Website founded by Milan Velimirović in 2006
23:28 UTC
| |
MatPlus.Net Forum General Isardam and en passant |
|
|
|
You can only view this page!
| Page: [Previous] [Next] 1 2 | (1) Posted by Bojan Basic [Friday, Dec 17, 2010 04:38] | Isardam and en passant How does en passant work in Isardam? That is, if moves that lead to Madrasi paralysis are forbidden, does that also hold for moves which lead to non-mutual Madrasi paralysis?
I assume that it holds (since I see no other logical interpretation). However, I was surprised to find a few problems in PDB which exploit e.p. in Isardam: P1067495, P1067550, P1067550, P1067606, P1067662, P1067773, P1068273, P1068318.
Popeye behaves very strange concerning this question. Namely, for wPd5, bPe7, Popeye permits 1.... e7-e5, but after that lists 2.d5-d6 and 2.d5*e6 e.p. as the only legal moves (that is—no other white piece may move). This makes no sense to me. | | (2) Posted by Cornel Pacurar [Friday, Dec 17, 2010 06:44] | @ Bojan RE: "How does en passant work in Isardam?"
It depends who you ask - Popeye or Winchloé's programmers! :) Our French chess problem friends are also debating this on their france-echecs forum, at the link provided earlier by Axel Gilbert (http://www.france-echecs.com/index.php?mode=showComment&art=20101216082147515).
Axel Gilbert
Donc avec PBb5 et PNa7, a7-a5 est illégal car le pion a5 serait paralysé.
Nicolas Dupont
Avec PBb5 et PNa7, il autorise a7-a5.
Nicolas Dupont
et Winchloé, il accepte a7-a5 avec un PNb5 ?
Axel Gilbert
Avec PNb5, Winchloé n'accepte pas a7-b5.
Winchloé et Popeye ne sont donc pas d'accord, c'est fâcheux.
En l'occurrence, je pense que c'est Chloé qui a raison... | | (3) Posted by Kevin Begley [Friday, Dec 17, 2010 13:32]; edited by Kevin Begley [10-12-17] | Win Chloe 3.06 also has an "Isardam Type B" (as well as: "Isardam Rex Inclusiv" & "Isardam type B Rex Inclusiv").
However, in none of these cases am I able to produce (reproduce?) an en passant capture (nor can I simply "pass" a pawn).
Have I misread the previous post, or was it not suggested that Win Chloe permitted en passant in this condition?
[edit: it seems I did misread... surely Win Chloe is correct: no en passant can be possible in Isardam.]
[aside: I would appreciate if somebody would please provide the rules for Isardam Type B.]
Oops, I stand corrected (again) -- I found the following in the WC database:
Jonathan Mestel & Alan Bell
Special Prize, Champagne Ty., Moscow, 2003
(= 14+13 )
PG12.5 Isardam
1.a4 c6 2.a5 b5 3.a6 Bb7 4.axb7 Sa6 5.b8=B Rxb8 6.d4 Rb6 7.Bf4 Qa8 8.Bb8 Sxb8 9.h4 a6 10.h5 g5 11.h6 Bg7 12.hxg7 h6 13.gxh8=B
Now, I am really confused!
I do not see the legal justification for 2...b5, or 10...g5.
The essential question: does Isardam prevent moves which would result in paralysis of my own unit, or of my opponent's unit (or both)?
Do we require an Isardam Type C? If so, then by combination (of Types B & C), we may require a Type D, as well. | | (4) Posted by Joost de Heer [Friday, Dec 17, 2010 19:39] | IIRC, in Isardam B, the Isardam rule doesn't apply to checks (so wRa5, bKb5, bRc5 is no check in Isardam, but is a check in Isardam B). | | (5) Posted by Axel Gilbert [Friday, Dec 17, 2010 22:38] | To sum up the "french" debate about en passant...
with BPa7 and WPb5, is a7-a5 legal ?
It leads to a position where WP threatens BP, therefore it should be illegal. Winchloe seems to understand it this way.
Maybe popeye states that only one pawn is "watched", so there's no "mutual" Madrasi paralysis then the move is legal. The en passant is legal after this (no reason to refuse it). Why Popeye can't take a move by another piece, I can't imagine.
