Fortran runtime error when restarting phonon calculations with ph_mat.x

[AldiSaraiva]

I am using the code to perform Raman calculations with QE 7.3.1, and overall, it seems to work very well. However, when I try to restart a phonon calculation in a cluster with ph_mat.x, I encounter the following error just before the phonon calculation completes.

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x7f1dea023cda in ???
#1  0x7f1dea022d75 in ???
#2  0x7f1de9e4fcff in ???
#3  0x7f1dea2a7766 in ???
#4  0x7f1dea2a9171 in ???
#5  0x7f1dea29f526 in ???
#6  0x7f1dea29f955 in ???
#7  0x40c296 in elphfil_epc_
	at ~/qe-7.3.1-ReleasePack/qe-7.3.1/QERaman/src/supp.f90:163
#8  0x40b3f9 in do_phonon2_
	at ~/Codes/Packages/qe-7.3.1-ReleasePack/qe-7.3.1/QERaman/src/do_phonon2.f90:153
#9  0x403b81 in phonon2
	at ~/Codes/Packages/qe-7.3.1-ReleasePack/qe-7.3.1/QERaman/src/phonon2.f90:78
#10  0x403a3c in main
	at ~/Codes/Packages/qe-7.3.1-ReleasePack/qe-7.3.1/QERaman/src/phonon2.f90:43

Is restarting phonon calculations with ph_mat.x officially supported, or is there a plan to implement this feature in future releases?

[AldiSaraiva]

After testing different configurations, I believe I have found the source of the issue where the array wq was not properly allocated when restarting the ph_mat.x calculation. This problem occurred in the file supp.f90.

To resolve it, I added the following line of code in supp.f90 at line 84:
ALLOCATE(wq(nqs))
Here's a snippet of where it was added:

INTEGER :: ftau(3,48)
INTEGER, EXTERNAL :: find_free_unit, atomic_number
ALLOCATE(wq(nqs))

filelph = TRIM(prefix) // '.elph'

This change allowed the code to continue without the segmentation fault error. While I'm not certain if this would be considered an "official" fix, it works as a workaround and might help others encountering a similar issue during restart.

[nguyen-group]

Hi @AldiSaraiva

I really appreciate your efforts to point out the issue and bug our code.
I did not realize this restart issue since I never restarted the program. But it is an important issue, and your fixing could help. I will work on this issue and make a patch.

Best,
Nguyen

[AldiSaraiva]

Hi Nguyen,

Thank you for your kind words and for acknowledging the issue. I'm glad that my input could be helpful, and I truly appreciate your efforts to look into it. Also, thank you for sharing your code with the community.

Best regards,

Aldi