

jhellis3
Posts: 390
Joined: Fri Aug 16, 2013 10:36 pm

Re: is this a fortress?

Post by jhellis3 » Wed Oct 09, 2019 11:10 pm

Of course, one is free to think whatever he or she chooses; I'm just interested in the objectively correct outcome of the position.

And I have no idea what engines someone else is using, but I strongly suspect that black giving up two pawns for no apparent compensation is far from optimal play.....

One can grab the latest Crystal source at https://github.com/jhellis3/Stockfish/tree/crystal

Post Reply