[Pw_forum] Problem with DFT + U calculations
Lai Jiang
jianglai at sas.upenn.edu
Mon Nov 14 05:31:49 CET 2011
Greetings all,
I'm having some problem getting convergence in my DFT +U calculations. My
system is anti-ferromagnetic double perovskite Sr2FeWO6. It is
experimentally an insulator but there are reports that state one needs
Hubbard U on Fe 3d orbital to account for the on site repulsion. However
after adding an U value it doesn't converge after 500 iterations and
sometimes the job got killed by the batch system (not walltime, something
like buffer overflow), typical pbs output looks like this (varies
on different platform):
Rank 16 [Wed Nov 9 23:21:11 2011] [c0-0c0s4n1] Fatal error in PMPI_Bcast:
Message truncated, error stack:
PMPI_Bcast(1308)......................: MPI_Bcast(buf=0x3e0def0,
count=26208, MPI_DOUBLE_PRECISION, root=1, comm=0xc4000000) failed
MPIR_Bcast(998).......................:
MPIR_Bcast_scatter_ring_allgather(842):
MPIR_Bcast_binomial(157)..............:
MPID_nem_gni_lmt_start_recv(920)......: Message from rank 0 and tag 2
truncated; 213408 bytes received but buffer size is 209664
[0] ERROR - nem_gni_error_handler(): a transaction error was detected,error
category 0x4 error code 0xb2e
Rank 0 [Wed Nov 9 23:21:11 2011] [c0-0c0s4n0] GNI transaction error
detected
[NID 00009] 2011-11-09 23:21:11 Apid 1944384: initiated application
termination
And the pw.x output ends like this:
atom 24 Tr[ns(na)]= 6.1782057
atom 24 spin 1
eigenvalues: 0.9740685 0.9841342 0.9842280 0.9899863 0.9914280
eigenvectors
1 0.0332876 0.0801355 -0.2854709 0.9362259 -0.1856276
2 0.0172615 0.0589306 0.8879786 0.1822775 -0.4177297
3 0.4670136 0.8765271 -0.0107306 -0.0778135 0.0861883
4 -0.0557289 -0.0272157 0.3597350 0.2894423 0.8848545
5 0.8816953 -0.4701756 0.0218145 0.0205958 0.0254629
occupations
0.990 -0.003 0.000 0.000 0.000
-0.003 0.986 0.000 -0.001 0.000
0.000 0.000 0.984 0.003 0.001
0.000 -0.001 0.003 0.976 0.003
0.000 0.000 0.001 0.003 0.988
atom 24 spin 2
eigenvalues: 0.0394193 0.0400804 0.1166033 0.1376013 0.9206564
eigenvectors
1 -0.0457703 -0.2123040 -0.9448844 -0.2058039 0.1329300
2 0.1857246 0.9542619 -0.1868327 -0.1235578 0.0686851
3 -0.0675780 -0.0156072 0.1108752 0.1546050 0.9792822
4 -0.9788987 0.1940608 -0.0050595 -0.0004723 -0.0638113
5 -0.0245350 -0.0800506 0.2448840 -0.9583712 0.1206087
occupations
0.134 -0.017 -0.005 0.020 -0.002
-0.017 0.049 -0.018 0.067 -0.011
-0.005 -0.018 0.093 -0.205 0.034
0.020 0.067 -0.205 0.851 -0.090
-0.002 -0.011 0.034 -0.090 0.127
nsum = 49.3436516
exit write_ns
total cpu time spent up to now is 4234.0 secs
total energy = -5515.60273544 Ry
Harris-Foulkes estimate = -5515.60272792 Ry
estimated scf accuracy < 0.00001279 Ry
total magnetization = 0.00 Bohr mag/cell
absolute magnetization = 32.50 Bohr mag/cell
iteration #102 ecut= 50.00 Ry beta=0.20
Davidson diagonalization with overlap
Application 1944384 exit codes: 1
Application 1944384 exit signals: Killed
Application 1944384 resources: utime ~4162s, stime ~88s
My input file is:
&control
calculation = 'scf'
title = 'Sr2FeWO6'
verbosity = 'high'
restart_mode = 'from_scratch'
pseudo_dir = '/workdir/jianglai/psp/gga'
etot_conv_thr = 1.D-6
forc_conv_thr = 2.D-4
disk_io = 'low'
tstress = .true.
tprnfor = .true.
wf_collect = .true.
nstep = 5000
/
&system
ibrav = -12
A = 11.301600
B = 5.613600
C = 15.885000
cosAC = -0.000366588948
nat = 80
ntyp = 5
nbnd = 351
ecutwfc = 50
occupations = 'smearing'
degauss = 2.D-3
nspin = 2
starting_magnetization(2) =0.5
starting_magnetization(3) =-0.5
lda_plus_u = .true.
hubbard_U(2) = 4
hubbard_U(3) = 4
/
&electrons
electron_maxstep = 500
startingwfc = 'random'
diagonalization = 'david'
mixing_mode = 'plain'
mixing_beta = 2.D-1
mixing_ndim = 8
conv_thr = 1.0D-8
/
&IONS
bfgs_ndim = 4
upscale = 25
pot_extrapolation = 'second_order'
wfc_extrapolation = 'second_order'
/
&CELL
cell_dynamics = 'bfgs'
press_conv_thr = 5.D-1
/
ATOMIC_SPECIES
Sr 87.62 Sr.upf
Fe1 55.847 Fe.semi.upf
Fe2 55.847 Fe.semi.upf
W 183.85 W.fhi.UPF
O 15.999 O.upf
ATOMIC_POSITIONS {crystal}
Sr 0.499500 0.012900 0.124550
Sr 0.500500 0.987100 0.875450
Sr 0.750500 0.512900 0.125450
Sr 0.749500 0.487100 0.374550
Sr 0.499500 0.012900 0.624550
Sr 0.500500 0.987100 0.375450
Sr 0.750500 0.512900 0.625450
Sr 0.749500 0.487100 0.874550
Sr 0.999500 0.012900 0.124550
Sr 0.000500 0.987100 0.875450
Sr 0.250500 0.512900 0.125450
Sr 0.249500 0.487100 0.374550
Sr 0.250500 0.512900 0.625450
Sr 0.249500 0.487100 0.874550
Fe1 0.250000 -0.000000 0.250000
Fe1 0.750000 -0.000000 0.750000
Fe2 0.000000 0.500000 0.000000
Fe2 0.500000 0.500000 0.500000
Fe2 0.250000 -0.000000 0.750000
Fe2 0.750000 -0.000000 0.250000
Fe1 -0.000000 0.500000 0.500000
Fe1 0.500000 0.500000 0.000000
W 0.250000 0.000000 0.000000
W 0.750000 0.000000 0.000000
W -0.000000 0.500000 0.250000
W 0.500000 0.500000 0.250000
W 0.250000 -0.000000 0.500000
W 0.750000 -0.000000 0.500000
W 0.000000 0.500000 0.750000
W 0.500000 0.500000 0.750000
O 0.023650 0.496000 0.129500
O 0.976350 0.504000 0.870500
O 0.226350 0.996000 0.120500
O 0.273650 0.004000 0.379500
O 0.023650 0.496000 0.629500
O 0.976350 0.504000 0.370500
O 0.226350 0.996000 0.620500
O 0.273650 0.004000 0.879500
O 0.523650 0.496000 0.129500
O 0.476350 0.504000 0.870500
O 0.726350 0.996000 0.120500
O 0.773650 0.004000 0.379500
O 0.523650 0.496000 0.629500
O 0.476350 0.504000 0.370500
O 0.726350 0.996000 0.620500
O 0.773650 0.004000 0.879500
O 0.359500 0.261000 0.012500
O 0.640500 0.739000 0.987500
O 0.890500 0.761000 0.237500
O 0.609500 0.239000 0.262500
O 0.359500 0.261000 0.512500
O 0.640500 0.739000 0.487500
O 0.890500 0.761000 0.737500
O 0.609500 0.239000 0.762500
O 0.859500 0.261000 0.012500
O 0.140500 0.739000 0.987500
O 0.390500 0.761000 0.237500
O 0.109500 0.239000 0.262500
O 0.859500 0.261000 0.512500
O 0.140500 0.739000 0.487500
O 0.390500 0.761000 0.737500
O 0.109500 0.239000 0.762500
O 0.121000 0.224000 0.488000
O 0.879000 0.776000 0.512000
O 0.129000 0.724000 0.762000
O 0.371000 0.276000 0.738000
O 0.121000 0.224000 0.988000
O 0.879000 0.776000 0.012000
O 0.129000 0.724000 0.262000
O 0.371000 0.276000 0.238000
O 0.621000 0.224000 0.488000
O 0.379000 0.776000 0.512000
O 0.629000 0.724000 0.762000
O 0.871000 0.276000 0.738000
O 0.621000 0.224000 0.988000
O 0.379000 0.776000 0.012000
O 0.629000 0.724000 0.262000
O 0.871000 0.276000 0.238000
K_POINTS (automatic)
2 4 1 1 1 1
I checked quantum espresso example files 25 and in README it says one needs
to set starting_ns_eigenvalues sometimes. I'm not sure whether it's the
case here but I'm gonna try. Still I don't quite understand the eigenvalues
and eigenvectors in write_ns routine. Are eigenvalues the energy of each of
the five d orbitals? The what are the eigenvectors?
That all said, if someone has experienced problems like this or knows why
convergence is so slow here, please let me know.
Thanks!
Lai Jiang
*********************************
Department of Chemistry
School of Arts and Sciences
University of Pennsylvania
*********************************
231 South 34th Street, Box 46
Philadelphia, PA 19104
Tel: 215-573-4241
Fax: 215-573-2112
Email: jianglai at sas.upenn.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20111113/8c441ae5/attachment.html>
More information about the users
mailing list