GW convergence

Please post here any issue related to convergence parameters in linear response GW and BSE calculations.

Moderators: Davide Sangalli, Daniele Varsano

haseebphysics1
Posts: 169
Joined: Sat Aug 17, 2019 2:48 pm

Re: GW convergence

Post by haseebphysics1 » Wed Aug 04, 2021 2:53 pm

Dear Dr. Daniele,

We have the following variables (I think, only in Yambo 5.0+). Can you please guide me, can we use them to replace the procedure mentioned here (http://www.yambo-code.org/wiki/index.ph ... alues_only)

Code: Select all

GWoIter=0                        # [GW] GWo self-consistent (evGWo) iterations on eigenvalues
GWIter=0                         # [GW] GW  self-consistent (evGW)  iterations on eigenvalues
Is it just providing the number of corresponding iterations (positive integer value)?

Thanks in advance,
Haseeb Ahmad
MS - Physics,
LUMS - Pakistan

User avatar
Daniele Varsano
Posts: 2992
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW convergence

Post by Daniele Varsano » Thu Aug 05, 2021 9:06 am

Dear Haseeb,
yes, as you said, you need to provide the number of iterations you want to perform.
Anyway, there were some issues on the automatic procedure that I'm not sure they have been fixed and I suggest you to follow the way indicated in the tutorial:
http://www.yambo-code.org/wiki/index.ph ... alues_only

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

haseebphysics1
Posts: 169
Joined: Sat Aug 17, 2019 2:48 pm

Re: GW convergence

Post by haseebphysics1 » Fri Aug 06, 2021 8:27 am

Dear Dr. Daniele,

I followed the tutorial (http://www.yambo-code.org/wiki/index.ph ... alues_only). And inserted the GfnQPdb and XfnQPdb in the new input (G1W1). However, the new calculation on the same -J directory did not seem to have calculated anything new. And it finished in few seconds after fitting the energies! Was not the new ndb.QP database would be created in the -J, which would be used in further BSE? The input, report, and log files are attached. It seems like the ndb.QP database is read correctly, but the gap is not opened any further and for any further iteration, how will I update the GfnQPdb and XfnQPdb?

Report file link: https://drive.google.com/file/d/1wuUl0T ... sp=sharing

Thanks,
You do not have the required permissions to view the files attached to this post.
Haseeb Ahmad
MS - Physics,
LUMS - Pakistan

User avatar
Daniele Varsano
Posts: 2992
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW convergence

Post by Daniele Varsano » Fri Aug 06, 2021 8:54 am

Dear Haseeb,
if you redirect in the same directory and ndb.QP si found it is not calculated but just read.
You need either to redirect the output in a different directory or change the name to the database.

More explicitly:

yambo -F g0w0.in -J g0w0

this will create the database ./g0W0/ndb.QP

build a g1w1.in file using:
GfnQPdb="E < ./g0W0/ndb.QP"
XfnQPdb="E < ./g0W0/ndb.QP"

and run:
yambo -F g1w1.in -J g1w1

repeat the procedure until convergence updating the previous directory in GfnQPdb and XfnQPdb in a g2w2.in file
GfnQPdb="E < ./g1W1/ndb.QP"
XfnQPdb="E < ./g1W1/ndb.QP"

yambo -F g2w2.in -J g2w2


Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

haseebphysics1
Posts: 169
Joined: Sat Aug 17, 2019 2:48 pm

Re: GW convergence

Post by haseebphysics1 » Fri Aug 06, 2021 9:04 am

Dear Dr. Daniele,

Thanks very much for the prompt reply. In using new -J each time, will not it would calculate the dipoles, PP databases, and ndb.HF_and_locXC database from scratch (because they are placed in a single directory)? So, in that case, can I just change the name of my ndb.QP which was calculated from G0W0 and hence changing the name of each ndb.QP that would be subsequently created in each iteration.

Is this a correct procedure?

Thanks in advance,
Haseeb Ahmad
MS - Physics,
LUMS - Pakistan

User avatar
Daniele Varsano
Posts: 2992
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW convergence

Post by Daniele Varsano » Fri Aug 06, 2021 5:31 pm

Yes it is correct,
But you can also copy or link dipoles etc in the new or in the SAVE directory. The SAVE dir is always read. Note that if you want to do self consistency also in W, the pp database needs to be recalculated.

Best
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

haseebphysics1
Posts: 169
Joined: Sat Aug 17, 2019 2:48 pm

Re: GW convergence

Post by haseebphysics1 » Sat Aug 14, 2021 8:02 am

Dear Yambo developers,

In GoWo calculations the gap was opened by almost 0.10 eV but the subsequent GW (G1W1 on top of G0W0) resulted in the negative corrections of 0.46 eV resulting in to decrease in the gap!! This was not what I was expecting. The calculations include SOC effects while I also decreased the band range in QPkrange slightly in the G1W1 calculations (but still including plenty of bands around Fermi). The inputs are also attached herein. Can you please suggest what I might be doing wrong here?

