I will try to process the results.

Tomas Brada

I ask you to make a software for processing results for the Linux system.

This is a problem for ice00, which he still has not solved.

Natalia Makarova
I made a very good progress on the result processing. But I need help. What outputs are desired? And in what format?
You want the CF ODL squares, that is for sure.
Then cf odl squares grouped by number of co-squares of each.
And then number of squares in each group ( count[1] = 1898 count[2] = 12 ... ).
I can give file with CF ODLSs together with their KF co-squares (like in your diagram).
I can make the output whatever you want!
Or should I replicate exact format as Belyshev's programs?

Second: in family_mar program, there is condition:

if(!is_simm(kf, lin))

, so the program will not produce symmetric(?) solutions. However the program klpmd does not have such condition. Is that correct? Which should be used for the PADLS experiment?

---

(me) I can give file with CF ODLSs together with their KF co-squares (like in your diagram).

Ignore this. This does not make sense to do and is different than diagram.

I made a very good progress on the result processing. But I need help. What outputs are desired? And in what format?

When processing the results, I need an out_klpmd.txt file, as well as files with groups of pairs of ODLS (three or more)

out_kf_3.txt
out_kf_4.txt
etc.

Second: in family_mar program, there is condition:

if(!is_simm(kf, lin))

, so the program will not produce symmetric(?) solutions. However the program klpmd does not have such condition. Is that correct? Which should be used for the PADLS experiment?

The PADLS experiment uses the family_mar program.

Since all symmetric solutions were found in another experiment, it is not necessary to look for them again.

XAVER
I received and processed your results for rows 61-80 in the PADLS experiment.
Found 151 unique CFs ODLS.

All is well.
Thanks.

XAVER
I received the processed results for October 2018.
All is well.
I will publish these results, as will be ready results for September.

Thank you very much.

XAVER
I received processed results for September.
All is well.
Thanks.

I will publish the results for September - October 2018 in the near future.

The results of the project ODLK1 for September - October 2018 are published here
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=46&postid=647#647
and here
https://boinc.progger.info/odlk/forum_thread.php?id=86&postid=2966#2966

XAVER
I received processed results for November.
All is well.
Thanks.

I will publish the results when the results for December are ready.

XAVER
I received processed results for December.
I will check these results tomorrow and publish the results for November - December 2018.

Thank you very much!

I published compatible programs for processing results to my repository.
They are two programs: postprocess and ortogoncnt. Both are multithreaded programs and work on linux system. They should work on windows too, if compiled.
Postprocess program reads the results from boinc project, finds additional CF ODLSs (like in zamyk.bat) and adds them to database file.
Ortogencnt program removes duplicate squares from database, computes statistics and outputs out_ortogon.txt, out_kf_X.txt, and out_kf_mates.txt files, like zamyk.bat.
More information in the Readme file.

The first program (postprocess) should have been part of the boinc application, it would save a lot of time in result processing. I think it should be added to new PADLS boinc application.

I am currently processing results from September and I will compare them to XAVER's result to validate. Hopefully I have not included error.

I published compatible programs for processing results to my repository.

Tomas Brada
thank you very much for your work.

Notice the readme.txt file

© 2019 Progger & Stefano Tognon (ice00)

ODLK1 project results for September 2018

407995 unique CFs ODLS.

Results processed by XAVER.
The uniqueness of the results is checked by Natalia Makarova in the framework of the ODLK1 project, without taking into account the results of the ODLK project.

Created on January 19, 2019

First you will receive other CFs ODLS data:

The Application odlk3 is 226322 CFs.
The Application odlkmax is 184328 CFs.

These results received XAVER.

So we have
226322 +184328 = 410650 CFs.

Now pay attention to this

The uniqueness of the results is checked by Natalia Makarova in the framework of the ODLK1 project, without taking into account the results of the ODLK project.

The database currently contains more than 5,000,000 unique CFs.
I took now only part of the database - from the project ODLK1.
I perform a check for uniqueness in stages, it is very inconvenient.
It is necessary to automate this process.
I will write about it in detail in the subject
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=54

Yes, I have long been asking ODLK and ODLK1 project administrators to automate the processing of results.
But this is still not done.
I processed the results a year by manual. Now I do not do it.

ice00 deals with this problem, but the problem is not solved.

I offered to publish the raw results from ODLK1 project.
Results in the ODLK project are available to everyone.

Processed results of the ODLK1 project for November - December 2018 are published here
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=46&postid=659#659
and here
https://boinc.progger.info/odlk/forum_thread.php?id=86&postid=2967#2967

XAVER
I received and processed your results for rows 81-100 in the PADLS experiment.
Found 184 unique CFs ODLS.
Fine!
Thanks.

I sent you a new 50 rows.

XAVER
I received and processed your results for rows 101-120 in the PADLS experiment.
Found 145 unique CFs ODLS.

All is well.
Thanks.

Tomas Brada
currently, the family_mar program works in one thread. It is very slow.

Is it possible to make this program multi-threaded?

PS. I saw multi-threaded programs when I participated in a contest for programmers.
My teammate wrote such programs.
These are good programs!
The user sets the number of threads.

To each!
Of course, you can make an artificial parallelization.

For example, run several programs in different folders.
It works well.

Or run several programs on different computers.

The version of family_mar in my repository is multi-threaded. It just needs to be compiled for your system. For example on my system:

git clone --recurse-submodules https://github.com/gridcoin-community/odlk-progs.git
cd odlk-progs
cmake -DCMAKE_BUILD_TYPE=Release .
make

For windows system example: https://github.com/tudelft3d/masbcpp/wiki/Building-on-Windows-with-CMake-and-MinGW