re 5-1 EGTB statistics

[vb4]

Hi Josh,

I have not rcvd any response from you and just doing a follow up to request those stat files that you alrady have and to see if you are planning on doing the remaining 5-1 tables. Again my email address is

vb4@prodigy.net

Thank you,

Les

[jshriver]

Hi Josh,

I have not rcvd any response from you and just doing a follow up to request those stat files that you alrady have and to see if you are planning on doing the remaining 5-1 tables. Again my email address is

vb4@prodigy.net

Thank you,

Les

Hrm not sure what you mean by stat files as right now I have no files pertaining to 5-1. I did try running FEG.exe to generate some but it kept crashing after a couple days running.

Once I have more free time I'm going to try it again, though probably not till around November.

Sorry for the delay.
-Josh

[vb4]

Hi Josh,

I think my comments were directed to Guido. I am not sure what happened maybe I screwed up <S>. Anyway Guido posted not to long ago that he has run a few of the 5-1 egtbs and I was inquiring about him sending me the stat files.

Sorry,

Les

Hi Josh,

I have not rcvd any response from you and just doing a follow up to request those stat files that you alrady have and to see if you are planning on doing the remaining 5-1 tables. Again my email address is

vb4@prodigy.net

Thank you,

Les

Hrm not sure what you mean by stat files as right now I have no files pertaining to 5-1. I did try running FEG.exe to generate some but it kept crashing after a couple days running.

Once I have more free time I'm going to try it again, though probably not till around November.

Sorry for the delay.
-Josh

[jshriver]

I generated KQK, KQQK, KQQQK, and in the process of generating KQQQQK

*crossing fingers* if it doesnt crash I'll share the egtb.

One thing I noticed that might be a problem, it's creating a lot of small files.
Last count was over 12k files. So wondering if it's crashing because it's hitting the max limit of files in a directory. Will monitor it more when I get off of work.

As of right now the KQQQQK directory is 1.2gigs

Is there a particular pawnless 5-1 you want to study first?

-Josh

[guido]

[quote="vb4"]Hi Josh,

I think my comments were directed to Guido. I am not sure what happened maybe I screwed up <S>. Anyway Guido posted not to long ago that he has run a few of the 5-1 egtbs and I was inquiring about him sending me the stat files.

Sorry,

Les

[quote="jshriver"]

Very strange!
I sent you an e-mail with 15 stats of 5-1 TBs on Sept,22nd and today, reading your message, I sent again the same e-mail with the attachment.
Is it correct your address?
If there are problems I can send the stats to this NG.

guido

[jshriver]

Very strange!
I sent you an e-mail with 15 stats of 5-1 TBs on Sept,22nd and today, reading your message, I sent again the same e-mail with the attachment.
Is it correct your address?
If there are problems I can send the stats to this NG.

guido

Can you send them to me too? jshriver <at> gmail.com

-Josh

[vb4]

Hi Josh,

I think my comments were directed to Guido. I am not sure what happened maybe I screwed up <S>. Anyway Guido posted not to long ago that he has run a few of the 5-1 egtbs and I was inquiring about him sending me the stat files.

Sorry,

Les

Very strange!
I sent you an e-mail with 15 stats of 5-1 TBs on Sept,22nd and today, reading your message, I sent again the same e-mail with the attachment.
Is it correct your address?
If there are problems I can send the stats to this NG.

Hi Guido,

I am not sure what happened so here is my email address again.

VB4@PRODIGY.NET

Please sned thme one more time and put in the subgect line EGTB stats. I wonder if my filter kicked them out although I dont think so. I will then reply back to you when I get them. I check my email every day so if you dont hear from me the day you send them then kindly check back here for my comments. Cant wait to see them. I assume you are only trying to send me the *.tbs stat files?

Looking for your response,

Thanks again Guido,

Les

guido

[vb4]

I generated KQK, KQQK, KQQQK, and in the process of generating KQQQQK

*crossing fingers* if it doesnt crash I'll share the egtb.

One thing I noticed that might be a problem, it's creating a lot of small files.
Last count was over 12k files. So wondering if it's crashing because it's hitting the max limit of files in a directory. Will monitor it more when I get off of work.

As of right now the KQQQQK directory is 1.2gigs

Is there a particular pawnless 5-1 you want to study first?

