The Chess Variant Pages
Custom Search




[ Help | Earliest Comments | Latest Comments ]
[ List All Subjects of Discussion | Create New Subject of Discussion ]
[ List Latest Comments Only For Pages | Games | Rated Pages | Rated Games | Subjects of Discussion ]

Comments/Ratings for a Single Item

Later Reverse Order EarlierEarliest
This item is a game information page
It belongs to categories: Orthodox chess, 
It was last modified on: 2015-03-16
 Author: H. G.  Muller. Chu Shogi. Historic Japanese favorite, featuring a multi-capturing Lion. (12x12, Cells: 144) (Recognized!)[All Comments] [Add Comment or Rating]
H. G. Muller wrote on 2020-10-14 UTC

Yes, HaChu was written by me, and yes, I am in practice the only WinBoard/XBoard developer. PM on WinBoard forum is not really a good way to reach me, as my inbox there is overflowing. TalkChess.com would be a better forum, but for private communication e-mail would be preferable. However, I usually prefer public over private communication; most things that are discussed tend to be useful for others as well.

The mnemonic pieces were indeed conceived from the wish to be able to play these horrendously large games with virtually no learning at all, because 95% of the piece symbols unambiguously indicates its move. So you only have to worry about a hand-full of (often very special and very powerful) pieces, such as Lions, Teaching Kings and Fire Demons.

It is great that you are working on a replacement for the 81Doju Chu-Shogi website. In practice it might be difficult to get the population of that website to migrate to yours, through, no matter how nice it is. Especially the Japanese players. Setting up a new site also means the loss of the database of all games that have been played so far. Perhaps it would have a better chance of success if you would just build a new client for the existing 81Dojo Chu server.


Alex Trick wrote on 2020-10-14 UTC

@Muller

Thank you very much for SVG pics! :) They're excellent!

In general we're going to popularize Chu Shogi and (I hope) the others Large-board shogi variants as well.

Mnemonic SVG images are extremely helpful for introducting the game for beginners and others who is afraid of japanese hieroghyphes. Especially including into account that Chu Shogi itself includes comparativelly so much pieces with different variants of move, that it's difficult to remember them in case you don't love it initially. What to tell about much larger variants of shogi with a much complicated move rules.

If we combine mnemonic pics with a pleasant (a bit beautiful) interface, move tips and learning for new players, add HaChu engine for playing with computer (I guess, your work?), and also add possibility to play with another peoples over network (planned in nearest future) - we'll get the platform that actually can be distributed around the world in a much easier way, than before.

For example, when 81dojo came into world, it popularized shogi around the world quicker. The analogue for the Chu Shogi - the Chu-Shogi Renmei 81client had a similar impact, but it still contains hieroglyphes (however it still has a nice interface through). For now Chu-Shogi Renmei 81client is under danger to be unavailable because of outdating of Flash Player technology. If after end of 2020 Flash Player apps will be actually blocked (forcefully), then the remaining variants to play over Internet would be a few. Maybe ChuDo will be one of most comfortable of them. That is the point - the more comfort - the more people coming, as there is easier to entry into the game.

I have registered on WinBoard forum, I'll take pleasure to contact you there via PM or any another way. If I understand correct, you're the only (or main) developer of WinBoard/XBoard programs, are you?


H. G. Muller wrote on 2020-10-13 UTC

@Alex - The mnemonic pics here were copied from WinBoard by taking screenshots, where I used a general-purpose drawing program to make the background transparent. WinBoard creates those images on-the fly, from components that represent the move in one direction (but i a way depending on the range in adjacent directions), plus a central dot. In order to draw an edge around that for the outline pieces, it creates a black image first, and overlays it with a slightly smaller white image, so that a black outline remains. The largest size is 33x33, though, and the images are not anti-aliased.

I remember having made a set of SVG images for each of the individual pieces, for XBoard, though. I am pretty sure this includes all the Chu-Shogi pieces, and probably also the Wa-Shogi pieces. I would have to dig that up; I could not find it on my website.

Interesting that there now is a new Chu-Shogi engine. How strong is it, compared to HaChu?

As to larger variants: I have a pretty strong engine for Tenjiku Shogi. (Not yet released; I am still planning to do that.)

I am not on Facebook.

[Edit] Oh, I see. The engine actually is HaChu. It is only the interface that is new.

[Edit2] I found the SVG pieces, and uploaded the whole project to http://hgm.nubati.net/Mnemonic-S.tar.gz .


Alex Trick wrote on 2020-10-13 UTC

Hello, H.G. Muller!

So, if I understand right, you weren't drawing it by yourself?

Recently one man named Joe Henbethydd have developed a Chu Shogi program with a pleasant interface. I took participation in its localization to English. You can try it yourself here http://anubhuuti.com/chudo/

As I understand, he is planning include multiplayer game over network in future, also I hope, he is planning include another Large-boards variants of shogi like Maka Dai Dai Shogi etc. But it is not exact. Time will tell.

