Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old December 4th 04, 07:13 PM
fhub
 
Posts: n/a
Default Problems with Arena/Chest/Popeye

(Doctor SBD) wrote in message ...

What you´re writing really seems more than strange, since I have absolutely
no problems here with both programs (and that´s also true for all others
using them, that I know).

I have a lot of trouble with using Arena with Chest and Popeye (UCI).


Ok, I assume you´re using my latest versions (ChestUCI 3.7 / PopeyeUCI 1.4),
but what´s your Arena version?
If you still have version 1.0, then you should upgrade (although I can´t
imagine, that this would be the reason for your problems).
Probably the best current version of Arena is 1.092 - if you don´t have
this version (or can´t get it anywhere), I could send it to you by email.

With Chest, I get a lot of incorrect solutions in helpmates (it gives some
of the correct solutions, but may miss numerable cooks).


Incorrect solutions? Absolutely impossible! Give me some examples!
For helpmates Chest isn´t the fastest one (because it has no special tricks
for this problem type), but I´ve never seen it producing wrong solutions.

Popeye rarely works, ...


What does this mean? What does Popeye do, if it ´does not work´?

... and both programs give me continual error messages about "incorrect FEN"
(which can somehow oddly be fixed by mirroring the position and mirroring it
back again).


Both program only state an "incorrect FEN", if it _is_ actually _incorrect_!
Furthermore this ´problem´ seems more likely to be a problem of the GUI
(i.e. of Arena), if at all.
I´m really wondering, how double-mirroring could solve this problem!?
Again - please give some FEN-examples here, where you get those errors!

Sometimes I get data write errors, and these are all legal positions.


Probably you mean the message "ERROR: File Write Error !".
This error is only reported by ChestUCI, if it can´t write its problem file
´ChestPro.txt´ - a reason for this could be, that you currently have this
file ´opened´ (maybe with a text editor) or if it´s ´write protected´, but
both cases seem quite unlikely!?

I have loaded Arena and the two solving programs on different computers,
always with the same results.


Since really a _lot_ of people are using these programs (without any problems),
I´m quite sure that there´s something wrong with your general installation
of Arena and/or those mate engines.
And of course with such little informations you´re giving it´s not easy to
really help you a lot - so I´m waiting for your answer with (hopefully)
some concrete examples.

Best regards,
Franz.
  #2   Report Post  
Old December 4th 04, 10:36 PM
Doctor SBD
 
Posts: n/a
Default

Incorrect solutions? Absolutely impossible! Give me some examples!

I entered the position of White: Kc5, Bd5, Nb3, Pe3, Black, Ka3, Ra4,Nb4, Pa5,
Pe4, and it shows 3 solutions (starting with the BN moving to d3, c2, and a2),
all model mates. No others are given.

I could give more, but I really have basically given up on these programs ever
giving me correct solutions in helpmates. Now I have no problem with any other
kind of mate in correctness, although the incorrect FEN does crop up often, and
again I just fix it by mirroring.

But with helpmates, I use the Arena/Chest duo for speed, but check for
soundness using Fritz/Mate 2.13.

I do not think I am the only one with these problems - one editor told me that
he occasionally gets incorrect problems marked C+ by the author(s), and you can
guess from what program.

What does this mean? What does Popeye do, if it ´does not work´?


It sits and does nothing. I call this "not working". I also get this Datei-
Schreib-Fehler or File write error very often with Popeye. Popeye is especially
aggravating with this error, and some days I cannot get it to do even one
problem.

I´m really wondering, how double-mirroring could solve this problem!?


I do not know, but if you came and looked at any of the computers I have used
these on, it does work. I will mirror to black, then to white, and the problem
will be fixed (for that problem only). Once it starts doing it (and it cannot
be predicted), you will have to continuously mirror positions. But on the next
start-up, it may work OK for awhile before proceeding with these errors.

With my installations, all I did was download and follow all instructions. And
I have wiped both computers clean of the programs and started from stratch, and
had the same problem again and again.

And yes, I am using the latest versions. I read instructions carefully, and
have followed them to the letter.

