On 01/08/2014 06:49 PM, Oleg Rubel wrote:
I tried to compile the FORTRAN code you attached. It has the include statement
use util, only: line_count
use structmod, only: struct, struct_read
use const, only: BUFSZ
I was able to find 'util.f' in the w2w package, but not
On 01/08/2014 01:20 AM, Oleg Rubel wrote:
I wonder if anybody can reproduce this error?
I can. I used only MPI-parallelism:
$ cat .machines
1: localhost localhost
$ x lapw1 -up -p
w2k_dispatch_signal(): received: Segmentation fault
$ cat uplapw1.error
** Error in Parallel LAPW1
** LAPW1
Dear Elias,
thank you for confirming the error. Now I know that it is not specific
to my compilation. Gavin emailed me that the problem might be related
to MKL. I wonder whether what is your version of MKL? Is it older than
10.3?
I will recompile lapw1 with some checkpoints and hope to gain more
Hi,
print option for lapw1 was never meant to work with mpi.
Why would you want to print such huge Hamiltonians in ascii.
regards
Robert
On 08 January 2014 PM 11:54:50 Oleg Rubel wrote:
Dear Elias,
thank you for confirming the error. Now I know that it is not specific
to my compilation.
Hello Robert,
good to hear from you :)
print option for lapw1 was never meant to work with mpi.
Why would you want to print such huge Hamiltonians in ascii.
I need Fourier coefficients, which we use to unfold the band structure
of a supercell. So far we dealt with small cases and output1 was
Dear Oleg,
I attach a little program for converting the ‘vector’ file to plain
text, maybe this can help. This was just a quick hack (derived from
‘join_vectorfiles’), but it worked for me.
As for reading ‘vector’ in Python, are you using f2py for that? I guess
you could take (the read-in
Dear Elias,
thank you so much for the help.
I attach a little program for converting the ‘vector’ file to plain text,
maybe this can help. This was just a quick hack (derived from
‘join_vectorfiles’), but it worked for me.
I tried to compile the FORTRAN code you attached. It has the include
Dear Users and Developers,
I attempt to print WF coefficients using the option 'WFPRI' in
'case.in1(c)' file. It works well for a serial run, but fails in
MPI-parallel execution (x lapw1 -p). The program crashes with a
segmentation fault after eigenvalues for the first k-point are written
in
8 matches
Mail list logo
