Unstoppable vs Untouchable pawns in a pawnrace?

[jesper_nielsen]

A question about chess. Though I am not sure this is the appropriate forum for it!

I am currently analysing the many (MANY!´) Pupsi losses from the recent WBEC tournament. And a lot of the losses comes from trading down into a lost pawn endgame.

So better pawn endgame evaluation is definitely on the todo list.

During the analysis I came across this position:

[d]8/2p1k2p/4p3/P5p1/2K5/6P1/P6P/8 b - - 0 37

And my feeble chess mind screams: "This is clearly won for black!"

Reasoning:
The black king can catch the passed white A pawn.

The white King can stop the passed black B and D pawns, but they are safe from him. They are "untouchable" for lack of a better term.

The white king is tied up to the stopping of the black passed pawns, and then the black king can catch and take the white passed pawn, making this is clearly won pawn endgame.

So the question: Is my analysis correct?

With careful play be black, I can find no way for the white king to take either of the black pawns.

Can this be made into a correct static evaluation of the position?
Have anybody tried something like it?

Kind regards,
Jesper

[michiguel]

A question about chess. Though I am not sure this is the appropriate forum for it!

I am currently analysing the many (MANY!´) Pupsi losses from the recent WBEC tournament. And a lot of the losses comes from trading down into a lost pawn endgame.

So better pawn endgame evaluation is definitely on the todo list.

During the analysis I came across this position:

[d]8/2p1k2p/4p3/P5p1/2K5/6P1/P6P/8 b - - 0 37

And my feeble chess mind screams: "This is clearly won for black!"

Reasoning:
The black king can catch the passed white A pawn.

The white King can stop the passed black B and D pawns, but they are safe from him. They are "untouchable" for lack of a better term.

The white king is tied up to the stopping of the black passed pawns, and then the black king can catch and take the white passed pawn, making this is clearly won pawn endgame.

So the question: Is my analysis correct?

With careful play be black, I can find no way for the white king to take either of the black pawns.

Can this be made into a correct static evaluation of the position?
Have anybody tried something like it?

Kind regards,
Jesper

White is lost. Unfortunately, it is very difficult to evaluate this statically.

Miguel

[zullil]

A question about chess. Though I am not sure this is the appropriate forum for it!

[d]8/2p1k2p/4p3/P5p1/2K5/6P1/P6P/8 b - - 0 37

And my feeble chess mind screams: "This is clearly won for black!"

So the question: Is my analysis correct?

Kind regards,
Jesper

Stockfish-1.6.2 (8 threads) seemed to agree with you, just before a stray cosmic ray killed it.

Code: Select all

info depth 29 score cp 1175 time 339929 nodes 3430811209 nps 10092728 pv e7d7 a5a6 d7c6 a6a7 c6b7 c4d4 b7a7 d4c4 a7a6 c4d3 a6a5 d3c4 a5a4 c4d4 a4a3 d4e3 a3a2 e3d4 a2b3 d4c5 g5g4 c5d4 e6e5 d4e5 c7c5 e5f6 c5c4 f6g7 c4c3 g7h7 c3c2 h7g6 c2c1q g6f5 c1d1 f5e4 d1e2 e4f5 
info currmove e6e5 currmovenumber 2
info currmove e7d6 currmovenumber 3
info nodes 3436912170 nps 10083859 time 340833 hashfull 999
info currmove h7h5 currmovenumber 4
info currmove e7f7 currmovenumber 5
info currmove c7c5 currmovenumber 6
info nodes 3446056030 nps 10079488 time 341888 hashfull 999
info currmove h7h6 currmovenumber 7
info currmove e7f6 currmovenumber 8
info currmove c7c6 currmovenumber 9
info currmove g5g4 currmovenumber 10
info currmove e7e8 currmovenumber 11
info currmove e7d8 currmovenumber 12
info currmove e7f8 currmovenumber 13
info depth 30
info currmove e7d7 currmovenumber 1
info nodes 3455170904 nps 10076645 time 342889 hashfull 999
info nodes 3463642990 nps 10071510 time 343905 hashfull 999
info nodes 3471902099 nps 10065643 time 344926 hashfull 999
info nodes 3481406276 nps 10063962 time 345928 hashfull 999
info nodes 3490849080 nps 10061794 time 346941 hashfull 999
info nodes 3500276070 nps 10059941 time 347942 hashfull 999
info nodes 3510342216 nps 10059526 time 348957 hashfull 999
info nodes 3521064565 nps 10061219 time 349964 hashfull 999
info nodes 3532251877 nps 10063940 time 350981 hashfull 999
info nodes 3542753719 nps 10065012 time 351987 hashfull 999
info nodes 3554193853 nps 10068566 time 352999 hashfull 999
info nodes 3565376436 nps 10071457 time 354008 hashfull 999
info nodes 3576068197 nps 10073062 time 355013 hashfull 999
info nodes 3586267329 nps 10073303 time 356017 hashfull 999
info nodes 3596975039 nps 10074883 time 357024 hashfull 999
info nodes 3607733879 nps 10076765 time 358025 hashfull 999
info nodes 3618795141 nps 10079142 time 359038 hashfull 999
info nodes 3629171456 nps 10079967 time 360038 hashfull 999
info nodes 3639457682 nps 10080287 time 361047 hashfull 999
info nodes 3650443014 nps 10082481 time 362058 hashfull 999
info nodes 3661220775 nps 10084285 time 363062 hashfull 999
info nodes 3671617151 nps 10084782 time 364075 hashfull 999
info nodes 3682220808 nps 10085899 time 365086 hashfull 999
info nodes 3693446409 nps 10089012 time 366086 hashfull 999
info nodes 3704067413 nps 10090379 time 367089 hashfull 999
info nodes 3715126185 nps 10092462 time 368109 hashfull 999
info nodes 3724207918 nps 10089696 time 369110 hashfull 999
info nodes 3733679130 nps 10087752 time 370120 hashfull 999
info nodes 3743385482 nps 10086426 time 371131 hashfull 999
info nodes 3754450063 nps 10088702 time 372144 hashfull 999
info nodes 3765159704 nps 10090177 time 373151 hashfull 999
info nodes 3775988541 nps 10091935 time 374159 hashfull 999
info nodes 3787429900 nps 10095128 time 375174 hashfull 999
info nodes 3798429990 nps 10097185 time 376187 hashfull 999
info nodes 3809368776 nps 10099336 time 377190 hashfull 999
Segmentation fault

[Michel]

Normally the outside passed pawn wins since black needs time to catch it while
white can eat the other black pawns.

However here white cannot eat the the other black pawns because there are two passed pawns.

It does not seem impossible to detect the relevant patterns.

[UncombedCoconut]

Stockfish-1.6.2 (8 threads) seemed to agree with you, just before a stray cosmic ray killed it.

Code: Select all

info nodes 3809368776 nps 10099336 time 377190 hashfull 999
Segmentation fault

Irreproducible bugs are always the most fun... I've never gotten SF to do that!