For popularization purposes of Chu Shogi there is a nice idea to use mnemonic pictures representing piece's moves. I really like it, and I use it as my memory sheets, and also I use it when teaching peoples how to play Chu Shogi.

It is nice if we would integrate such mnemonic pics into ChuDo program of Joe Henbethydd, but unfortunately, the pics are somebit a low quality (I mean low resolution). It is nice, if we have at least 200x200 sized images, or (which is better) vector images.

At the moment we don't have any artist which could draw for us such image assets.

So the question is - where did you actually get the mnemonic pics? Have you drawn them by yourself, or you just used already existing images, which is located in the link you provided?

Anyway nice to meet you, and hope we shall some time play Chu Shogi or another Large-board variant of shogi.

P.S. By the way, do you have any page on Facebook? There is a Chu Shogi group, maybe we can find each another there and write one to another sometimes?

Sincerelly yours Alex Trick


H. G. Muller wrote on 2020-07-16 UTC

They are in graphicsDir=/membergraphics/MSdaishogi/ .

Just go to the web page http://chessvariants.com/membergraphics/MSdaishogi to see how they are named.


Daphne Snowmoon wrote on 2020-07-16 UTC

If you are okay, Can i know where and how did you use the mnemonic characters?


dax00 wrote on 2020-06-13 UTC

Using my (pre-adjustment) piece value table, under early middlegame, the value ranges given are as follows:

  • Pawn -2.5/4/6 (Tokin 9/9.5/10)
  • Go-between 8/8.5/10
  • Bronze 6/8/9
  • Silver 8.5/10.5/11
  • Leopard 9.5/10.5/11
  • Lance 14.5/16 (White Horse 25/26/27)
  • Kirin 18/19/20
  • Bishop 24/26/29

Then I adjust the values using things I know about the position, which can easily swing them a further 20% for ranged pieces, 40% for small pieces, and 200% for individual pawns.

I wrote the table to suit my personal style, at the behest of one Andrej Trauner, so it may not be applicable to others. I believe it a safe assumption that my table is the most thorough estimate of chu shogi piece values in existence, especially when combined with the list of adjustments, which only exists in my head.

I don't put too much weight on material advantage when playing. It only accounts for roughly 35% of my evaluation, compared to about 60% for purely positional considerations.

 


H. G. Muller wrote on 2020-06-13 UTC

Well, it estimates about 450 for both Kirin and Lance! (to compare: 480 for Bishop.) It differs a bit each time, as the estimation process involves random sampling.


dax00 wrote on 2020-06-13 UTC

It takes many tempi to attempt to promote a Lance. Edge attacks have never been my favorite, so I haven't had much experience in playing for such promotions, until very late in games. However, based off opponents who have tried to move along the edge files, and my defense against such attacks, I have concluded that early edge pushes require at least 20 tempi of preparation to be mildly successsful, against someone who is not exactly interested in the same. A player can comfortably defend, at least for a while, but the edge file attacker can get through at least one promotion exchange if they put enough in it, this usually being some exchange like Dragon-promoting for Rook. A player must be particularly careless to allow a Lance to promote safely and flee.

Overall, I don't rate Lances that highly, perhaps around the same as a Kirin.


H. G. Muller wrote on 2020-06-13 UTC

Yes, of course it is very weak. It searches only 2 ply, and that isn't even enough to see a mate-in-1 threat coming. You can switch it to 3 ply, but in a game as large as Chu Shogi that would probably already make it think too long. This is JavaScript, not a C program!

I am sure I could make a search that plays much stronger with the same number of nodes. I put this one together in two days. But that is not meant as an excuse: I intend this thing to be weak.  It is meant to be a demo for people that have never played the game before, and visit the page the diagram is on to learn about its rules, so they can have a sparring partner to get a whif of how the rules work out in practice.

Also note that it doesn't use any variant-specific knowledge other than the rules. It has to guess the piece values itself, and I am amazed of how reasonable these guesses appear to be. I made a version that also tries to weight in promotability, by making some weighted average of the values of base piece and promoted form, where the weight of the latter increases with the ease of its promotion and the relative gain of intrinsic value. That also looks reasonable, except that the current algorithm then puts an unexpectedly high value on the Lance. It seems this has the most favorable prospects for promotion, so that it assumes you would do your best to preserve it until it can actually promote, rather than lose it in tactics to clear the way for others. (Of course it doesn't realize the Lance is file bound, so that you really don't have much choice in the matter...). But now that I think about it, perhaps this is not such a silly thought. The Lance itself is pretty useless in combat, can reach the zone in a single move when the time comes, and a White Horse is one of the more valuable pieces.


dax00 wrote on 2020-06-13 UTC

So, I switched to Firefox. It appears to be working better now. Still very weak AI

 


Aurelian Florea wrote on 2020-06-13 UTC

I thought the AI would be easy picking but it is not :)!