This is the result of the report from the G2W2 (when reading G1W1 ndb.QP database)

Code: Select all

 [04] External/Internal QP corrections
 =====================================

 [RD././G1W1/ndb.QP]-------------------------------------------------------------
  Lattice constants                                :  10.49403  31.75089  10.56345
  Coulomb cutoff potential                         : none
  Electronic Temperature                           :  300.1998 [K]
  Bosonic    Temperature                           :  300.0997 [K]
  Green`s function   energies                      : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
  wavefunctions                                    : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
  PPA    diel. fun.  energies                      : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
  wavefunctions                                    : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
  QP tot states                                    :  2400
  GWo SC iterations                                : 0
  GW  SC iterations                                : 0
  GW solver                                        : Newton
  GW approximation                                 : PPA
  PPA imaginary Energy                             :  27.21138 ev
  GW SC iterations                                 : 0
  dS/dw steps                                      :  2
  dS/dw step                                       :  0.100000 ev
  X G`s                                            :   695 used
  X G`s                                            :   695 disk
  X bands                                          :     1  1000
  X poles                                          :  100.0000 o/o
  X e/h E range                                    : -1.000000 -1.000000 ev
  X xc-Kernel                                      : none
  X BZ energy Double Grid                          : no
  Sc/G bands                                       :     1  1000
  Sc/G damping                                     :  0.100000 ev
  Sc bands terminator                              : no
  Sx RL components                                 :   5353
                                                   
  QP @ state[ 1 ] K range                          :    1   40
  QP @ state[ 1 ] b range                          :   347   406
  GF energies kind                                 :  E < ./gw_conv-20EXX_80VXX_40FFTG_bands1-1000_5NGBklx/ndb.QP + E Fit
  GF WF`s kind                                     : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
  Xs energies kind                                 : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
  Xs WF`s kind                                     : Perdew, Burke & Ernzerhof(X)+Perdew, Burke & Ernzerhof(C)
 - S/N 009346 ---------------------------------------------- v.05.00.02 r.19567 -

 E<./G1W1/ndb.QP[ PPA@E  27.21138 = XG 695 = Xb 1-1000 = Scb 1-1000]

 [dE_from_DB-Nearest K] Exact matches       :  100.0000 [o/o]
  


  [04.01] FIT on E < ./G1W1/ndb.QP
  ================================

  = FIT report for E<./G1W1/ndb.QP[ PPA@E  27.21138 = XG 695 = Xb 1-1000 = Scb 1-1000] =
  = First column (conduction), Second column (valence) =

   Gap correction (database)     : -0.461772 [eV]
               (FIT, actual)     :  0.184843 [eV]
   Energies 0th order            :  0.564467  0.841396 [eV]
   Energies 1st order            :  1.018448  0.838082
                            Error:  0.002501  0.003491
   

  [QP_apply] Action to be applied: E<./G1W1/ndb.QP[ PPA@E  27.21138 = XG 695 = Xb 1-1000 = Scb 1-1000]

  [05.01]  QP corrections report
  ==============================

  [X+QP] === General ===
  [X+QP] Electronic Temperature                     :  0.258779E-1   300.300    [eV K]
  [X+QP] Bosonic    Temperature                     :  0.258606E-1   300.100    [eV K]
  [X+QP] Finite Temperature mode                    : yes
  [X+QP] El. density                                :  0.72091E+24 [cm-3]
  [X+QP] Fermi Level                                :  0.797699 [eV]

  [X+QP] === Gaps and Widths ===
  [X+QP] Conduction Band Min                        :  0.856257 [eV]
  [X+QP] Valence Band Max                           : -0.906222 [eV]
  [X+QP] Filled Bands                               :   376
  [X+QP] Empty Bands                                :   377  1000
  [X+QP] Direct Gap                                 :  1.802209 [eV]
  [X+QP] Direct Gap localized at k-point            :   9
  [X+QP] Indirect Gap                               :  1.762479 [eV]
  [X+QP] Indirect Gap between k-points              :   1   9
  [X+QP] Last valence band width                    :  0.510743 [eV]
  [X+QP] 1st conduction band width                  :  1.274906 [eV]
   
   
  [coarse-grid] Minimum Bare Gap        [ev]:  2.224229
  [coarse-grid] Minimum Gap Correction  [ev]: -0.461750

Thanks,

PS (after post): It seems like a similar issue was reported here earlier too!
viewtopic.php?f=30&t=2002&p=10308&hilit ... +GW#p10308
I'm not sure whether the parallelization problem exists also in v5.0.2 or not!
You do not have the required permissions to view the files attached to this post.
Haseeb Ahmad
MS - Physics,
LUMS - Pakistan

User avatar
claudio
Posts: 359
Joined: Tue Mar 31, 2009 11:33 pm
Contact:

Re: GW convergence

Post by claudio » Tue Aug 31, 2021 11:11 am

Dear Haseeb Ahmad

I don't know if this is due to SOC etc...
please provide us DFT and Yambo input we will try to reproduce this bug

best
Claudio
Claudio Attaccalite
[CNRS/ Aix-Marseille Université/ CINaM laborarory / TSN department
Campus de Luminy – Case 913
13288 MARSEILLE Cedex 09
web site: http://www.attaccalite.com

Post Reply