Reminder : in Madrasi, with BPa7, WPb5, WKb4 1... a5 does not give check (the pawn is considered paralysed immediately after playing its move), but black can't ignore the threat : after a random move, a5 un "unparalysed" (b5 has lost is right to take en passant) and could take the king. It's a common trick in stalemate problems. | | (6) Posted by Kevin Begley [Saturday, Dec 18, 2010 11:17]; edited by Kevin Begley [10-12-18] | Thanks Joost... I assumed that was the difference, but wanted confirmation.
@Axel,
It seems to me that en passant must be considered legal, by default, in Isardam.
I concede this rule is counter-intuitive (and may add considerable complexity to an already hyper-complex condition).
If you assume otherwise, as the default position, you risk being cooked by this move.
A number of problems already make use of en passant, and at least one adds material to avoid the possibility of an EP claim against it.
Therefore, if 1...a5 is (by default) legal (in the diagram below), we must conclude one of two things:
CASE 1) Isardam (by default) is defined according to paralysis of the opponent's units, or
CASE 2) Isardam (by default) is defined according to MUTUAL paralysis.
Either one being the case, one must be very careful to consider the impact upon non-reciprocal attacking units (such as hoppers)!
e.g., scheme
(= 3+3 )
Isardam
b) wGd8->a8
c) bGh6->g6
Stipulation: With black to play (assuming 1...a5 is legal), is it legal to move his Grasshopper to the 8th rank?
a) diagram
In either CASE (1 or 2), 1...Gf8 must be illegal (paralyzes the opponent's unit, and enters into mutual paralysis).
b) wGd8 shifted to a8
In either CASE (1 or 2), 1...Gf8 must be legal (does not paralyze the opponent's unit, nor does it enter into mutual paralysis).
c) bGh6 shifted to g6.
Here's the real question!
CASE 1 says I can not paralyze my opponent's units, thus the move (1...Gg8) must be illegal.
CASE 2 says the paralysis is not mutual, thus the move (1...Gg8) is completely legal. | | (7) Posted by seetharaman kalyan [Saturday, Dec 18, 2010 12:19]; edited by seetharaman kalyan [10-12-18] | Any move which result in Madrasi type paralysis is illegal. So moving your own piece into paralysis is definitely illegal. So, 1...a7-a5 which results in temporary paralysis of the pawn is illegal. It is high time that Popeye or winChloe correct this if necessary. | | (8) Posted by Kevin Begley [Saturday, Dec 18, 2010 12:32]; edited by Kevin Begley [10-12-18] | Not so fast -- the problem here is neither Popeye's nor Win Chloe's...
There are various interpretations for what constitutes "Madrasi Paralysis" -- do not assume that your interpretation is the only one valid.
Furthermore, a number of problems (several have been cited) already allow en passant capture.
Because they do, by default, you must assume the worst case (en passant is legal), to avoid being cooked...
Otherwise, you must allow for Isardam Type C (in the interest of preserving works which were assumed correct, when published).
And, I've already noted that by combination of Type B & C, you may get Type D.
The problem only compounds, if Type C is allowed to have to two separate CASES (as explained above).
And then, there is the matter of Rex Inclusive...
My suggestion would be for Isardam to default to CASE 2 (w/ legal en passant) -- that is: only MUTUAL paralysis is illegal in Isardam.
If somebody requires a new Type (to avoid en passant, or to make paralysis based upon the opponent's units) the burden is theirs to introduce it.
[EDIT: There is no Fairy Codex which maintains official, complete definitions of fairy elements.
...no group with the authority to mitigate special case rules not explicitly considered by an inventor.
...no group which maintains any version control of fairy elements (even for modified rules in the FIDE Condition, called Chess]
...no group which regularly schedules TTs, in the interest of eliciting ambiguous rules.
...no group which sanctions consistent, time-tested fairy elements (encouraging inventors to stay consistent to established models).
There is not even an established consistency about simple promotion duals, or final move duals in orthodox selfmates!
Instead, we prefer to rely upon programmers to locate and solve all our own problems for us.] | | (9) Posted by Bojan Basic [Saturday, Dec 18, 2010 15:03] | The problem with Popeye is that it is inconsistent with itself in this case. Namely, it doesn’t allow non-mutual paralysis, en passant capture being the only exception. That is, in Kevin's scheme, moving black grasshopper to the 8th rank is considered illegal in all three phases a), b) and c).