SBD


  #3   Report Post  
Old December 4th 04, 10:59 PM
Joost de Heer
 
Posts: n/a
Default

I entered the position of White: Kc5, Bd5, Nb3, Pe3, Black, Ka3, Ra4,Nb4,
Pa5, Pe4, and it shows 3 solutions (starting with the BN moving to d3, c2,
and a2), all model mates. No others are given.


Stipulation?

If it's a h#2, there's no solution, if it's a h#3, it's horribly cooked.

I do not think I am the only one with these problems - one editor told me
that he occasionally gets incorrect problems marked C+ by the author(s), and
you can guess from what program.


I've never seen Popeye give wrong solutions, except in cases where there's
ambiguity in a fairy definition (like rebirth squares of a locust in anticirce,
or the combination madrasi+marscirce, but I doubt you're using those....)

I don't use Chest, as I hardly ever compose orthodox things.

Joost
--
Du hast mein Herz zerrissen, meine Seele geraubt
Das es so enden würde hätt` ich nie geglaubt [Aus der Ruinen -]
Ohne Rücksicht auf Verluste, hast Du meine Welt zerstört [L'Âme Immortelle]
Eine Welt, die vor kurzem nur uns beiden hat gehört
  #4   Report Post  
Old December 5th 04, 09:11 AM
Anders Thulin
 
Posts: n/a
Default

Joost de Heer wrote:

If it's a h#2, there's no solution, if it's a h#3, it's horribly cooked.


How is that relevant?

I can imagine essentially two cases he a) the program claims to give
a complete solution, or b) the programs claims only to solve problems that
follow current Codex (or some similar canon).

In the first place, it's a bug to leave out any line of a solution.
No dicussion here.

In the second, it's perfectly all right to give only the first solution
found -- but the program is not useful for testing problems unless you know
by other means that the problem conforms to Codex. Now helpmates are tricky
-- they're the only problem class I know where multiple solutions are allowed,
but they're expected to be stated in the stipulation. If a solving program
doesn't have that kind of stipulation form, it simply must give all solutions
to helpmates.

Unfortunately, very few programs authors seem willing to state what their
programs actually should do, and, if there are known shortcomings, what
those are. (And no, a user shouldn't have to read source code for that.)

If there was some way to compare the solutions from one program with
those from another, it would help a lot -- but that would require that
some kind of 'standard solution format' to be formulated and implemented.
It wouldn't help immediately to establish if a program gives a complete
solution, but it would help establishing if there are any differences between
two programs, and what they are. And by comparing, say, known 'true'
solutions (say, dump certain solutions from various endgame databases),
there would be some ground for knowing how much to trust the programs
in question. It would also be great testing material for anyone trying
to create a new solving program. (Actually ... I can see a kind of weird
informal problem tourney he for problems that demonstrate errors in
solving programs.)

I do not think I am the only one with these problems - one editor told me
that he occasionally gets incorrect problems marked C+ by the author(s), and
you can guess from what program.


I've never seen Popeye give wrong solutions, except in cases where there's
ambiguity in a fairy definition (like rebirth squares of a locust in anticirce,
or the combination madrasi+marscirce, but I doubt you're using those....)


There is at least one known case where Popeye can't give complete solutions to
plain, orthodox mates, and so may report fewer solutions. It's not well documented
-- I learnt about it only after reporting the case as a bug after running into
it in one of the early problem books I made, and it part of the reason why I'm
using Problemist for full solutions. It's unfortunate that there's no list of these
bugs ...

--
Anders Thulin ath*algonet.se http://www.algonet.se/~ath
  #5   Report Post  
Old December 5th 04, 09:51 AM
Doctor SBD
 
Posts: n/a
Default

It's unfortunate that there's no list of these
bugs ...


I am glad to see Mr. Thulin, well-known in the problem world for his
contributions, to confirm that there are indeed some bugs in some of these
programs.

Another bug that happened with me, happened with Chest. In a helpmate, where
there where two solutions, promoting to a bishop and promoting to a rook (the
position was *something like* - this is from human memory only, Wkh6, Ng3, Ng1;
B Kh8, Pg2 -this was for practice only).

After 1... Nf1 there should be two ways to promote to effect the helpmate, gxfR
and gxfB. Then white plays Ne5 and mates on g6 or f7, depending on the
promotion (R or B head back to g8).

