Not logged inRybka Chess Community Forum
Forum RybkaChess Help Search Register Login
Up Topic Rybka Support & Discussion / Aquarium / Tree exchange
- - By Kappatoo (****) [de] Date 2012-03-15 16:55
I have noticed that many guys here are very secretive about their opening analysis. In a sense, I can understand this, since many of you are just pretty competitive. On the other hand, I would guess that most of us are no competitors (thinking about the WBCCC, this may not be so clear, but okay). I, for instance, will never play a freestyle tournament and will never play with an engine on the chessbase server, and will very likely never again play a correspondence tournament. The chances that I will meet one of you OTB are extremely slim.

Notice, moreover, that even of some of you are potential competitors, it can still be beneficial for both to exchange information.
I recall that billyrabar provided his huge Sveshnikov tree for free, as did Mark with his Caro Kann tree. (There are also a number of book-makers who share their work.) This is of course very generous, but I would guess that not everyone is comfortable with giving away his/her precious work for free.

So I thought the next-best thing would be to exchange information. If A has analyzed line a and needs information about line b and the reverse is true for B, why don't they just exchange their results? Or maybe C and D both have information about line c and agree to co-operate. Many ways of sharing information are conceivable. I would guess that some such co-operation among some forum members already takes place, but my thought was that the chances of reaching like-minded people are bigger if there is a public place to 'advertise'.

What do you think?
Parent - - By mattchess (*) [us] Date 2012-03-15 17:16
I think that is a great idea :cool:

Perhaps the moderators would be willing to make a new forum group for this called "IDEA Exchange" and a new thread could be created for each submitted project.  Maybe we start by picking an opening and have volunteers deeply analyze interesting positions and then share the results that we could export into our own master trees. 

Personally, lately I have been looking at lines that come out of the King Fianchetto modern/pirc and scotch openings. 

Should come up with a standard post template that would provide the information people need to evaluate if they want to use the tree.  Perhaps:

GENERAL DESCRIPTION:
ROOTS:  <Paste the FEN positions here>
ENGINE:
HARDWARE: (general description of your processor, memory)
ANALYSIS QUALITY SETTINGS:   <Time> <and/or> <Depth>
TOTAL ANALYSIS TIME:
TOTAL POSITIONS IN TREE:

Then attach the project

and people could comment in each thread, with a unique thread for each project (if we get a sub-forum as suggested above)

I like this idea so much I think it deserves its own website!  Might need to look into that.
Parent - - By saurus_ (**) [gb] Date 2012-03-16 08:39
I'm using Aquarium and IDEA mainly for my opening preparation. I'm playing conventional chess on an very amateur basis. It is very unlikely that my direct competitors are using the same tool. Even than that wouldn't be an issue for me as I'm playing for fun only.

Because I'm a CAE Engineer my analysis quality is much better than my practical chess skills.

So, If there would be same method for exchange, I would love to contribute everything a can give.

But organizing an working exchange basis for IDEA is not as simple matter.

Ideally there is database with root nodes. Additionally quality information's are required as mentioned below. I don't think hardware is big factor. But Ply depth and engine matter...

I'm looking forward to contribute

Saurus
Parent - By Kappatoo (****) [de] Date 2012-03-19 23:00
I don't like to work with fixed depth. One problem I see is that, e.g., depth = 18 may be acceptable in a middlegames, but it is much less so in an endgame.

