The ABINIT Users Mailing List ( CLOSED )

[Takeshi Nishimatsu <takeshi@physics.rutgers.edu>]

Dear Xavier and all,

Xavier Gonze wrote:
> For the DOS with the tetrahedron method,
> one should avoid having tetrahedra that have corners with the same 
> energies, as mentioned by Matthieu.
> A shifted (shiftk 1/2 1/2 1/2) grid will usually present such a problem.
> So, you should try a non shifted grid (shiftk 0 0 0) for the DOS.
> Moreover, if the band gap happens at the Gamma point, you will have it 
> correctly reproduced in your DOS.

Thank you for your advice. With the attached input file, I got a less peaky 
result.
I set "ngkpt  8 8 8" (even numbers) to get some k-points on zone boundaries.

Caculation of the attached input file goes fine with abinis or single-process 
abinip.
But I got SIGSEGV with multi-process abinip at DATASET 3. I could not figure 
out what
is the problem. My environment is Linux, gfortran-4.2.2 (FCFLAGS=-O3), and 
mpich2-1.0.6.
The abinip compiled with ifort 10.0.025 has same problem, too.
Does someone has the same problem or not?

PATCH: It seems that gfortran does not like meaningless ";".
--- abinit-5.4.3/src/13io_mpi/hdr_comm.F90~     2007-08-30 16:04:06.000000000 
-0400
+++ abinit-5.4.3/src/13io_mpi/hdr_comm.F90      2007-10-12 11:42:28.000000000 
-0400
@@ -186,7 +186,7 @@
  hdr%so_psp =list_int(1+index:npsp   +index) ; index=index+npsp
  hdr%symafm=list_int(1+index:nsym       +index) ; index=index+nsym
  hdr%symrel=reshape(list_int(1+index:nsym*3*3   +index),(/3,3,nsym/))
-                                                 ; index=index+nsym*3*3
+                                                  index=index+nsym*3*3
  hdr%typat  =list_int(1+index:natom      +index) ; index=index+natom
 end if
 deallocate(list_int)



=== SrTiO3.in 
==================================================================
### Definition of the k-point grid
kptopt 1
nshiftk 1
shiftk  0 0 0
ngkpt  8 8 8

ndtset 3
#DATASET 1
tolvrs1  1.0d-20
tolwfr1  0.0d0
  iscf1  7
prtden1  1

#DATASET 2
tolvrs2  0.0d0
tolwfr2  1.0d-18
getden2  1
getwfk2  1
prtden2  0
 prtwf2  0
  iscf2  -2
prtdos2  2

#DATASET 3
tolvrs3  0.0d0
tolwfr3  1.0d-18
getden3  1
getwfk3  1
prtden3  0
 prtwf3  0
  iscf3  -2
prtdos3  3
ratsph3  2.7212056200 1.1432843056 2.6456165750 2.6456165750 2.6456165750 

##### CONVERGENCE PARAMETERS
ecut 20.0

acell 7.26 7.26 7.26

rprim  1     0     0
      0     1     0
      0     0     1

#Definition of the atom types
ntypat 3
znucl 38 22 8

#Definition of the atoms
natom 5
typat 1*1 1*2 3*3
xred
 0.0 0.0 0.0 
 0.5 0.5 0.5 
 0.0 0.5 0.5 
 0.5 0.0 0.5
 0.5 0.5 0.0


### Definition of the SCF procedure
nband 35
nstep 200
diemac 12.0
occopt 7 tsmear 5.0d-3
pawecutdg  30.0


=== files 
======================================================================
SrTiO3.in
SrTiO3.out
SrTiO3._i
SrTiO3._o
SrTiO3.
sr_ps.abinit.paw
ti_ps.abinit.paw
o_ps.abinit.paw



=== dos.gp 
=====================================================================
#!/usr/bin/env gnuplot
##
set key left

set title "`pwd | sed 's%^.*/%%'`" font 'Times-Roman,18'

set terminal postscript landscape enhanced color 'Times-Roman' 22
set output 'dos.eps'

plot [-2:1] [0:400] \
    '< tail --line=+15 SrTiO3._o_DS2_DOS'        using 2:3 title 'total' w l,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0001' using 2:3 title 'Sr 4s' w l,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0001' using 2:4 title 'Sr 3p' w l,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0002' using 2:3 title 'Ti 3s' w l,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0002' using 2:4 title 'Ti 2p' w l,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0002' using 2:5 title 'Ti 3d' w l 
lt 9,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0003' using 2:($3*3) title ' O 2s' 
w l lt 7,\
    '< tail --line=+20 SrTiO3._o_DS3_DOS_AT0003' using 2:($4*3) title ' O 2p' 
w l lt 8


Sincerely,
--
Takeshi Nishimatsu
http://loto.sourceforge.net/feram/   Fast MD program for perovskite-type 
ferroelectrics