Furthermore, check this scheme (this is what I also mentioned in the first post, but maybe I wasn't clear enough).
(= 2+1 )
The only possible sequence of two halfmoves (B->), according to Popeye, are 1.... e7-e5 2.d5-d6 and 1....e7-e5 2.d5*e6 e.p. Popeye considers 1.... e7-e5 2.Qb1-~ illegal!
It becomes more interesting if we move the queen to a1. The third legal sequence then appears: 1.... e7-e5 2.Qa1*e5, but again that's all. Therefore, seems that Popeye allows the double-step, but the pawns mustn't remain next to each other after the following move. Does anybody have an explanation for the logic behind this? Personally, I think it's a bug, and that the programmers' intention was to make non-mutual paralysis illegal (as is in WinChloe). | | (10) Posted by Kevin Begley [Saturday, Dec 18, 2010 15:42]; edited by Kevin Begley [10-12-18] | This may indeed be a bug... however, there is another distinct possibility...
Consider en passant captures in Madrasi -- the forward stepping Pawn is considered paralyzed for a single move (and as such, strange things occur, such as staleamates in positions appearing to be checkmates).
Perhaps there is some strange logic at work, which demands that the pawns be immediately separated?!
I would be surprised if the popeye team went to the effort to program this (particularly given the fact that they have not treated non-mutual attacks, such as in the case of grasshoppers)...
My conclusion: there may actually be no such thing as a C+ Isardam problem (save when en passant is mathematically impossible, or specify "Isardam Type C"). | | (11) Posted by Axel Gilbert [Saturday, Dec 18, 2010 16:28] | Well the problem is that we have published problems with both conventions : "ep paralysis" illegal or ep paralysis (and therefore ep) legal.
For instance :
(= 3+3 )
sh#15
Isardam + Circé
by Günter Jordan
7 Kxa1 (Bc1) then 15 Kd6 Bf4# - 16 e5 illegal.
At this point, I don't see how with we shall decide which definition is better. I do prefer forbiding any "one-way" Madrasi paralysis but who am I to speak ?
As for the computer program (and I do agree of course that we can't rely on how conditions are programmed to define them) :
My version of Popeye (PopeyeUCI) does not allow "ep paralysis" (therefore, the problem above is C+ both with PopeyeUci and Winchloé) so I suppose that I depends on the version. I don't get the strange behavior quoted upon by Bojan : 1... e7-e5 is impossible.
So... I can't conclude yet !
As for the current composition tournament, I'll give you my decision (in accordance with the judges) asap. | | (12) Posted by seetharaman kalyan [Saturday, Dec 18, 2010 18:26] | As Kevin rightly pointed out, it would be better if an official clarification from our federation is given on such issues. All the important fairy forms should be brought under the Codex and a small Committee should be formed and allowed to give quick rulings. Such a committee will of course use the discussions in Forums like this. Is anyone willing to "Bell the cat" ? | | (13) Posted by Kevin Begley [Saturday, Dec 18, 2010 18:30]; edited by Kevin Begley [10-12-18] | Mystery is solved... [edit: at least for the non-retro section of this TT, it seems en passant is acceptable.]
Michel Caillaud (the non-retro Judge for the present TT) renders the following opinion (available on PDB: PROBID='P1098716'):
"Note that this solution is illegal according to Popeye that doesn't allow 3.f5, but legal according to Winchloé; difference between the 2 programs seems to be that Popeye applies observation of one piece by another to reject the move under Isardam criteria, and Winchloé applies abitility to capture the other piece."
[edit: seems he has the programs swapped here.]
However, Michel does not address the larger question -- if en passant is possible, then the rules governing hoppers are not presently computer testable (the potential impact of this is considerable, for virtually any problem using non-reciprocal fairy units).
With such problems already existing in both states, I see no way to avoid a third (Isardam Type B already exists) Type of Isardam.
ps: There's a 50% chance any of us will volunteer to entangle the bell around the neck of Schrödinger's cat. :) | | (14) Posted by Kevin Begley [Saturday, Dec 18, 2010 19:33]; edited by Kevin Begley [10-12-18] | Ignoring Type B, Rex Inclusive, and the potential combination of Types, I see 3 distinct Types of Isardam possible:
1) Mutual Isardam
A player is prevented from causing MUTUAL Madrasi-Paralysis.
