[Wien] Segmentation fault in w2w

[Rubel, Oleg]

Thank you for sharing that!

Would it be possible to share (maybe privately) the case.struct file and _exact_ sequence of commands (incl. initialization, scf run, etc.) with all options. Then it will be possible to reproduce and debug.

Thank you in advance
Oleg

--
Oleg Rubel (PhD, PEng)
Department of Materials Science and Engineering
McMaster University
JHE 359, 1280 Main Street West, Hamilton, Ontario L8S 4L8, Canada
Email: rubelo at mcmaster.ca
Tel: +1-905-525-9140, ext. 24094
Web: http://olegrubel.mcmaster

________________________________________
From: Wien <wien-bounces at zeus.theochem.tuwien.ac.at> on behalf of Niraj Aryal <debonairniraj at gmail.com>
Sent: Friday, September 11, 2020 14:02
To: A Mailing list for WIEN2k users
Subject: [Wien] Segmentation fault in w2w

Dear w2w experts and developers,

I am using Wien2k version 19.2 in scientific linux 7.5 using intel compilers (2018).

I am trying to wannierize f-electron system with antiferromagnetic magnetic ordering using w2w version 2.0 in the presence of SOC.

After self-consistent calculations, these are the steps I am following for the wannierisation:

init_w2w (to write case.klist, case.inwf, case.win etc)

 x lapw1 -up -p
 x lapw1 -dn -p
 x lapwso -up -orb -p

x w2w -so -up -p  --> segmentation fault here
x w2w -so -dn -p  --> segmentation fault here
..........

Everything works perfectly if I use small kmesh (about 100 kpoints) for the wannierisation.
I have about 50 bands to wannierize.
When I increase the number of kpoints, w2w crashes giving segmentation fault.

I found that w2w can write case.mmn and case.eig with no problems but the seg fault occurred while writing case.amn.
I confirmed the following hoping that it could help in the debugging process:

1) If I decrease the number of bands to wannierize, the number of kpoints in the wannierisation can be slightly increased without seg fault.
2)The problem persists without -so and -p (i.e. without parallel calculation) flag.

In my opinion, there are some hard coded array size value(s) in the w2w program in amn/mmn subroutines which give segmentation fault when the number of projections*bands*kpoints exceeds that value.
This was confirmed by doing calculations in a simpler system like silicon with a large number of kpoints and bands.

I tried looking into the w2w source files and found few hard coded values.
However, I am hesitant to make changes because I am worried this could ruin other things.

This problem is similar to the one posted last year:
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg18896.html
which was not solved.
I would greatly appreciate your suggestion/feedback regarding this problem.
You could also guide me to the right subroutines if you know the details of the code.

Thank you.

Sincerely,
Niraj Aryal
Research Associate
Brookhaven National Lab.
Upton, NY