Fairy-Max 4.8 released

[hgm]

a) Wf : the average of all affected squares which could be covered at all by a chess piece (this concerns all ever possible targets of defense or attack)

These 'ab -initio' piece-value determinations are extremely tricky, and I stopped trusting them. If someone would have asked me the value of the Nightrider, I would have guessed somewhat below Rook (like you find), and perhaps even below Commoner (=non-royal King). After all, the latter two have mating potential, which the Nightrider lacks.

But not So!

Replace the Rooks with Nightriders (H) in the opening position for one side, and the side having the Rooks gets a beating! Even Pawn odds (f-Pawn) is not enough to overcome the difference: H+H beats R+R+P still significantly. Despite the fact that the Nightrider side is handicapped by having no castling rights, in addition to the Pawn odds!

[smrf]

Hello H.G. ,

please do not forget, that tthis part of my evaluation model merely describes the STATIC component of a chess piece. The DYNAMIC positionally aspects are not covererd directly by it. Therefore it is not as easy as you concluded. In my model the positional influence values are somehow related to k/x of the average static values for dominated squares. Thus "weaker" pieces have the stronger positional influence.

Reinhard.

[hgm]

I know it is not easy. This is why I reverted to simply determining the piece values empirically, from unequal-material games. And the opening value of the Nightrider definitely is about 60 cP above that of the Rook.

I guess what an ab-initio theory of piece values should weight in quite heavily is the number of directions a piece can move in, as well as the average distance covered. The Nightrider does have on the average fewer moves as a Rook on an empty board, but they are not as easily blocked. The Knight has on the average fewer moves than a (Commoner) King, and being Leapers blocking of rays plays no role. But yet the Knight is the stronger piece.

[mageofmaple]

Congratulations on making a universal interface for large-board chess games! But where can I download it? I've read the thread, but don't seem to be able to find a link.

Thanks!
Greg

[hgm]

Correct, there is no link because it is not released yet. I don't want to release as long as it contains known bugs. And currently saving the PGN does not work correctly (although it did work before): all moves with Chancellor and Archbishop are simply omitted!

This must have to do with the way I tried to implement more different piece characters in PGN and FEN (or edit) than there are display symbols. (Such that the display symbols can be used for multiple pieces.) I am not happy with this solution at all, and I am going to change it by adding a new option

/PieceToChar="PNBRQKpnbrqk"

that can assign the letter by which a piece is indicated in FEN and PGN arbitrarily, so that the user can assign anny external piece indicator and any font symbol independently to each member of the (necessarily limited) set of internal piece representations.

I will also have to increase the number of internal pieces beyond the 11 that I have now, as Shogi seems to need more. (The problem is that most pieces can promote, and that the promoted versions, although they move the same, have to remain distinguishable by origin, as on capture they refer to the original piece when they are dropped again on the board. So there need to be 5 flavors of Gold General, depending on if they are promoted Pawn, Lance, Knight, Silver General, or an original Gold General. And the Rook and Bishop have their own promoted forms, Dragon King and Dragon Horse. Together with King that makes 14, which can all occur in the same game.)

I will fix all this soon, but currently I am playing in the Dutch Open in Leiden, and it proves difficult to program and operate a Chess engine for over-the-board games at the same time...

[mageofmaple]

That makes sense. Thanks for all your hard work! It will terrific to have a universal interface flexible enough to even handle Shogi.

P.S. Chu Shogi a semi-popular large-board Shogi has 30 piece types! Of course, this makes things more complex, as there are only 26 letters.
ChessV allows two-letter notations, provided one of the following: either the first letter of your two-letter notation is not a separate notation, or, if it is, preceed the two-letter notation with an underscore - the underscore is never displayed but it is stored as part of the FEN to force it to read as a two-letter piece and not two one-letter peices. But you probably don't want to get this complicated yet And there are probably no winboard engines playing Chu Shogi, but support for Betza's Chess With Different Armies (one of my personal favorites) requires a lot of different piece types also.

[hgm]

And there are probably no winboard engines playing Chu Shogi, but support for Betza's Chess With Different Armies (one of my personal favorites) requires a lot of different piece types also.

Yes, but in Betza's game the are not really needed all in the same game. His armies typcally consist (just like in FIDE Chess) of 5 piece types plus pawns. And even if you are allowed to prmote to piece types the opponent has, this means at most 9 piece types plus pawns, as they share the the King.

Btw, Fairy-Max allows 15 piece types to participate in a game (put pieces that move as mirror images, such as white and black Pawns, are considered different). It should have no problem at all playing most of Betza's pieces. Only the 'Short Rook' of the 'Remarkable Rookies' would be a problem in the current Fairy-Max version. But all these combinations of Ferzes, Wazirs, Dabbabahs, that is really a piece of cake. I should work on implementing minor promotions, though.