Autonomous Subproject

Message boards : News : Autonomous Subproject
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 7 · 8 · 9 · 10 · 11 · 12 · 13 . . . 15 · Next

AuthorMessage
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 565 - Posted: 7 Jan 2019, 1:21:33 UTC - in response to Message 564.  

For PADS experiment:
I compiled both of the basic generators with freebasic compiler and verified that they still produce same results. Then I created a program to link both the generators and kanonizator together. This program finds KF DLKs from a "row" in just 25 seconds. The original took about 8 minutes.

It is very good!
Do your programs run in Windows system?

Unfortunately, family_mar still takes much longer time to process it. This means that creating BOINC application will be easier.

Yes you are right.
I hope that someday ice00 will agree to launch this experiment in the ODLK1 project.
I found a programmer who is ready to help him with this. This programmer is actively working with BOINC projects.

I am waiting for one batch to finish. When it does, i will send you the results and all other results that I have.

OK.
ID: 565 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 566 - Posted: 7 Jan 2019, 7:57:39 UTC

XAVER
I received and processed your results (results_xav04.txt).
Found 281 unique CFs ODLS.

All is well.
Thanks!
ID: 566 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tomas Brada

Send message
Joined: 22 Jan 18
Posts: 63
Credit: 373,625
RAC: 0
Message 567 - Posted: 7 Jan 2019, 20:58:09 UTC - in response to Message 565.  

Unfortunately, family_mar still takes much longer time to process it. This means that creating BOINC application will be easier.

Yes you are right.
I hope that someday ice00 will agree to launch this experiment in the ODLK1 project.
I found a programmer who is ready to help him with this. This programmer is actively working with BOINC projects.


Mistake. I mean that with my changes to the programs, creating a BOINC app will be easier.

And that family_mar still takes majority of the time. The function that takes the most time is trans_dlx::poisk_simm . I do not understand the math, so I will not work on that function.

I updated the repository. The programs now print most of the messages in English. Added support for input/output arguments to family_mar, so scripting will be easier. I finished the PADLS combined program, it too supports arguments. Other changes in the commit list.

The main algorithm of PADLS generators is not published.

I did not test the programs on Windows yet. I would have to use virtual machine. Windows user should test: install GIT, CMake, C++ compiler, FreeBasic compiler and dependencies. Anyway, I think it should work. I did not use any Linux specifics. So once you can compile, it should work.

$ ./psevdoass.exe -abk tmp.txt 4 3 7 2 1 9 5 6
I9= 2
I9= 3
I9= 4
I9= 5
I9= 6
I9= 7
I9= 8
I9= 9
SQUARES GENERATED 89312
ASS_DLK10A: 89312
Step took: 0.565635 s
I1= 2
I1= 3
I1= 4
I1= 5
I1= 6
I1= 7
I1= 8
I1= 9
SQUARES GENERATED 478835
PSEVDOASS_DLK_NEW: 478835
Step took: 3.11591 s
Have DLS: 568147
Found CF: 304574
Step took: 0.467369 s
КФ записаны в файл:     tmp.txt
Step took: 0.130915 s
Total run time: 3.71423 s

$ ./family_mar.exe some_squares.txt tmp.txt
Поиск марьяжных ДЛК (кроме симметричных) для семейства ЛК
Have LS: 193
ETA: 0h 0.0666667m 
Checked DLK: 116437
Run Time (s): 6.64513
Found Fancy DLS: 0
Они записаны в файл tmp.txt
ID: 567 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tomas Brada

Send message
Joined: 22 Jan 18
Posts: 63
Credit: 373,625
RAC: 0
Message 568 - Posted: 7 Jan 2019, 21:09:25 UTC

$ ./psevdoass.exe
Expected 9 arguments
Experimental generator of Pseudo-Associative Diagonal Latin Squares.
generator.exe -abk output X1 X2 X3 X4 X5 X6 X7 X8
 -a : use ASS_DLK10A generator
 -b : use PSEVDOASS_DLK_NEW generator
 -k : find canonicial form
 output : file to write results to
 X1-8   : generator parameters

$ ./family_mar.exe 
Поиск марьяжных ДЛК (кроме симметричных) для семейства ЛК
Expected 2 arguments
Search for Fancy Diagonal Latin Squares.
family_mar.exe input output
 input : file to read Latin Squares from
 output: file to write Fancy Diagonal Latin Squares to
ID: 568 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 569 - Posted: 8 Jan 2019, 3:00:16 UTC - in response to Message 567.  


