ABINIT Discussion Forums

Hi all,
I posted this on the message board under a different topic, but haven't received any replies so I thought I'd try here as well.

I am attempting to run anaddb (interested in raman) for calcite with a rhombohedral primitive cell. I have been getting segmentation faults. They are not always in the same place in the code, but they all seem to boil down to "syntax errors" in the fortran files (which I know cannot be right because the anaddb fortran files all run fine when I use tutorial input files). I can't tell where the error is coming from. Everything seems to run fine: abinit, mrgddb until I get to anaddb and then I get this weird segmentation fault. There are some warnings in abinit, which I know about because I am ignoring some of the necessary convergence studies as at this stage I just want to see how close I can come and then I will refine everything later. Is it possible I am running out of memory?

The error looks like this when I run anaddb in gdb:
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7b85e08 in _gfortran_reshape_r8 () from /usr/lib/libgfortran.so.3

The output file contains nothing.
The log file contains nothing.

I am running Abinit-6.8.1 on a 64-bit Dell precision T7500 linux workstation running Ubuntu 10.10.

I have attached all the relevant input files.

I am using the following pseudopotentials:
20ca.2.hgh
6c.pspnc
8o.pspnc
from the Psps_for_tests directory.

Every time I try to solve this I feel like I've hit a wall. Any help would be greatly appreciated.

Many thanks,
loniarf

Dear loniarf,

I analyzed the ddb file and received no segmentation fault, but there are lots of "NaN" in the output file. You could run the test packages to make sure if there is any problem with your compilation.

Sincerely,
Guangfu Luo

Guangfu Luo,
Thank you for your reply. I am relieved to see that the problem does not appear to be my input parameters.
I have run the basic tests (tests_in, tests_libxc, tests_fast) as well as the response function tests (tests_v2).
There are also do not appear to be errors during the use of configure, but I have posted here the results of configure as a .txt file.
Please let me know if you see anything strange.
Thanks again,
Lonia

Dear Guangfu Luo,
It seems to me (after successful re-compilation and tests) that the problem may be with my computing system.
What are the specs of the system that you used to successfully run my input?
If I can't get my own system running, would it be possible for you to send me the output files that your run generated?
Thank you once again,
Loniarf

Dear Loniarf,

I use Intel Ifort-9.1.052 and Intel mpi 4.0.0.028 to compile ABINIT, and they worked pretty well for the abinit versions I used. The following line is my configure setting for abinit-6.8.1.

./configure FC=/opt/intel/impi/4.0.0.028/bin64/mpiifort enable_mpi=yes --enable-mpi-io=no --enable-64bit-flags=yes --with-trio-flavor=none

As you requested, I attached the anaddb analysis results of your files, and a successful Raman spectrum analysis of my own for your check.

Sincerely,
Guangfu Luo