Website founded by
Milan Velimirović
in 2006

22:27 UTC
ISC 2021



Remember me

Forgot your
Click here!
to create your account if you don't already have one.

Rating lists


MatPlus.Net Forum General Madrasi
You can only view this page!
Page: [Previous] [Next] 1 2
(21) Posted by Thomas Maeder [Monday, Jul 21, 2008 18:08]

I must say that I did not understand very well how it works in Popeye, I tried different possible rules, but each time there are exceptions.

So I would much appreciate to be explained how it works. Perhaps, Thomas, will you agree to spend a little time for that ?

Popeye doesn't implement Madrasi as regularly as I had imagined from only looking at the Anticirce+Madrasi case.

After a quick look through the (very complex!) Popeye source code, it seems that Madrasi observation is pretty much the same thing as orthodox attack (ignoring self-check), with some exceptions:
- Grid chess
- Mono-/Bichromatic chess
- Edge mover
- Koeko and Antikoeko
- Geneva chess

I can understand the first three, because they are static restrictions.

As for Koeko and Geneva chess, they are comparativley young condititions (especially Geneva chess), so it may be that their inventors defined Madrasi to behave this way.
(Read Only)pid=2498
(22) Posted by Guy Sobrecases [Monday, Jul 28, 2008 10:04]

I heard that similar debates had occured in the Jurassic period of Anticircé.
I do not want to add confusion to this nice discussion, but I confirm that Anticircé type Calvet is needed in our joint pb with Peter (at least in the first phase), kindly a/m by Vlaicu.
(Read Only)pid=2526
(23) Posted by Jacques Rotenberg [Monday, Jul 28, 2008 14:34]; edited by Jacques Rotenberg [08-07-30]

here it seems to be slightly different

In Anticirce you had a strange debate : may or may not a piece be allowed to capture on its rebirth square? After a (very) long debate it appeared a good solution : to let the two options possible, and just to specify it (when needed) with the stipulation.

Here you can only wish the situation to be similar!

In Wchloe, it seems clear that C. P. choosed the following option :
When you have Madrasi + condition "X", you calculate Madrasi effects with the position under condition "X".
In Popeye it could be that the choosen option is :
When you have Madrasi + condition "X", you calculate Madrasi effects with the position under orthodox rules.

Unfortunately, it is not like that! Popeye only mainly behaves that way, as explained Thomas.

So in the meanwhile, here is what we have :

Wchloe :
When Madrasi + "X", you calculate Madrasi effects with condition "X". (and it is not a programming error!!)

Popeye :
When Madrasi + "X", you calculate Madrasi effects with orthodox rules, BUT NOT ALWAYS !!.

In any case the situation seems to be perfectible
(Read Only)pid=2528
(24) Posted by Ian Shanahan [Friday, Aug 1, 2008 07:07]

I believe this dilemma is easily resolved (though not for computer programmers!) by invoking what ought to become known as the "Ridley Principle".

Some years ago in "The Problemist", an analogous debate erupted over a composition which utilised Circe and Monochrome Chess. The English problemist Mark Ridley settled the matter by making the stipulations 'non-commutative' (as a mathematician would put it): that is, the order in which the stipulations are presented is absolutely crucial - the first listed condition taking precedence over whatever follows. Hence Circe Monochrome is NOT the same as Monochrome Circe, because in the latter, rebirths onto the opposite-coloured squares are forbidden (i.e. Monochrome 'overrides' Circe).

So here, if Madrasi is listed first, then paralysis takes place in relation to orthodox chess; if listed after another condition, then that condition 'overrides' Madrasi.

Does this help?
(Read Only)pid=2558
(25) Posted by Juraj Lörinc [Friday, Aug 1, 2008 11:34]

In a sense it would help, at least on human recognition level.

However the current solving programs (at least here discussed Popeye and WinChloe) are treating fairy conditions put in for a problem as a set (not ordered), not a n-tuple or sequence (ordered). As set, there are internal rules for resolution of fairy condition interaction, specific for individual programs.