Mistake. I mean that with my changes to the programs, creating a BOINC app will be easier.

Tomas Brada
sorry, I misunderstood you.

You have made an excellent version of the PADLS experiment software for the Linux system.
Thanks.
This is exactly what ice00 needs!

ice00
now you can run the experiment in a BOINC project without problems?

Tomas Brada can help you with this.
Please tell us your opinion.

In addition, you could participate in autonomous subproject.
You have a server that you can download with this experiment.
Software is now for the Linux system.
Do you want to participate in the experiment?
ID: 569 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tomas Brada

Send message
Joined: 22 Jan 18
Posts: 63
Credit: 373,625
RAC: 0
Message 570 - Posted: 8 Jan 2019, 10:14:14 UTC - in response to Message 569.  

You have made an excellent version of the PADLS experiment software for the Linux system.
Thanks.
This is exactly what ice00 needs!

ice00
now you can run the experiment in a BOINC project without problems?

Tomas Brada can help you with this.
Please tell us your opinion.


My modified program is not limited to Linux system, it should compile and run on Windows system too.

For boinc it is better to have a single executable that uses the boinc api. Currently it is two executables (generator and family_mar checker). It is possible to run in wrapper mode, but boinc api is better, because it can report % progress and allow suspend. I could combine the generator and checker. The second step, adding boinc api is harder for me, because a library is missing a file on my system and to get it, I would have to install it manually, which takes time.

How many WUs do you have, or will have, for this experiment?

How soon would you like to see the results?
ID: 570 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 571 - Posted: 8 Jan 2019, 12:30:28 UTC - in response to Message 570.  
Last modified: 8 Jan 2019, 12:32:56 UTC

The second step, adding boinc api is harder for me, because a library is missing a file on my system and to get it, I would have to install it manually, which takes time.

Please see this message.
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=45&postid=309#309

Are we talking about this?

How many WUs do you have, or will have, for this experiment?
I already have 1235 rows, this is 1235 WUs.
In the future, there may be more than 15,000 WUs.

How soon would you like to see the results?

I'm not hurrying.
But it does not depend on me, but on ice00. I do not have access to the server project ODLK1.
ID: 571 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tomas Brada

Send message
Joined: 22 Jan 18
Posts: 63
Credit: 373,625
RAC: 0
Message 572 - Posted: 8 Jan 2019, 13:55:50 UTC - in response to Message 571.  

The second step, adding boinc api is harder for me, because a library is missing a file on my system and to get it, I would have to install it manually, which takes time.

Please see this message.
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=45&postid=309#309
Are we talking about this?

Yes, that's it. Adding boinc api to single well-behaved program is simple. It is just adding #include boinc_api.h, boinc_init, boinc_fraction_done, boinc_finish and boinc_resolve_filename. I can do that, but I wont be able to test, because #include boinc_api.h is not working for me, because a library on my pc is incorrect. I do not want to solve the library issue now.
ID: 572 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 573 - Posted: 8 Jan 2019, 14:08:23 UTC

Tomas Brada
generators are written in BASIC.
Is this an obstacle to using the API BOINC?

I was told that it is necessary to rewrite the generators on C++.
It's right?
ID: 573 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 574 - Posted: 8 Jan 2019, 14:15:00 UTC - in response to Message 572.  

I can do that, but I wont be able to test, because #include boinc_api.h is not working for me, because a library on my pc is incorrect. I do not want to solve the library issue now.

Unfortunately, we have no response from ice00 :(
Either he does not read the topic, or he does not want to answer.
ID: 574 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tomas Brada

Send message
Joined: 22 Jan 18
Posts: 63
Credit: 373,625
RAC: 0
Message 575 - Posted: 8 Jan 2019, 14:20:12 UTC

https://github.com/gridcoin-community/odlk-progs
psevdoass: Add family_mar searching, with paralel support



I linked the PADLS with family_mar searching. Now only think left for boinc app is to add the boinc api calls (previous message).

I also added parallel execution support. On my PC one workunit runs for only 15 minutes. I verified the output matches.