But Chest could only find the one solution in this case (promotion to rook) and
when I tried to force gxf1(B), it immediately set the position to 1/2-1/2
(insufficient material). This is strange, as white has no more "sufficient
material to mate" than before, but since Black does, I suppose it can find that
solution.

But again, this is only from memory, so details of the exact circumstances may
differ a bit from what has occurred. But in this case Mate 2.13 from Fritz
failed me as well, but not through the same means! It simply recognized no
promotions as leading to mate, and came up with a bunch of longer cooks.

Again, it could well be something I have done, but it may also be the program,
as well as how I am handling it.

SBD


  #6   Report Post  
Old December 5th 04, 10:53 AM
Anders Thulin
 
Posts: n/a
Default

Doctor SBD wrote:

But Chest could only find the one solution in this case (promotion to rook) and
when I tried to force gxf1(B), it immediately set the position to 1/2-1/2
(insufficient material). This is strange, as white has no more "sufficient
material to mate" than before, but since Black does, I suppose it can find that
solution.


Could that be an Arena/Chest problem? It seems as if 'chess playing'
knowledge had been injected, and that doesn't sound like pure Chest.

I've seen something vaguely similar with Problemist, involving short
and long notation: in long notation, the solution contained

2. ... Q7xa4 3. Rd6#
2. ... Q1xa4 3. Rd2#

but with 'short notation' enabled, only one appeared as

2. ... Qxa4 3. Rd6#

(I think this has been fixed, though. The problem is:
Andrews: Q3r3/3q3b/1p6/2p1RpB1/2pk1rpK/6Pp/BP1pR3/8 #3)

As I don't use Popeye for solution details (only for initial soundness
testing), I have no examples handly, but I have an old letter from Norbert
Geissler documenting a problem:

"It's right, that in a major solving tourney a problem was tested
by the judge with Popeye and it found three variations.
But the solvers in the tourney found four variations, which confused
the director (he had computer tested it, right ?).
Since Popeye doesn't consider defenses which differ only at move three
or later to the threat they should defend, this happened."

And one thing few program authors bother about are identifying short
solutions: typically you have to wade through all possible lines
yourself. (Problemist does the right thing he very easy to distinguish
between no sol., multiple sol., and short sol.)

I've been thinking of creating test files that, along with some auxiliary
software, could be used to evaluate solving programs, mainly as a help
to build up trust in a new solving engine (or a new release of one)
without having to test several hundred positions manually.

Popeye and Chest could probably be matched with a bit of work ... may be more
difficult with pure Windows/programs like Problemist or something like Alybadix.


--
Anders Thulin ath*algonet.se http://www.algonet.se/~ath
  #7   Report Post  
Old December 5th 04, 11:18 AM
fhub
 
Posts: n/a
Default

Ok, let me start with your helpmate ´problem´:

I entered your following position (with stipulation h#3):
8/8/8/p1KB4/rn2p3/kN2P3/8/8 b - -

Here´s the solution file of Popeye:

1 (Ka3*b3 Time = 0.050 s)
2 (Ka3-a2 Time = 0.440 s)
3 (Ka3-b2 Time = 1.040 s)
1.Sb4-a2 Bd5-c6 2.Ra4-b4 Sb3-d2 3.Rb4-b3 Sd2-c4 #
1.Sb4-a2 Bd5-c6 2.Ra4-b4 Sb3*a5 3.Rb4-b3 Sa5-c4 #
1.Sb4-a2 Kc5-c6 2.Sa2-c3 Sb3-a1 3.Ka3-b4 Sa1-c2 #
1.Sb4-a2 Kc5-c6 2.Sa2-c3 Sb3-d4 3.Ka3-b4 Sd4-c2 #
1.Sb4-a2 Kc5-c6 2.Ka3-b4 Sb3-a1 3.Sa2-c3 Sa1-c2 #
1.Sb4-a2 Kc5-c6 2.Ka3-b4 Sb3-d4 3.Sa2-c3 Sd4-c2 #
1.Sb4-a2 Kc5-b6 2.Sa2-c3 Sb3-a1 3.Ka3-b4 Sa1-c2 #
1.Sb4-a2 Kc5-b6 2.Sa2-c3 Sb3-d4 3.Ka3-b4 Sd4-c2 #
1.Sb4-a2 Kc5-b6 2.Ka3-b4 Sb3-a1 3.Sa2-c3 Sa1-c2 #
1.Sb4-a2 Kc5-b6 2.Ka3-b4 Sb3-d4 3.Sa2-c3 Sd4-c2 #
1.Sb4-a2 Sb3-a1 2.Sa2-c3 Kc5-c6 3.Ka3-b4 Sa1-c2 #
1.Sb4-a2 Sb3-a1 2.Sa2-c3 Kc5-b6 3.Ka3-b4 Sa1-c2 #
1.Sb4-a2 Sb3-d2 2.Ra4-b4 Bd5-c6 3.Rb4-b3 Sd2-c4 #
1.Sb4-a2 Sb3-d4 2.Sa2-c3 Kc5-c6 3.Ka3-b4 Sd4-c2 #
1.Sb4-a2 Sb3-d4 2.Sa2-c3 Kc5-b6 3.Ka3-b4 Sd4-c2 #
1.Sb4-a2 Sb3*a5 2.Ra4-b4 Bd5-c6 3.Rb4-b3 Sa5-c4 #
4 (Sb4-a2 Time = 9.390 s)
1.Sb4-c2 Kc5-c6 2.Ka3-b4 Sb3-c1 3.Sc2-a3 Sc1-a2 #
1.Sb4-c2 Kc5-b6 2.Ka3-b4 Sb3-c1 3.Sc2-a3 Sc1-a2 #
5 (Sb4-c2 Time = 10.820 s)
1.Sb4-d3 + Kc5-b5 2.Sd3-b2 Sb3-a1 3.Ra4-d4 Sa1-c2 #
1.Sb4-d3 + Kc5-b5 2.Sd3-b2 Sb3-d2 3.Ra4-d4 Sd2-b1 #
1.Sb4-d3 + Kc5-b5 2.Ra4-d4 Sb3-a1 3.Sd3-b2 Sa1-c2 #
1.Sb4-d3 + Kc5-b5 2.Ra4-d4 Sb3-d2 3.Sd3-b2 Sd2-b1 #
1.Sb4-d3 + Kc5-b5 2.Ra4-d4 Sb3*d4 3.Sd3-b2 Sd4-c2 #
6 (Sb4-d3 + Time = 13.510 s)
7 (Sb4*d5 Time = 13.670 s)
8 (Sb4-c6 Time = 14.000 s)
9 (Sb4-a6 + Time = 14.060 s)

And here´s the output of ChestUCI:

HelpMate in 3 found ! (3 Solutions in 00:00)
3/3 00:01 -M3 1. ... Nc2 2.Kc6 Kb4 3.Nc1 Na3 4.Na2+
3/3 00:01 -M3 1. ... Na2 2.Nxa5 Rb4 3.Bc6 Rb3 4.Nc4+
3/3 00:01 -M3 1. ... Nd3+ 2.Kb5 Nb2 3.Na1 Rd4 4.Nc2+

So where is here any problem?
ChestUCI gives the same _main_ solutions as Popeye - but:
Of course ChestUCI does NOT (by default!) give ALL POSSIBLE variations!
In it´s default setting ChestUCI only gives all _different_ PVs, i.e.
all main lines with a different _first_ key move!
I´ve choosen this behaviour, because otherwise the complete solution tree
would be really _huge_ for most problems.
If you would like to see the _complete_ solution, you could select
´EngineMode=Expert´ in ChestUCI´s configuration AND remove the ´U´ in the
´ExpertOptions´ (so they should be ´-Sr´) - in this case you can find
really ALL solution variants in the solution file of ChestUCI, which can
be displayed by selecting ´Show Solution´ in the engine configuration menu.

You can see - there´s absolutely nothing wrong with ChestUCI, but of course
(as the name ´Expert mode´ says) this needs _some_ experience with Chest ...