Hi Josh,

It really does not matter which you do, I am humble and accept any work you are working hard to develope. When I rcv the stats files from Guido I will send them to you.

Thanks and good luck,

Les

-Josh

[guido]

Hi Josh and vb4,

All is very strange! I tried to send the stats 3 times to vb4 and once to Josh, but I never received an answer, so I decided to put them here in an attachment.

The file stats.txt contains 15 stats files of 5-1 TBs. Unfortunately I don't use the same format of Eugene Nalimov, so it could be difficult to compare my stats with his. Moreover in any case there would be some differences between them; in my TBs:

- The total number of positions (legal + illegal) is larger for about 10-15% than that of EN TBs
because I don't eliminate from the files the positions in which there is a check made by a man close
to the opponent king.

- In many cases the number of legal positions is lesser than that of EN TBs because I eliminate from the computation positions duplicated by symmetry when both the kings are on the main diagonal (a1-h8). So the numbers of positions with a given result can be different and should always be equal_or_lesser than EN numbers.

- I don't keep into account ep capture. This can bring some small differences in endings where ep
capture is possible (not in these 15 stats and in all 5-1 endings).

- For the names of my files I use the Italian (and Spanish) names of men (King = (R)e, Queen = (D)onna, Rook = (T)orre, Bishop = (A)lfiere, Knight = (C)avallo, Pawn = P)edone) and colours (White = (B)ianco, Black = (N)ero). So rddddr is equivalent to kqqqqk, etc.

I concatenate all the single files together inserting an initial line with the English name of the ending.
If someone finds any difficulty in these files or big differences with EN stats, write me pls.

If someone needs to have lists of the positions with a specific result, I can extract and send these lists (if they are not too long!) with the men indicated in English notation. In short I'll insert an option to have the output list in FEN format (at present it is not in this format). Can someone
suggest me a site for an exact definition of EPD/FEN?

I have intention to produce other 5-1 endings, but before continuing and generate bigger files I have to modify my program to reduce the use of disk.

Guid

[guyhaw]

Note - White to move draws, where White has the 5 men - g

[jshriver]

Greetings,

FEG.exe did complete a 5-1 egtb successfully.. or *I guess successfully*

However it seems odd.
When I had checked the status.. the temp directory had over 1.6gigs of data for the current egtb.

Once it finished, cleared the temp data... the resulting 5-1 egtb was less than 100 megs. Think it was around 64-70megs. This doesnt sound right.

I'm really not sure how to test it. I'm not sure if the temp data was just raw data, and that the resulting set was compressed.. but I still can't believe it can compress over 1.6gig (possibly more) into around 64-70megs.

Anyone here familiar with FEG.exe and give some pointers?

If the results are correct, hopefully in a month or two once I find a home I can resume tests.

Till then I unfortunately have to put all my hardware into storage.

-Josh

[mbourzut]

Th ese results are quite reasonable, especially if there were many like pieces in the ending, as FEG does not implement permutation symmetry until the final compression stage.

-Marc

Greetings,

FEG.exe did complete a 5-1 egtb successfully.. or *I guess successfully*

However it seems odd.
When I had checked the status.. the temp directory had over 1.6gigs of data for the current egtb.

Once it finished, cleared the temp data... the resulting 5-1 egtb was less than 100 megs. Think it was around 64-70megs. This doesnt sound right.

I'm really not sure how to test it. I'm not sure if the temp data was just raw data, and that the resulting set was compressed.. but I still can't believe it can compress over 1.6gig (possibly more) into around 64-70megs.

Anyone here familiar with FEG.exe and give some pointers?

If the results are correct, hopefully in a month or two once I find a home I can resume tests.

Till then I unfortunately have to put all my hardware into storage.

-Josh

[jshriver]

Th ese results are quite reasonable, especially if there were many like pieces in the ending, as FEG does not implement permutation symmetry until the final compression stage.

-Marc

The egtb I had generated was knnnnk. Thanks for clearing up the size problem for me. Once things settle a bit I'll put some new files into the queue.

Hope all is well.
-Josh

[vb4]

Th ese results are quite reasonable, especially if there were many like pieces in the ending, as FEG does not implement permutation symmetry until the final compression stage.