Recoding would be quite painful, I guess.
(Read Only)pid=2560
(26) Posted by Guy Sobrecases [Monday, Aug 4, 2008 00:21]

I think that an interesting point is to know the definition of Madrasi given by the inventor. So far, I could find three different ones on the web.

It seems that the first Madrasi problems were published by Abdul Jabbar KARWATKAR in "The Hindu", issued in July 1979. For sure, one can find here the original definition.

In the probable case that no Mat Plus member can find this newspaper in his library, one can also find an original #2 published by "The Problemist" in 1980.

Abdul Jabbar KARWATKAR
The Problemist 1980
(= 14+8 )

‡2 (14+8)

I unfortunately do not know precisely the issue.

The definition of the condition detailled in "The Problemist" reproduces, for sure, the original definition which had been edited in "The Hindu" some weeks before.
(Read Only)pid=2571
(27) Posted by Guy Sobrecases [Friday, Aug 15, 2008 00:09]

Thanks to Alain Bienabe, I have been able to read the definition of Madrasi, written in The Problemist, March 1981 :

« When a man, other than a K, attacks an enemy man of the same kind, both men become inert. Hence a capture between equals would be illegal. »

This definition probably sticks to the original one, edited in «The Hindu» in 1979.
(Read Only)pid=2619
(28) Posted by Thomas Maeder [Friday, Aug 15, 2008 22:59]

When a man, other than a K, attacks an enemy man of the same kind, both men become inert. Hence a capture between equals would be illegal.


This doesn't sound like a usable definition, though, because the word "attack" isn't defined. E.g. does a pinned piece attack anything (apart from the piece that pins it, if the pinned piece can move along the pin line)?

Also, the "definition" would cause a black pawn f4 to become temporarily "inert" if White plays g2-g4 (provided the black pawn "attacks" the white pawn), which sounds very original.
(Read Only)pid=2621
(29) Posted by Frank Richter [Saturday, Aug 16, 2008 13:20]

As far as I remember, the "en passant trick" was already used in Madrasi problems (by K. Wenda?)
(Read Only)pid=2623
(30) Posted by Joost de Heer [Saturday, Aug 16, 2008 13:43]

Quite interesting, the usual application of Madrasi allows one-way paralysis, the original definition doesn't. A poor translation job to English from the original source, or was it really intended this way?
(Read Only)pid=2624
(31) Posted by Mario Richter [Saturday, Aug 16, 2008 15:08]

Maybe an example composed by the inventor himself shows his intention:

A. J. Karwatkar
Eteroscacco 8 Ott-Dic 1979
(= 6+4 )

a) #3 (orthodox)
b) #3 Madrasi
(Read Only)pid=2625
(32) Posted by Thomas Maeder [Sunday, Aug 17, 2008 22:05]; edited by Thomas Maeder [08-08-17]

Maybe an example composed by the inventor himself shows his intention:


[Solution of b) 1.Rb4+ axb4+ 2.a4! (with temporary paralysis of PPa4b4) Ka5+ 3.Bxb4#]

So non-reciprocal Madrasi paralysis, as supported by Popeye and (probably) other solving programs, doesn't conform to the original Madrasi. Ouch!
(Read Only)pid=2629
(33) Posted by Kevin Begley [Monday, Aug 18, 2008 01:41]

ouch indeed...

Frankly, I don't like the inventor's intent as much as the way the rules have evolved...
More interesting to have just one pawn paralyzed in cases of en passat, in my humble opinion.

Clearly, this comes down to a version/naming issue (do we call the common version something like Modern Madrasi, or do we call the original version something like Original Madrasi, or what?).

The problem, comes down to the age old dilemma for fairy problems -- no fairy codex, no higher authority to resolve uncertainty.
Thus, editors and programmers make their own decisions, and solvers (and composers) are left to guess.