As for your other problems (illegal FEN, file write error, ...):
Sorry, but I really have no idea how I could help you! :-(
As I already said there are hundreds of users having ansolutely no problems
with both of these programs, so it´s clear (at least for me), that it´s
definitely not the fault of ChestUCI/PopeyeUCI.
Maybe there´s any incompatibility with your OS, with any system DLLs, with
programs running in the background ... - I don´t know!
I´m quite sure, that I would be able to solve your problems within minutes,
if I would sit in front of your PC - but of course I can´t give any diagnosis
from a distance, and so I see no chance here in this forum.

Regards,
Franz.
  #8   Report Post  
Old December 5th 04, 01:03 PM
fhub
 
Posts: n/a
Default

Joost de Heer wrote in message ...
I entered the position of White: Kc5, Bd5, Nb3, Pe3, Black, Ka3, Ra4,Nb4,
Pa5, Pe4, and it shows 3 solutions (starting with the BN moving to d3, c2,
and a2), all model mates. No others are given.


Stipulation?

If it's a h#2, there's no solution, if it's a h#3, it's horribly cooked.


Hello Joost,

sorry for my dumb question, but since I´m not a native English speaker
maybe I don´t understand this ´cooked´ correctly.

I´ve always thought, a problem is ´cooked´ if there exists a _shorter_
solution than intended (or stated in the problem)!?

But in the helpmate example above the is no h#2 solution at all -
it´s actually a h#3 as demanded.
So why on earth should this problem be ´horribly cooked´?

Regards,
Franz.
  #9   Report Post  
Old December 5th 04, 01:35 PM
Doctor SBD
 
Posts: n/a
Default


And here´s the output of ChestUCI:

HelpMate in 3 found ! (3 Solutions in 00:00)
3/3 00:01 -M3 1. ... Nc2 2.Kc6 Kb4 3.Nc1 Na3 4.Na2+
3/3 00:01 -M3 1. ... Na2 2.Nxa5 Rb4 3.Bc6 Rb3 4.Nc4+
3/3 00:01 -M3 1. ... Nd3+ 2.Kb5 Nb2 3.Na1 Rd4 4.Nc2+

So where is here any problem?


If one is composing, this is a big problem.
If one is simply solving, this is a big problem.

Both composer and solver need to see all solutions, not just the 3 best. For
example, I cannot be assured, using Chest/UCI, that my problem is
computer-tested to the rigor that a judge or editor would expect (unless I do
as you note, and if that is the case, this should be noted to all problemists
using the software)!

And yes, I found that "show solution" option that shows all solutions - but
should that be an option or mandatory?

SBD
  #10   Report Post  
Old December 5th 04, 03:26 PM
Joost de Heer
 
Posts: n/a
Default

sorry for my dumb question, but since I´m not a native English speaker
maybe I don´t understand this ´cooked´ correctly.

I´ve always thought, a problem is ´cooked´ if there exists a _shorter_
solution than intended (or stated in the problem)!?


Technically, this composition is dualistic, if 3 solutions were intended
(A cook is a solution with a different 1st move, a dual is a solution
where deviation from the intended solution happens later in the solution)

See the Codex (http://www.sci.fi/~stniekat/pccc/codex.htm):

Article 9 - Cook

(1) Subject to paragraph (2), a chess composition is called cooked if it
has a solution that differs in its first move from the author's solution.

Article 10 - Dual

(1) Subject to paragraph (2), a dual is said to occur if, after the
first move, there is more than one method of satisfying the stipulation.

Joost
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Women's Olympic Team - - Problems Just Starting Richard Van Riper rec.games.chess.politics (Chess Politics) 0 March 28th 04 10:05 PM
Positional problems books kempelen rec.games.chess.analysis (Chess Analysis) 15 September 27th 03 07:51 PM
Collection of my Diary Problems Fernando rec.games.chess.analysis (Chess Analysis) 4 September 10th 03 12:26 AM
Problems with Frank Niro Sam Sloan rec.games.chess.politics (Chess Politics) 6 July 22nd 03 08:44 PM
Problems with Frank Niro Sam Sloan rec.games.chess.misc (Chess General) 1 July 21st 03 08:13 PM


All times are GMT +1. The time now is 02:04 PM.

Powered by vBulletin® Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
Copyright ©2004-2018 ChessBanter.
The comments are property of their posters.
 

About Us

"It's about Chess"

 

Copyright © 2017