-Marc

The egtb I had generated was knnnnk. Thanks for clearing up the size problem for me. Once things settle a bit I'll put some new files into the queue.

Hi Josh,

Hey that is really great you attempting to generate these 5-1 files. For my personal self Josh as I explained in previous postings I am only interested in the *.tbs stat files. Thank you so much for giving this a go and whenever you find time send me the *.tbs stat file. Looking forward to more of them <s>.

Thanks,

Les
Hope all is well.
-Josh

[guido]

Hi Josh,

Hey that is really great you attempting to generate these 5-1 files. For my personal self Josh as I explained in previous postings I am only interested in the *.tbs stat files. Thank you so much for giving this a go and whenever you find time send me the *.tbs stat file. Looking forward to more of them <s>.

Thanks,

Les
Hope all is well.
-Josh[/quote][/quote]

Hi vb4,

are you still interested in my 5-1 EGTB stats?
I have some more kqqqbk, kqqqnk, kqqqpk, kqqrrk, kqqbbk, kqqnnk).
In my stats there is another difference with Nalimov and other stats because I set as illegal also
some situations of double checks (e.g. two knights and others).
If you are interested I'll send it as an attachment because your private address always rejected my
messages. With Josh address I hadn't any problem.
A curiosity: I generated also knnnkb and I found that there is one position (Ka1, Nh1, Nb3, Nh8, kc2, bg8) in the wtm file where white wins in 96 moves!

Guido

[Kirill Kryukov]

A curiosity: I generated also knnnkb and I found that there is one position (Ka1, Nh1, Nb3, Nh8, kc2, bg8) in the wtm file where white wins in 96 moves!

Guido

Hi Guido! In KBNNKN you will find checkmate in 109! (I collect longest checkmate positions on this page).

[vb4]

Hi Josh,

Hey that is really great you attempting to generate these 5-1 files. For my personal self Josh as I explained in previous postings I am only interested in the *.tbs stat files. Thank you so much for giving this a go and whenever you find time send me the *.tbs stat file. Looking forward to more of them <s>.

Thanks,

Les
Hope all is well.
-Josh

[/quote]

Hi vb4,

are you still interested in my 5-1 EGTB stats?
I have some more kqqqbk, kqqqnk, kqqqpk, kqqrrk, kqqbbk, kqqnnk).
In my stats there is another difference with Nalimov and other stats because I set as illegal also
some situations of double checks (e.g. two knights and others).
If you are interested I'll send it as an attachment because your private address always rejected my
messages. With Josh address I hadn't any problem.
A curiosity: I generated also knnnkb and I found that there is one position (Ka1, Nh1, Nb3, Nh8, kc2, bg8) in the wtm file where white wins in 96 moves!

Hi Guido,

Absolutley still interested in them. Since you have experienced problems for some reason with my email address let me make a suggestion. To be sure I get them post them here as an attachment and if you can please try one more time to my email address. It is vb4@prodigy.net I have never had a problem with anyone sending me email so I am puzzled but perhaps in the subject line you can put "Chess" Thanks again,

Les

Guido[/quote]

[guido]

A curiosity: I generated also knnnkb and I found that there is one position (Ka1, Nh1, Nb3, Nh8, kc2, bg8) in the wtm file where white wins in 96 moves!

Guido

Hi Guido! In KBNNKN you will find checkmate in 109! (I collect longest checkmate positions on this page).

Hi Kirill,

Thanks for your interesting collection of the longest chechmates.
I have not yet generated kbnnkb, so I didn't know such info.
But the reason of my surprise is in the fact that in the position mentioned by me in the endgame knnnkb, the black bishop is not captured in the first move(s) (in this case the white would have won in less moves), and obtaining a draw for the black would seem very easy because it would be sufficient to change the black bishop with any of the three knights, as knnk is always a draw with few particular exceptions. This explains my surprise.
In the endgame kbnnkn mentioned by you the black knight should capture or change with the white bishop only, because if the black
knight changes with another knight, the endgame kbnk, differently from knnk, is generally easily won by white.

Ciao
Guido

[jshriver]

Seems the first run wasn't right, but it only took 2 days to regenerate KNNNNK
The egtb is 1.5 gigs. However I dont see any statistics or logs files. Will have to dig around some more.

-Josh