./psevdoass.exe -abkmt output.txt 4 3 7 2 1 9 5 6
I9= 2
I9= 3
I9= 4
I9= 5
I9= 6
I9= 7
I9= 8
I9= 9
SQUARES GENERATED 89312
ASS_DLK10A: 89312
Step took: 0.565902 s
I1= 2
I1= 3
I1= 4
I1= 5
I1= 6
I1= 7
I1= 8
I1= 9
SQUARES GENERATED 478835
PSEVDOASS_DLK_NEW: 478835
Step took: 3.16772 s
Have DLS: 568147
Found CF: 304574
Step took: 0.596584 s
Search for Fancy by Family Diagonal Latin Squares (not symmetric)
nb_threads: 16
Checked DLS: 118931254
CPU time: 12127.2
Step took: 811
Found Fancy DLS: 20
Fancy DLS written to:   output.txt
Step took: 0.00805 s
Total run time: 12130.9 s
ID: 575 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Tomas Brada

Send message
Joined: 22 Jan 18
Posts: 63
Credit: 373,625
RAC: 0
Message 576 - Posted: 8 Jan 2019, 14:22:55 UTC - in response to Message 573.  

generators are written in BASIC.
Is this an obstacle to using the API BOINC?
I was told that it is necessary to rewrite the generators on C++.
It's right?

Not needed. The API can be used as long as part of the program is in C++. The generators can stay in basic.
ID: 576 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 577 - Posted: 8 Jan 2019, 14:31:05 UTC - in response to Message 576.  

generators are written in BASIC.
Is this an obstacle to using the API BOINC?
I was told that it is necessary to rewrite the generators on C++.
It's right?

Not needed. The API can be used as long as part of the program is in C++. The generators can stay in basic.

Thanks.
It's good.
ID: 577 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 578 - Posted: 8 Jan 2019, 14:44:03 UTC

XAVER
please tell us how experiment PADLS started.
All is well?
ID: 578 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
XAVER
Avatar

Send message
Joined: 27 Dec 18
Posts: 12
Credit: 5,932
RAC: 0
Message 579 - Posted: 8 Jan 2019, 19:53:28 UTC

All is well. Currently i am checking row 9 out of 25. Shall i send you all 25 or do you prefer smaller batches (say 10 rows)?

Kind regards
XAVER
ID: 579 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 580 - Posted: 9 Jan 2019, 2:06:43 UTC - in response to Message 579.  

Shall i send you all 25 or do you prefer smaller batches (say 10 rows)?

You can send any portions.
ID: 580 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 581 - Posted: 9 Jan 2019, 6:33:04 UTC

XAVER
I received and processed your results for rows 1-10 in the PADLS experiment.
Found 87 unique CFs ODLS.

All is well.
Thanks!
ID: 581 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 582 - Posted: 9 Jan 2019, 7:55:41 UTC
Last modified: 9 Jan 2019, 15:22:42 UTC

Dear participants of ODLK and ODLK1 projects!

In the autonomous subproject start a new experiment PADLS (pseudo-associative DLSs).
The first results are obtained from XAVER.
I also perform this experiment on my PC.

Join now!

My script works in the Windows system.
Tomas Brada has modified software for the Linux system.

To get WUs write to me.
ID: 582 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 22 Oct 17
Posts: 470
Credit: 0
RAC: 0
Message 586 - Posted: 10 Jan 2019, 18:14:05 UTC

XAVER
I received and processed your results for rows 11-20 in the PADLS experiment.
Found 107 unique CFs ODLS.

All is well.
Thanks!
ID: 586 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ice00
Project administrator
Project developer

Send message
Joined: 28 Oct 17
Posts: 108
Credit: 46,930
RAC: 16
Message 587 - Posted: 10 Jan 2019, 18:15:12 UTC

Sorry for the delay in read forum but I'm in trouble with Electronic Invoice that is mandatory in Italy from 1/1/2019

So far when testing BASIC program using free compiler in server did gave compilation error, so for this it was needed a C++ conversion.

I'll test the compilation of Github code on testing server as soon as possible

thanks
ID: 587 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 . . . 7 · 8 · 9 · 10 · 11 · 12 · 13 . . . 15 · Next

Message boards : News : Autonomous Subproject


©2019 Progger & Stefano Tognon (ice00)