The longer we wait, the more it spins towards chaos.

I'd like to see regular tourney (perhaps semi-annual) asking for problems that demonstrate new cases of uncertainty.
This gives two years to resolve uncertainties, and the resolution by whatever authority committee can be published in the next round.

That's my 2 cents.

Kevin Begley.
(Read Only)pid=2630
(34) Posted by Ian Shanahan [Monday, Aug 18, 2008 04:10]; edited by Ian Shanahan [08-08-18]

I seem to recall, from Cedric Lytton's fairies column in "The Problemist" during the early 1980s, a composition (by Syzonenko?) with grasshoppers and the stipulation Madras* Chess [sic!] This was Madrasi with non-reciprocating paralysis, I think. This would definitely be worth investigating.
(Read Only)pid=2631
(35) Posted by Jean-Marc Loustau [Monday, Aug 18, 2008 14:43]

The non-reciprocity of paralysis by grasshoper in Madrasi was the theme of the 1986 Andernach tourney...
(Read Only)pid=2633
(36) Posted by Frank Richter [Monday, Aug 18, 2008 16:07]

There are a lot of other possibilities of "one-direction" paralysis in Madrasi (combination with KöKo, Imitator, neutral pieces etc.). Arnold Beine wrote an article about this theme (harmonie 75, 09/2003), and also we started a theme tourney some years ago (award see
(Read Only)pid=2634
(37) Posted by Eric Huber [Monday, Aug 18, 2008 20:45]

The Madrasi definition we've been using for some years seems to offer more possibilities than the original one. Shouldn't we keep both definitions and specify, for each problem, which one we use or have used ?

We have the same problem with Republican Chess. ( See this recent talk on Mat Plus forum ) Some problems have been composed using the peculiarity of the Poisson definition (it's mandatory to put the opposite King on the board when there's a mate) whereas the original van Gool definition says something different about that particular issue.
(Read Only)pid=2636
(38) Posted by Kevin Begley [Tuesday, Aug 19, 2008 02:13]


Forcing Solvers to plow through countless special case rules for each fairy element is not acceptable.

Unless rules are standardized, fairy problems fall into the chaos common within the chess variant crowd -- where every poor fool must plow through winded definitions to understand that a wizard is just a camel, or maybe this time it's called a sorceror, or a mage, or man-witch, or whatever they decide to call the eye of newt weilding tenderfoot (which is really a camel). And when you finally find a camel that acts like a camel, keep reading -- a map from Dune may be overlayed onto the board, and you'll be warned about camels making ryhthmic movement (see also wormsign, see also checkmate eternal).

Some standard rules, from some authority, would be nice.
Start with a small codex -- things everybody accepts: Grasshoppers, Nightriders, oops -- I was going to say Madrasi, ...
Work up to the complex issues, and develop some methodology for preserving problems which were developed without standard rules.

It's either standards now, or we shall all, oneday, be composing Lucas Cycles.

(Read Only)pid=2639
(39) Posted by seetharaman kalyan [Monday, May 10, 2010 07:59]

there was a special issue of 'feenscach' containing a long article on Madrasi (1981?). It contained a definition slightly different to the original but okayed by Mr.Abdul Jabbar Karwatkar. I think that definition used the term 'observed' for the first time. If I remember right, a piece 'observed' was defined as as an opponent King in that place being in check.
(Read Only)pid=5353
(40) Posted by Alfred Pfeiffer [Thursday, Jun 3, 2010 00:29]

The text was: "MADRASI - ein Indermärchen" by -be- (=Bernd Ellinghoven), published in "feenschach" 65, 15. Mai 1983, pages 74-91.
AJK's first rule is named as "old Madrasi" (altes Madrasi), and the simple "Madrasi" is the new rule (accepted by AJK), allowing also asymmetric paralysings.
(Read Only)pid=5441

No more posts
Page: [Previous] [Next] 1 2

MatPlus.Net Forum General Madrasi