I must confess that I am a bit more serious about my OTB play. So, while I am willing to share my analysis with (almost) each person individually, I would feel much less comfortable sharing it with everyone.
I analyzed some opening lines which are not in my repertoire out of curiosity. You inspired me to share some of those.
Parent - By Kappatoo (****) [de] Date 2012-03-19 22:43
Thanks for your enthusiasm. :) Let's see how things develop.
Parent - - By Dadi Jonsson (Silver) [is] Date 2012-03-16 09:12
Start sharing. As soon as 10 projects have been shared, I'll create a new board on the forum for this purpose.
Parent - By mattchess (*) [us] Date 2012-03-16 16:15
I'll post some project trees related to B06 modern and pirc lines this weekend
Parent - - By saurus_ (**) [de] Date 2012-03-17 10:49
Dadi,

how can share projects?
Attachment: franz1-1.xml - Französisch E2-E4_E7-E6 (2k)
Parent - - By saurus_ (**) [de] Date 2012-03-17 11:09
The attachment contains ~120.000 positions after E2-E4  E7-E6 French Defense

Engine: Houndini 1.5 32bit
Quality: min 17 ply
Attachment: franz1.hsh (53k)
Attachment: franz1_discarded.elm (128B)
Attachment: franz1_discarded.hsh (144B)
Parent - By Kappatoo (****) [de] Date 2012-03-19 22:43
Cool, thanks! I will have a look at it (as soon as you find a way to share it :)).
Parent - - By keoki010 (Bronze) [us] Date 2012-03-17 15:13
Dadi, had an article, I think, I'll try to find it if he doesn't answer.  You posted the tree config. You need to post the project epd file or the *.hsh and *.elm files to be able to see the tree.  If you post the epd, put the .xml file in >AquariumData>Config>TreeConfig, stop Aquariium and restart it to initialize the tree config. Put the files where they need to go as shown in the tree. Or in IDEA use manage>Add Results...
Hope that helps.
Parent - - By saurus_ (**) [de] Date 2012-03-18 20:53
Keokio10,

I couldn't upload the *.elm file because it is bigger than 6.000k (it is 16.000k) :fat:
Parent - - By Dhanish (***) [in] Date 2012-03-19 01:13

> I couldn't upload the *.elm file because it is bigger than 6.000k (it is 16.000k)


Without the *.elm file, your tree cannot be seen by others. Have you tried compressing it using the tree utilities?
Parent - - By saurus_ (**) [gb] Date 2012-03-19 07:58
If the idea of exchange should work the limit of 6.000k need to be increased for the exchange area.

Ofcause I can splitt up the tree in smaller trees, but this would add extra work to me and to everbody how want to use his work.

Can the Aquarium team help on this?

For me an working exchange database would be more than a reason to buy the next version of Aquarium...
Parent - By Dhanish (***) [in] Date 2012-03-19 15:00

> If the idea of exchange should work the limit of 6.000k need to be increased for the exchange area.


The Forum administrators may not want to pay for the bandwidth required for the transfer of large files. Of course, there are plenty of sites which specialize in hosting files meant for transfer, you can upload there and give the link here.

But did you try to compress the file using Tree Utilities? I have found that .elm files of 100 MB come down to a few MB after compression.
Parent - By keoki010 (Bronze) [us] Date 2012-03-19 15:40 Edited 2012-03-19 15:42
You can try the tree utilities or export the idea tree to an epd file.  Anything else on the forum we'll have to get Dadi's help.
If you split the file with 7z you could put all 3 together. the elm hsh and xml It shouldn't take a lot of time to do it.
Parent - - By Dadi Jonsson (Silver) [is] Date 2012-03-19 17:46
One option is to get a (free) DropBox account to share your files.
Parent - - By CumnorChessClub (***) [gb] Date 2012-03-21 09:56
Another option which is free and unlimited is to use Opera unite http://unite.opera.com/overview/
Parent - By Dadi Jonsson (Silver) [is] Date 2012-03-23 20:47
I had a quick look and this looks like a simple method to share files.
Parent - - By saurus_ (**) [gb] Date 2012-03-21 06:38
http://dl.dropbox.com/u/68271550/French%20Defence/French_Defence.7z

Ok, after zipping the file it was still 7.500k (limit is 6.000k)

So I went the second way of creating a DragBox Account.

I don't think that the charing of projects will fly if this requires an external space. :confused:

The link should now contain all information in on 7z file.

