NAN correction in gw.qp file

[ljzhou86]

I am doing a gw calculation on a 0D system containing 40 atoms using yambo 4.3.2 and "yambo -g n -p p -V par -r" to generate input file as

The input file as"
gw0 # [R GW] GoWo Quasiparticle energy levels
ppa # [R Xp] Plasmon Pole Approximation
rim_cut # [R RIM CUT] Coulomb potential
HF_and_locXC # [R XX] Hartree-Fock Self-energy and Vxc
em1d # [R Xd] Dynamical Inverse Dielectric Matrix
NLogCPUs=8 # [PARALLEL] Live-timing CPU`s (0 for all)
PAR_def_mode= "workload" # [PARALLEL] Default distribution mode ("balanced"/"memory"/"workload")
X_all_q_CPU= "1 1 9 8" # [PARALLEL] CPUs for each role
X_all_q_ROLEs= "q k c v" # [PARALLEL] CPUs roles (q,k,c,v)
X_all_q_nCPU_LinAlg_INV= 4 # [PARALLEL] CPUs for Linear Algebra
X_Threads=8 # [OPENMP/X] Number of threads for response functions
DIP_Threads=8 # [OPENMP/X] Number of threads for dipoles
SE_Threads=8 # [OPENMP/GW] Number of threads for self-energy
SE_CPU= "1 1 72" # [PARALLEL] CPUs for each role
SE_ROLEs= "q qp b" # [PARALLEL] CPUs roles (q,qp,b)
RandQpts= 2000000 # [RIM] Number of random q-points in the BZ
RandGvec= 123 RL # [RIM] Coulomb interaction RS components
CUTGeo= "box XYZ" # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere X/Y/Z/XY..
% CUTBox
45.50000 | 45.50000 | 45.50000 | # [CUT] [au] Box sides
%
CUTRadius= 0.000000 # [CUT] [au] Sphere/Cylinder radius
CUTCylLen= 0.000000 # [CUT] [au] Cylinder length
CUTwsGvec= 0.700000 # [CUT] WS cutoff: number of G to be modified
EXXRLvcs= 90737 RL # [XX] Exchange RL components
FFTGvecs= 90737
Chimod= "Hartree" # [X] IP/Hartree/ALDA/LRC/PF/BSfxc
% BndsRnXp
1 | 360 | # [Xp] Polarization function bands
%
NGsBlkXp= 4 Ry # RL # [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 | 400 | # [GW] G[W] bands range
%
GDamping= 0.10000 eV # [GW] G[W] damping
dScStep= 0.10000 eV # [GW] Energy step to evaluate Z factors
DysSolver= "n" # [GW] Dyson Equation solver ("n","s","g")
%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 1| 55|66|
%
""

The calculation finished normally, but I found the correction of qp gap is strange showing "NaN" in gw.qp file containg the GW correction.

#
# __ __ ________ ___ __ __ _______ ______
# /_/\/_/\ /_______/\ /__//_//_/\ /_______/\ /_____/\
# \ \ \ \ \\::: _ \ \\::\| \| \ \\::: _ \ \\:::_ \ \
# \:\_\ \ \\::(_) \ \\:. \ \\::(_) \/_\:\ \ \ \
# \::::_\/ \:: __ \ \\:.\-/\ \ \\:: _ \ \\:\ \ \ \
# \::\ \ \:.\ \ \ \\. \ \ \ \\::(_) \ \\:\_\ \ \
# \__\/ \__\/\__\/ \__\/ \__\/ \_______\/ \_____\/
#
#
# GPL Version 4.3.2 Revision 134. (Based on r.15658 h.afdb12
# MPI+OpenMP Build
# http://www.yambo-code.org
#
# GW [Newton Solver]
#
# PPA imaginary pt [ev]: 27.21138
# RIM G`s : 123
# RIM random pts : 2000000
# GW SC iterations :0
# dS/dw steps : 2
# dS/dw step [ev]: 0.10000
# X G`s [used]: 14363
# X G`s [disk]: 14363
# X bands : 1 360
# X poles [o/o]: 100.0000
# X e/h E range [ev]:-1.000000 -1.000000
# X xc-Kernel : none
# X BZ energy Double Grid: no
# X BZ Double Grid points:0
# Sc/G bands : 1 400
# Sc/G damping [ev]: 0.10000
# Sc bands terminator : no
# Sx RL components : 90879
# QP @ K 001 - 001 : b 055 - 066
#
# K-point Band Eo E-Eo Sc|Eo
#
1.00000 55.00000 -0.36408 NaN NaN
1.00000 56.00000 -0.20616 NaN NaN
1.00000 57.00000 -0.10570 NaN NaN
1.00000 58.00000 -0.09662 NaN NaN
1.00000 59.00000 -0.09638 NaN NaN
1.00000 60.00000 0.00000 NaN NaN
1.00000 61.00000 1.77565 NaN NaN
1.00000 62.00000 2.15941 NaN NaN
1.00000 63.00000 2.15947 NaN NaN
1.00000 64.00000 2.40576 NaN NaN
1.00000 65.00000 2.80214 NaN NaN
1.00000 66.00000 2.87236 NaN NaN
"
What 's the reason and how to fix it?? THanks a lot.
Best