This is the simplest form -- no EP captures -- and is fully testable (but only with Win Chloe).
2) Opponent Isardam
A player is prevented from Madrasi-Paralyzing an Opponent's unit.
This is a more complex form -- EP captures may be possible -- but, is not completely testable (only partially by popeye).
3) Self Isardam
A player is prevented from making any move which Madrasi-Paralyzes his own unit (not implemented anywhere).
Not implemented (even partially) in any software.
Unless the judges insist upon a specific type, I presume composers would have a free choice, providing they specify. | | (15) Posted by Bojan Basic [Saturday, Dec 18, 2010 22:06] | Axel’s comment that PopeyeUCI doesn’t allow en passant paralysis intrigued me, since PopeyeUCI is based on the earlier version of Popeye (the current one is 4.55). This made me research a bit, and here is the result: v4.47 and the prior ones do not manifest this strange before—they simply forbid en passant. Therefore, the conclusion is that this is, after all, a bug in Popeye, and that v4.47 and the earlier ones can be used for testing.
QUOTE Michel Caillaud (the non-retro Judge for the present TT) renders the following opinion (available on PDB: PROBID='P1098716'):
"Note that this solution is illegal according to Popeye that doesn't allow 3.f5, but legal according to Winchloé; difference between the 2 programs seems to be that Popeye applies observation of one piece by another to reject the move under Isardam criteria, and Winchloé applies abitility to capture the other piece."
This has nothing to do with the en passant capture, but with the combination of Isardam and other fairy conditions (in this case, Immun Chess). | | (16) Posted by Kevin Begley [Sunday, Dec 19, 2010 11:15] | >"This has nothing to do with the en passant capture, but with the combination of Isardam and other fairy conditions (in this case, Immun Chess."
It does relate to the question of whether only mutual madrasi-paralysis is prevented by Isardam, which goes to the heart of the en passant dilemma.
Also, it illustrates that this judge is willing to consider alternative interpretations when confronted by ambiguous rules (which should be no surprise to anyone familiar with his work). | | (17) Posted by Kevin Begley [Sunday, Dec 19, 2010 12:31]; edited by Kevin Begley [10-12-19] | Upon further reflection, I realize my classifications for Isardam problems are inexact (if not outright erroneous)...
So, here are my amends:
Consider the following two "Isardam with Grasshoppers" problems (note both are C+ in Win Chloe).
Kankuh Kobayashi
1st HM, Probleemblad, 2000
(= 2+3 )
h=2.5 b) d5 <> f5 (C+ Win Chloe)
= Grasshoppers
Isardam (Opponent Type)
a) 1…Sf2 2.Sd8 Ré3 3.Cd5+ Sd4=
b) 1…Sg2 2.Sç8 Rg3 3.Cf5+ Sg4=
In the final positions (both phases), note why the black King is immobile -- it appears to be what I had called "Opponent Madrasi."
Opponent Madrasi -- A player is only prevented from making moves which result in Madrasi-Paralysis to his opponent's unit.
Any move of the bK would result in madrasi-paralysis upon the wG (but neither mutual-paralysis, nor paralysis to black's own G).
I had incorrectly stated that "Opponent Madrasi" is partially implemented by newer versions of popeye, but I had it exactly wrong.
Nor is "Opponent Madrasi" precisely Win Chloe's Type of Isardam implementation!
As we shall soon see, there are FOUR (4) Isardam Types... not three (3), as I had originally suspected.
Peter Harris
feenschach, 2005
(= 5+8 )
h=3 b) -bGd3 (C+ Win Chloe)
= Grasshoppers
Isardam
a) 1.Gd2 Kg1 2.Gd1 Gf3 3.Gh1* Gf2=
b) 1.K6f4 Kg2 2.Gdd6 Gb3+ 3.Gh1 Gb2=
Just consider the position after the emboldened asterisk.
Black has just moved bGd1-h1, to prevent wGf3 from capturing the bKd5.
The mechanism which prevents capture on d5 is not "Opponent Madrasi," as I would previously have expected; but, "Self Madrasi."
Self Madrasi -- A player is only prevented from making moves which result in Madrasi-Paralysis to his own unit.
If white Gf3 captures on d5, it results in madrasi-paralysis to the wGd5 (white's own unit), but not to any opponent unit!
Thus, Win Chloe implements a fourth type of Isardam: "Any Isardam" -- a combination of Self- and Opponent- Isardam.
Ignoring Isardam Type B, and Rex Inclusives, there are four distinct types of Isardam (two are important, two are not implemented):
1) Mutual Isardam : A player is only prevented from making moves which result in MUTUAL Madrasi-Paralysis (EP possible).
2) Opponent Isardam : A player is only prevented from making moves which result in Madrasi paralysis to his opponent's units (EP possible).
3) Self Isardam : A player is only prevented from making moves which result in Madrasi-Paralysis to his own units (EP not possible)
4) ANY Isardam : a player is prevented from making moves which result in Madrasi-Paralysis to any unit (EP not possible).
[edit: Type 4 may equally be considered to be a combination of Type 2 and Type 3.]
Win Chloe fully implements type 4, as does older popeye versions (prior to 4.47).
The vast majority of published problems fall under this category.
The later versions of popeye seemed to [partially] implement either Mutual- or Opponent- Type Isardam (but more likely, these were only bugs).
Nevertheless, there are several problems published which do allow en passant captures in Isardam.
Therefore, at least one of these Types (perhaps both) must be valid (though, neither is presently implemented in software).
Which Type of Isardam is expected in the Noel TT?
Unless stated otherwise, I would presume any Type (providing the rules are clearly specified).
Which form is the default type of Isardam?
Most would say type 4, though a fair argument could be made that en passant should be presumed legal (to avoid cooks, etc).
Most of us who favor defaulting to allowing en passant captures, I suspect, probably prefer Type 1 (over Type 2).
I leave the matter in the capable hands of the Fairy Rules Council (which is yet to be appointed).
[The rest of us can rejoice, knowing that many items of business must be considered before this matter hits their docket!] | | (18) Posted by Michel Caillaud [Sunday, Dec 19, 2010 18:13] | There are "complementary conditions" in Winchloé that can extend the "basic conditions".
The Proofgame by Jonathan and Allan is presented in Winchloé with "complementary condition" "Pas de prise en passant" (no en passant capture), thus allowing move 2...b5 no allowed in "basic Isardam".
My view can be summed up like this : I don't care the rule, I care the problem.
Of course, it is more confortable to have computer-tested problem. Thus, due to additional work for testers and judges, one can expect that composers using unusual variants are doing this cleverly... | | (19) Posted by Kevin Begley [Sunday, Dec 19, 2010 21:04]; edited by Kevin Begley [10-12-19] | >The Proofgame by Jonathan and Allan is presented in Winchloé with "complementary condition" "Pas de prise en passant" (no en passant capture), thus allowing move 2...b5 no allowed in "basic Isardam".
Indeed -- my apologies for overlooking (and thus not mentioning) this additional fairy element: no en passant capture.
The added condition seems not to have been manufactured for db consistency -- it is part of the original problem (see PROBID='P1011756').
But, regardless... there are Isardam problems (w/ no other condition) showing actual en passant captures.
...to see some of these, just search: K='isardam' and sol='ep'
If these are legal (in some type of Isardam) it would also impact hoppers (and other units with non-reciprocal attacking possibilities). | | (20) Posted by Michel Caillaud [Sunday, Dec 19, 2010 22:42] | << The added condition seems not to have been manufactured for db consistency -- it is part of the original problem (see PROBID='P1011756').>>
Yes, I know (I was the judge). Composers added the condition as they could not test the problem with Popeye that refused 2...b5 at the time...and I probably made the entry in Winchloe.
<< But, regardless... there are Isardam problems (w/ no other condition) showing actual en passant captures.
...to see some of these, just search: K='isardam' and sol='ep'
If these are legal (in some type of Isardam) it would also impact hoppers (and other units with non-reciprocal attacking possibilities).>>
?? I made the search in Winchloe and found only 3 problems. 1 by Juraj Loric, a hybrid triplet where the e.p. is not in the Isardam phase, and 2 "Last move?" by Arno Tungler. I suspect that if Christian had checked these 2 when entering the base, he would have declared them as "No solution"... | | Read more... | Page: [Previous] [Next] 1 2
MatPlus.Net Forum General Isardam and en passant |
|
|
|