If some one can try out this is working.

thanks saurus
Parent - By Dhanish (***) [in] Date 2012-03-21 07:39

> If some one can try out this is working.


Yes, the tree is OK. Thank you, though I have given up playing the French!
Parent - - By mattchess (*) [us] Date 2012-03-21 08:11
How many positions are in that project?  The file sizes seem huge compared to my own projects - want to compare
Parent - - By saurus_ (**) [gb] Date 2012-03-21 11:52
There are about 126.000 positions in the project. (it is sill incrasing)

The tree is optimised for learning the opening. So it is more wide than deep.

I'm a patzer and most of my oponents are also patzer. Therefore I'm also looking for moves what good players are unlikely to play.

On the other side it doesn't makes sense for me to look to deep into the lines, because there will be nearly always  be a mistake of one of the players within the first 10 movers.

For advanced players this tree will be of limited value.

Ofcause you can alwasy use it as starting point to expand the lines evern further.

Do you also see my color coding and root nodes?
Parent - - By Kappatoo (****) [de] Date 2012-04-08 20:17
Only 126.000? I downloaded your tree, it is displaying incredible numbers. (Cf. dhanish's tree shot above - that's also what I'm seeing.)
Parent - By saurus_ (**) [de] Date 2012-04-11 19:18
Kappatoo,

The n-values are rubbish. I guess this has to do with me intensive usage of the master tree. 126.000 should be correct size.

I don't know how to correct the n-values.
Parent - - By Dadi Jonsson (Silver) [is] Date 2012-03-23 20:50

> I don't think that the charing of projects will fly if this requires an external space. :confused:


Once we have enough files shared, I'll see what I can do.
Parent - - By Vempele (Silver) [fi] Date 2012-03-23 21:31 Edited 2012-03-23 21:35
So people won't share files until they can do so effortlessly, and you won't make it effortless for them until they've shared enough files. Is this an experiment to see which appears first, the chicken or the egg?
Parent - By Dadi Jonsson (Silver) [is] Date 2012-03-23 21:52
I doubt I will come up with something that requires less effort than DropBox (not to mention effortless). It's more about making the files easier to access after they have been shared. While there are just a few files, all in the same topic, there isn't much need for organizational overhead.
Parent - - By nimzo5 (**) [us] Date 2012-03-17 12:51 Edited 2012-03-17 14:05
I like this idea and would be willing to participate if the analysis fits any of the openings I am interested in.

Alternately, I would be willing to discuss starting a new project and dividing up the work.
Parent - - By Kappatoo (****) [de] Date 2012-03-19 22:54
Good. Any proposals as to which lines you would like to investigate?
Parent - By nimzo5 (**) [us] Date 2012-03-24 12:22
I am currently cleaning up and organizing my analysis so that it would be sensible to someone besides myself.

Without outlining my repertoire I am interested in d4 openings from White's perspective and the Sicilian. I could do some e4 as well, but it hasn't been my primary focus to date.
Parent - - By ppipper (*****) [es] Date 2012-03-23 23:19
I am interested in this idea. I dont mind sharing my trees, provided I can get similar amount of information and, on average, of similar quality.
Parent - - By Kappatoo (****) [de] Date 2012-03-24 09:58
Great. In your case, it won't be easy to provide you with similar quality, though.
Parent - - By nimzo5 (**) [us] Date 2012-03-24 12:24
Maybe we should choose a particular starting point, divide the principal variations and compare to see if we like each others analysis?
Parent - - By Kappatoo (****) [de] Date 2012-03-24 13:06
Sounds good. Let's see if we find some more volunteers.
As to the lines: I can make one or more proposals if no-one else does. Let's see what others say.
Parent - - By ppipper (*****) [es] Date 2012-03-24 20:15
if there are not any suggestions, I would go for CaroKann, English or Dutch
Parent - By nimzo5 (**) [us] Date 2012-03-25 11:10
hmm, I have some analysis in these openings esp. from Black's perspective facing the English and nh3 systems vs the Dutch. I will take a closer look at what I have done already in those.
Parent - By ppipper (*****) [es] Date 2012-03-24 14:00 Edited 2012-03-24 14:05
yes, +1.

