My Community

Posted: **Wed Mar 15, 2023 11:29 am**

Dear developers,

Thank you for uploading the patch file.
I also encountered a similar error, so I applied the patch file to the source code of vasp.6.4.0 and recompiled it.
However, there still seems to be a problem with my ML_FFN file.

Could you please check if the broken part of my ML_FFN file is the same as that of this thread?

I used makefile.include.intel_omp to compile vasp.6.4.0 and ran the job using hybrid MPI (MPI+OpneMP).

The attached folder contains a ML_FFN file generated by ML_ISTART=0 and an error message that occurred by ML_ISTART=2.

This is the error message.

Code: Select all

forrtl: Invalid argument
forrtl: severe (39): error during read, unit 503, file /path/to/my/working/dir/ML_FF
Image              PC                Routine            Line        Source             
vasp_std           0000000001D3965E  for__io_return        Unknown  Unknown
vasp_std           0000000001D6A112  for_read_seq          Unknown  Unknown
vasp_std           00000000005F4EE2  Unknown               Unknown  Unknown
vasp_std           00000000005F4B86  Unknown               Unknown  Unknown
vasp_std           000000000065D8C7  Unknown               Unknown  Unknown
vasp_std           0000000000724278  Unknown               Unknown  Unknown
vasp_std           000000000072ECAA  Unknown               Unknown  Unknown
vasp_std           00000000010DE375  Unknown               Unknown  Unknown
vasp_std           0000000001BA962E  Unknown               Unknown  Unknown
vasp_std           0000000000418F9E  Unknown               Unknown  Unknown
libc-2.17.so       00002B49506443D5  __libc_start_main     Unknown  Unknown
vasp_std           0000000000418EA9  Unknown               Unknown  Unknown
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed

  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
100   185    0    29  100   156    315   1695 --:--:-- --:--:-- --:--:--  2010
{"status":200,"message":"ok"}

I also tried ML_MODE=refit according to the instructions in the URL below but encountered the same error.
URL: https://www.vasp.at/wiki/index.php/Known_issues

Best regards,
Kai

Posted: **Wed Mar 15, 2023 4:42 pm**

Hello Kai!

I just checked your ML_FFN file and everything seems ok, there are no empty sections which are symptoms of the bug reported in this forum topic. Also, I was able to run a short ML_ISTART=2 trajectory with your input files without any problems. I tried the GNU compiler and two versions of the Intel compiler. From the run script you provided I could identify that you are using the Intel compiler version 2018.5.274 on your machine. Do you maybe have a newer version available too which you could test? I did use Intel's oneAPI 22.1.2 and 23.0.0.

Since this seems to be a separate issue, could you please open a new topic with your answer, I will then move our posts to this topic. Thank you!

All the best,
Andreas Singraber

Posted: **Fri Mar 17, 2023 5:23 am**

Dear Andreas Singraber,

Thank you for checking my ML_FFN file.

I have opened a new topic about the comment I posted on this thread.

I will test another version of compilers and report to you if there is any progress on that.

Best regards,
Kai

Posted: **Mon Apr 10, 2023 12:29 pm**

Dear developers,

Thank you for the advice.

I confirmed that the Fortran error didn't occur by compiling with intel oneAPI 23.0.0.

However, there were still some errors in the output of "make test".

==================================================================
SUMMARY:
==================================================================
The following tests failed, please check the output file manually:
bulk_BN_PBE_Libxc bulk_BN_SCAN_Libxc NiO_2E4W

I understand that the "Libxc" job failed because the -DUSELIBXC option was not used when compiling VASP.6.4.0.
However, I don't know how to fix a problem of NiO_2E4W.
An error message of NiO_2E4W is probably due to MPI but other jobs finished successfully.

I ran the test calculation twice and both times encountered the error.

I set -DVASP2WANNIER90 to compile option and used wannier90-2.1.0.

Do you know anything about the error?

Best regards,
Kai

My Community

forrtl: severe (39): error during read, unit 503, file /path/to/working/dir/ML_FF

Re: VASP 6.4 ML_ISTART=2 doesn't work

Re: VASP 6.4 ML_ISTART=2 doesn't work

forrtl: severe (39): error during read, unit 503, file /path/to/working/dir/ML_FF

Re: forrtl: severe (39): error during read, unit 503, file /path/to/working/dir/ML_FF