GW, Bethe-Salpeter …
Moderators: maryam.azizi, bruneval
-
Zhang Xiaole
- Posts: 46
- Joined: Thu Sep 15, 2011 7:13 am
Post
by Zhang Xiaole » Wed Mar 26, 2014 2:39 am
Dear all,
I used abinit-7.6.2 to calculate the energy gap with hybrid functional. Everything went well in the test files(t41.in), however, when i changed the input file for my system which is a hexagnal structure the code would complain a bug like following:
Code: Select all
--- !BUG
message: |
iloop,igb,mgb= 2 77 76
about to overwrite gbound.
src_file: sphereboundary.F90
src_line: 251
...
leave_new : decision taken to exit ...
calling XMPI_ABORT...
MPI_ERROR_STRING: MPI_ERR_UNKNOWN: unknown error
i don't known why i can't append my input and log files.
Thanks in advanced.
sheleon
Phd student
Phys Department
Shanghai JiaoTong university, Shanghai, China
-
Zhang Xiaole
- Posts: 46
- Joined: Thu Sep 15, 2011 7:13 am
Post
by Zhang Xiaole » Wed Mar 26, 2014 2:38 pm
these are input and log files
-
Attachments
-
bamgf.in- input file
- (3.61 KiB) Downloaded 367 times
-
- log.out
- log file
- (77.78 KiB) Downloaded 326 times
Phd student
Phys Department
Shanghai JiaoTong university, Shanghai, China
-
martinspenke
- Posts: 13
- Joined: Wed Feb 12, 2014 2:26 pm
Post
by martinspenke » Thu Mar 27, 2014 1:13 am
I think you miss the screening part in your input file if you wanna use gw feature !
Bests
-
Zhang Xiaole
- Posts: 46
- Joined: Thu Sep 15, 2011 7:13 am
Post
by Zhang Xiaole » Thu Mar 27, 2014 2:20 am
martinspenke wrote:I think you miss the screening part in your input file if you wanna use gw feature !
Bests
Thanks for your reply.
Since it is a hybrid functional calculation, just part of the GW calculation, i don't thing there is something missing.
sheleon
Phd student
Phys Department
Shanghai JiaoTong university, Shanghai, China
-
martinspenke
- Posts: 13
- Joined: Wed Feb 12, 2014 2:26 pm
Post
by martinspenke » Thu Mar 27, 2014 7:08 am
I tried your input files and got the same error.
I guess this is one of those compiler dependent errors.
So try to compile abinit with gfortran and rerun the calculation.
Let me know of it .
Bests
-
Zhang Xiaole
- Posts: 46
- Joined: Thu Sep 15, 2011 7:13 am
Post
by Zhang Xiaole » Thu Mar 27, 2014 12:24 pm
martinspenke wrote:I tried your input files and got the same error.
I guess this is one of those compiler dependent errors.
So try to compile abinit with gfortran and rerun the calculation.
Let me know of it .
Bests
Unfortunately, changing compiler dos not make any difference.
Phd student
Phys Department
Shanghai JiaoTong university, Shanghai, China
-
martinspenke
- Posts: 13
- Joined: Wed Feb 12, 2014 2:26 pm
Post
by martinspenke » Thu Mar 27, 2014 5:56 pm
Okay then the bug must be removed by developers. The bug is due to FFT grid.
So any way if you wanna calculate the gap energy of your system, switch to an all electron code ELK, for instance.
http://elk.sourceforge.net/Cheers
-
Zhang Xiaole
- Posts: 46
- Joined: Thu Sep 15, 2011 7:13 am
Post
by Zhang Xiaole » Fri Mar 28, 2014 6:05 am
martinspenke wrote:Okay then the bug must be removed by developers. The bug is due to FFT grid.
So any way if you wanna calculate the gap energy of your system, switch to an all electron code ELK, for instance.
http://elk.sourceforge.net/Cheers
Thanks.
sheleon
Phd student
Phys Department
Shanghai JiaoTong university, Shanghai, China