In order to be organized, we probably have to exchange our emails, or do it everything publicly? I must admit I like the idea to share my analysis with the ones who share theirs.
Parent - By ppipper (*****) [es] Date 2012-03-24 14:03
of course trees quality is important, but IMO what makes the main difference in order to win tournaments is certain type of chess comprehension, the one that makes you to choose one move over others, even if they have better evaluations.
Parent - - By Kappatoo (****) [de] Date 2012-04-08 20:25 Edited 2012-04-08 20:37
I am trying to upload my tree on Topalov's Nxf7 in the Anti-Moscow: (Let's see if I manage that ...)

1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 e6 5. Bg5 h6 6. Bh4 dxc4 7. e4 g5 8. Bg3
b5 9. Be2 Bb7 10. O-O Nbd7 11. Ne5 Bg7 12. Nxf7 *
r2qk2r/pb1n1Nb1/2p1pn1p/1p4p1/2pPP3/2N3B1/PP2BPPP/R2Q1RK1 b kq - 0 12


It is a pretty small tree; I am sure many of you have deeper analyses on this position than I do. If you are nevertheless interested, please have a look. Any comments are welcome.
If you would like to hear my assessment on what I consider to be the critical lines, feel free to ask, of course.
Attachment: D43Topalovs12.Nxf7.elm (520k)
Attachment: D43Topalovs12.Nxf7.hsh (872B)
Parent - By nimzo5 (**) [us] Date 2012-04-09 00:17
Thanks for sharing this, I am busy for the next couple days but will look at it ASAP.
- - By buffos (Silver) [gr] Date 2012-03-17 11:08
Why tree-project sharing is not good at this stage

The basic problem is that there will be positions with incorrect evaluations (for example best move is 0.00 and there is a move with score 0.80 that would "never be expanded" by idea in a very large tree) which the user getting the tree will never know about them.

Without "watching" the tree expanding, you will never actually understand the position, how it failed, why... etc... Those are REALLY important information

For this to work, we need a "logging" utility, that would write down, the progress of the best line, how it changed, how many times, why... give details about the lines... etc.
Only with that information in hand you can continue to work on some else's tree.

Ofcourse if i know the other person, i trust his skills, then i would certainly go ahead and use the tree.

At least this is my opinion
Parent - - By Kappatoo (****) [de] Date 2012-03-19 22:53
I agree with most of what you say after "... not good at this stage." However, I don't see how this supports your first sentence. Of course, one should not blindly trust such a tree, one has to be able to read it correctly, etc. But the tree nevertheless contains a lot of information which may be very useful.

If I were to share a tree, I would be willing to answer questions on which moves I consider to be most promising, which lines I think are critical, at which points the tree is more reliable and at which less so, etc. So I agree with you that one can easily be misled if one only follows the tree's main lines (based on eval).
Parent - - By Wrath of the Titans (****) [sa] Date 2012-03-24 15:49
I suggest to have one common settings to all, I'm willing to exchange projects.
Parent - - By ppipper (*****) [es] Date 2012-03-24 20:38
yes,  that is of course needed. Otherwise the tree would be a total mess
Parent - - By nimzo5 (**) [us] Date 2012-03-28 00:00
So what would that standard be? I typically use multiple engines in IA to test my idea results.
Parent - By mattchess (*) [us] Date 2012-04-09 04:35
I think it would be difficult/impractical to integrate someone else's project into your master tree due to engine/setting differences etc - however viewing the tree as stand alone would provide useful information and perhaps provide insight into how you might want to generate your own project.
Up Topic Rybka Support & Discussion / Aquarium / Tree exchange

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill