problem with -M option with spinorial calculations

Questions and doubts about the spin-dependent version of Yambo.

Moderators: Davide Sangalli, andrea marini, Conor Hogan, myrta gruning

problem with -M option with spinorial calculations

Postby hlee » Mon Dec 09, 2013 9:34 pm

Dear Developers:

I am starting to use Yambo (GPL Version 3.4.0 Revision 2132) for quasi-particle calculations with spin-orbit coupling (SOC).
First, I calculated the GW corrections for bulk Si with SOC following the following steps:
(1) scf calculation with Quantum ESPRESSO (QE) v5.0.2 (+patches) using the following input
Code: Select all
&control
    calculation='scf',
    restart_mode='from_scratch',
    prefix='Si',
    outdir='./'
    pseudo_dir= './',
!    wf_collect=.true.,
    verbosity ='high',
/&end
 &system
    ibrav=2,
    celldm(1)=10.183,
    nat=2,
    ntyp=1,
    ecutwfc=25.0,
    nbnd=20,
    force_symmorphic= .true.,
    noncolin=.true.
    lspinorb=.true.
/&end
 &electrons
    conv_thr = 1.0d-8
/&end
ATOMIC_SPECIES
 Si  28.086  Si.rel-pbe-rrkj.UPF
ATOMIC_POSITIONS
 Si 0.00 0.00 0.00
 Si 0.25 0.25 0.25
K_POINTS (automatic)
 4 4 4 1 1 1

(2) nscf calculation with QE using the following input
Code: Select all
&control
    calculation='bands',
    restart_mode='from_scratch',
    prefix='Si',
    outdir='./'
    pseudo_dir= './',
    wf_collect=.true.,
    verbosity ='high',
/&end
 &system
    ibrav=2,
    celldm(1)=10.183,
    nat=2,
    ntyp=1,
    ecutwfc=25.0,
    nbnd=100,
    force_symmorphic= .true.,
    noncolin=.true.
    lspinorb=.true.
/&end
 &electrons
    conv_thr = 1.0d-8
    diago_full_acc=.true.,
/&end
ATOMIC_SPECIES
 Si  28.086  Si.rel-pbe-rrkj.UPF
ATOMIC_POSITIONS
 Si 0.00 0.00 0.00
 Si 0.25 0.25 0.25
K_POINTS (automatic)
 6 6 6 0 0 0

(3) creation of core databases using the following command
Code: Select all
p2y -N -S

(4) initialization for Yambo using the following command
Code: Select all
yambo -N

(5) generation of the input file for Yambo
Code: Select all
yambo -d -k hartree -g n -p p

For testing purposes, I changed the input file slightly as follows:
Code: Select all
#  ____  ____     _       ____    ____  ______      ___
# |_  _||_  _|   / \     |_   \  /   _||_   _ \   ."   `.
#   \ \  / /    / _ \      |   \/   |    | |_) | /  .-.  \
#    \ \/ /    / ___ \     | |\  /| |    |  __". | |   | |
#    _|  |_  _/ /   \ \_  _| |_\/_| |_  _| |__) |\  `-"  /
#   |______||____| |____||_____||_____||_______/  `.___."
#
#             GPL Version 3.4.0 Revision 2132
#                http://www.yambo-code.org
#
em1d                         # [R Xd] Dynamical Inverse Dielectric Matrix
gw0                          # [R GW] GoWo Quasiparticle energy levels
ppa                          # [R Xp] Plasmon Pole Approximation
HF_and_locXC                 # [R XX] Hartree-Fock Self-energy and Vxc
EXXRLvcs=  10          Ry    # [XX] Exchange RL components
Chimod= "Hartree"            # [X] IP/Hartree/ALDA/LRC/BSfxc
% QpntsRXp
  1 | 16 |                   # [Xp] Transferred momenta
%
% BndsRnXp
   1 | 100 |                 # [Xp] Polarization function bands
%
NGsBlkXp= 1            Ry    # [Xp] Response block size
% LongDrXp
 1.000000 | 0.000000 | 0.000000 |        # [Xp] [cc] Electric Field
%
PPAPntXp= 27.21138     eV    # [Xp] PPA imaginary energy
% GbndRnge
   1 | 100 |                 # [GW] G[W] bands range
%
GDamping= 0.100000     eV    # [GW] G[W] damping
dScStep= 0.100000      eV    # [GW] Energy step to evalute Z factors
DysSolver= "n"               # [GW] Dyson Equation solver (`n`,`s`,`g`)
%QPkrange                    # [GW] QP generalized Kpoint/Band indices
  1| 16|  1| 20|
%
%QPerange                    # [GW] QP generalized Kpoint/Energy indices
  1| 16| 0.0|-1.0|
%

(6) test run using 8 cores

When running without -M option, calculation ended successfully.
However, when using -M option, calculation stopped resulting the following outputs:
In r-* file,
Code: Select all
...
 [04] Bare local and non-local Exchange-Correlation
 ==================================================

 [EXS] Plane waves : 169

 QP @ K 001 - 016 : b 001 - 020

 [Distribute] Average allocated memory is [o/o]: 100.0000
 [FFT-HF/Rho] Mesh size: 14  14  14

 [WF loader] Normalization (few states)  min/max  :0.812E-16  1.00

 [xc] Functional Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
 [xc] LIBXC used to calculate xc functional

 [WARNING]Not enough states to calculate rho

 XC HF and DFT [eV] @ K [1] (iku): 0.00      0.00      0.00
 <1|HF|1> =       NaN       NaN <1|DFT|1> = -10.27369 -.472E-16
 <2|HF|2> =       NaN       NaN <2|DFT|2> = -10.27369 0.330E-15
 <3|HF|3> =       NaN       NaN <3|DFT|3> = -11.35864 0.472E-16
 <4|HF|4> =       NaN       NaN <4|DFT|4> =       NaN       NaN
 <5|HF|5> =       NaN       NaN <5|DFT|5> = -11.35687 -.472E-16
 <6|HF|6> =       NaN       NaN <6|DFT|6> =       NaN       NaN
 <7|HF|7> =       NaN       NaN <7|DFT|7> =       NaN       NaN
 <8|HF|8> =       NaN       NaN <8|DFT|8> = -11.35687 0.944E-16
 <9|HF|9> =       NaN       NaN <9|DFT|9> = -9.818784 -.177E-16
 <10|HF|10> =       NaN       NaN <10|DFT|10> = -9.818784 -.148E-15
 <11|HF|11> =       NaN       NaN <11|DFT|11> = -9.813917  0.000000
 <12|HF|12> =       NaN       NaN <12|DFT|12> =       NaN       NaN
 <13|HF|13> =       NaN       NaN <13|DFT|13> =       NaN       NaN
 <14|HF|14> =       NaN       NaN <14|DFT|14> =       NaN       NaN
 <15|HF|15> =       NaN       NaN <15|DFT|15> =       NaN       NaN
 <16|HF|16> =       NaN       NaN <16|DFT|16> = -10.54242  0.000000
 <17|HF|17> =       NaN       NaN <17|DFT|17> = -9.879955  0.000000
 <18|HF|18> =       NaN       NaN <18|DFT|18> =       NaN       NaN
 <19|HF|19> =       NaN       NaN <19|DFT|19> =       NaN       NaN
 <20|HF|20> =       NaN       NaN <20|DFT|20> = -9.879955  0.000000
...
 [05] Dynamic Dielectric Matrix (PPA)
 ====================================

 [Distribute] Average allocated memory is [o/o]: 19.50000

 [ERROR] STOP signal received while in :[05] Dynamic Dielectric Matrix (PPA)

 [ERROR] Error in parallel wf distribution


As shown above, I strongly suspect the warning message, "[WARNING]Not enough states to calculate rho".
When calculating without SOC, there is no problem both with -M option and without -M option.
Dr. Hyungjun Lee
Institute of Theoretical Physics, EPFL
hlee
 
Posts: 29
Joined: Mon Jul 15, 2013 2:09 pm

Re: problem with -M option with spinorial calculations

Postby Davide Sangalli » Tue Dec 17, 2013 9:39 pm

Dear Dr. Hyungjun Lee,
you are right.

Unfortunately the -M option was never tested with n_spinor=2.
Try to replace the file WF_states_setup.F in src/wf with the one attached and re-compile yambo.

Let me know if it works. In case I'll submit the bug-fix to the distributed version

Best regards,
Davide
You do not have the required permissions to view the files attached to this post.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
http://www.ism.cnr.it/en/davide-sangalli-cv/
http://www.max-centre.eu/
User avatar
Davide Sangalli
 
Posts: 315
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy

Re: problem with -M option with spinorial calculations

Postby hlee » Wed Dec 18, 2013 3:28 pm

Dear Dr. Davide Sangalli:

Thank you very much for fixing this bug.
It works!

Sincerely,
Dr. Hyungjun Lee
Institute of Theoretical Physics, EPFL
hlee
 
Posts: 29
Joined: Mon Jul 15, 2013 2:09 pm


Return to Spin-dependent and spinorial calculations

Who is online

Users browsing this forum: No registered users and 1 guest

cron