[Daniele Varsano]

Dear Zhou Liu-Jiang,
hard to say what went wrong, can you please post the report file?
my first guess is that something went wrong with the plasmon-pole approximation.
Even if most probably not related to your problem, about the cutoff potential I would use a sphere instead of a cube.
Is that input file generated with the same version of yambo you are using? I find strange you have the

Code: Select all

 CUTwsGvec

variable defined, but the ws shape it is not listed in

Code: Select all

CUTGeo 

legend.
A suggestion is to download the latest version of the code and repeat your calculation (eventually also removing PAR_def_mode variable as you are assigning the cpu roles by hands) by generating the input file by scratch. If the problem persists please post your input files (also qe with pseudos) as we will need to reproduce the problem in order to spot what is going wrong.
Best,
Daniele

[ljzhou86]

Dear Daniele
[quote=If the problem persists please post your input files (also qe with pseudos) as we will need to reproduce the problem in order to spot what is going wrong.
[/quote]

I have tried these methods you suggested, however, the problem actually persists. I have attached the pseudos and input and output files of qe and yambo to you. Hope it can actually be fixed.

Thanks a lot.

[Daniele Varsano]

Dear Zhou ,
thanks for reporting, I have opened an issue and we will look at this shortly.
Best,
Daniele

[Davide Sangalli]

Dear Zhou,
thanks you for reporting.
Can you try to inspect your ndb.pp and ndb.pp_fragment_1 files with ncdump to check if they already contain some NaN ?

Best,
D.

[ljzhou86]

Dear Zhou,
thanks you for reporting.
Can you try to inspect your ndb.pp and ndb.pp_fragment_1 files with ncdump to check if they already contain some NaN ?

Best,
D.

I searched the "NAN" by using "ncdump ndb.pp | grep NaN " and "ncdump ndb.pp_fragment_1 | grep NaN ", and got nothing.

[ljzhou86]

Dear Zhou ,
thanks for reporting, I have opened an issue and we will look at this shortly.
Best,
Daniele

Dear Daniele,

Did you figure out this "NAN" issue?

[Daniele Varsano]

Dear Zhou,
unfortunately I still not have found the time to reproduce the problem and look into that.
I hope to find it very soon.
Daniele

[luonn]

Dear Daniele and Zhou,
I meet the same problem in a G0W0 calculation starting from HSE06. I get the SAVE fold from a scf calculation and the G0W0 calculation also finished normally. Besides, compared with the calculations from a PBE starting point, is there anything else should be taken care of when dealing with the input parameters? Thanks a lot!

[Daniele Varsano]

Dear Nannan Luo,
GW starting from HSE it is not allowed at the moment. There are different treatments of the divergence of the exchange term that make Yambo and QE not compatible resulting in a nonnegligible source of error. I have inserted barriers in p2y in order to prevent the use of HSE, but probably they have not yet percolated in the GPL release.
The allowed hybrid functionals at the moment is PBE0. We have recently also extended to the use of Gau-PBE but I still have to push in the GPL release.
When using PBE0 (or HF), anyway particular care of the divergence treatment have to be considered, ie to use the same technique wich are "vcut_ws" or "vcut_spherical" in exxdiv_treatment and the corresponding use of the coulomb cutoff in Yambo with the same geometry.
Best,
Daniele