H. G. Muller wrote on 2020-06-12 UTC

Capture bridges now also work. The anti-trading rules in Chu Shogi are so complex because they are applied to a piece capable of double capture, so that a trade could be accompanied by a sizable material gain, and they did not want to outlaw that. The same could in principle happen when several piece types of unequal value are subject to this rule. (As is the case in Tengu Dai Shogi, where Lion and Lion Dog can also not capture each other when they are protected.)

I solved it by specifying a value threshold, and when the combined gain of the capture is larger than the value of the capturer plus this threshold it can be made safely. Because the piece values are only known to the diagram, the option to specify this is tradeThreshold=N, where N is the number of the most-valuable piece that can not act as a capture bridge. So in this case I specified N=2, as the GB was defined as the second piece.

This appears to work, as can be easily tested by dragging a white Lion to i6 and a black one to g7, switch the AI on, and then drag various white pieces to h6 (and then take back 2 moves for the next try). For every piece black should then play Lnxh6xi6, except for P and GB, where it will play something else.


H. G. Muller wrote on 2020-06-12 UTC

For me it plays BT-g10 instead of Lnxe7. Your browser must be using an old cached script.

BTW, I sort of patched the counter-strike problem now, by paying more attention to what the lastest move was when invoking the AI, and making it aware of a preceding non-Lion x Lion when I set it thinking. This too isn't perfect yet. (It would miss a locust capture of the Lion by a Falcon or Eagle.) But for now it should do.

The point is that I want to replace the entire old move-entry system of the diagram, which had a move generator that directly highlighted squares, in interaction with what already had been clicked, by one based on the new move generator in the AI. The latter just generates a list of all moves, complete with all side effects. And because it can search ahead, it will be able to judge even the more complex legality conditions perfectly. The highlighting during move entry should then be based on that list, and the moves would automaticaly be in AI format, so that they can be fed back into it when it goes thinking for a reply. So it would be a bit of a waste to spend too much time patching up the old system for better interaction with the AI.


Aurelian Florea wrote on 2020-06-12 UTC

I had managed to clear the cache but the error still goes. I use microsoft edge.


dax00 wrote on 2020-06-12 UTC

Same with me.

1. Nf5 Cd11 2. Ne7 Ci11 3. Nxd8xe7 Ng8 4. h5 Nxe7


Aurelian Florea wrote on 2020-06-12 UTC

Wow cool thing supporting the imitator. I remember earlier versions were not able. Does the AI work also with it? This could sound silly, but random initial positions are supported?


H. G. Muller wrote on 2020-06-12 UTC

Well, for me it doesn't capture the Lion, after these moves. I don't know what browser you are using, but on FireFox I can clear the cache by keeping Shift pressed during the page reload.

The diagram already supports a Betza notation for a Joker (I for Imitator). See the 'Betza sandbox' comment on Betza notation.


Aurelian Florea wrote on 2020-06-12 UTC

And by the way, talking about previous moves, is there a way they will be taken into account as I am thinking about the joker in my apothecary chess game.


Aurelian Florea wrote on 2020-06-12 UTC

The game started from regular initial position and it went like this : 1. Ng5 Cd11 2. e5 Ci11 3. j5 Nf8 4. h5 Tg10 5. Ng6 Nxg6 . I'm not sure what needs refreshing.


H. G. Muller wrote on 2020-06-12 UTC

That is not enough information to set up the position. Can you post the game that it prints below the diagram, up to the position where it happened?

And did you make sure your browser cache was refreshed, so that you are using the updated diagram script, rather than an old cached one? (BTW, counter-strike is still a problem, because the AI does not take prior moves into account. So it wouldn't know that non-Lion x Lion took place.)


Aurelian Florea wrote on 2020-06-12 UTC

@HG,

Hello,

I had just tried it and my advanced lyon defended by the DH got captured by the AI. I then tried it again. It seems to happend to player's lyon on 6th rank once the second dk pawn got moved. Maybe if you try reproducing such a position you may find the bug!...


H. G. Muller wrote on 2020-06-12 UTC

The AI of the diagram now also has a vague notion of the Lion anti-trading rules. It isn't perfect yet, because it does not apply the rule for 'bridge capture' yet; you can just generally forbid capture of protected pieces of a certain type. It also doesn't limit it to distant captures, but I guess that only means it unjustly forbids moves that you would never want to make anyway.


Aurelian Florea wrote on 2020-06-10 UTC

Good job, Adam!...


dax00 wrote on 2020-06-10 UTC

Excellent, passed all my tests. Thanks again! I will be using this preset from now on.


25 comments displayed

Later Reverse Order EarlierEarliest

Permalink to the exact comments currently displayed.