GW self-consistency on eigenvalues

Various technical topics such as parallelism and efficiency, netCDF problems, the Yambo code structure itself, are posted here.

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano, Conor Hogan, Nicola Spallanzani

sponce
Posts: 12
Joined: Tue Oct 04, 2011 9:52 am

GW self-consistency on eigenvalues

Post by sponce » Thu Jan 11, 2018 12:08 pm

Dear all,

I have two questions:
1) We can run Yambo self-consistently using keywords:
XfnQPdb = 'E < ./ndb.QP'
GfnQPdb = 'E < ./ndb.QP'

What is happening to the states above the states explicitly considered in the GW calculations? Are the DFT value taken or are all the states above shifted by the last state correction ?

2) Because such eigenvalue correction might break the f-sum rule, we would like to do a k-resolved scissor shift of the eigenvalues. Basically taking the CBM GW shift and applying it to all the states above for each k-points.
We tried this manually with a script that reads the ndb.QP file and update the SAVE/ns.db1 file (this file was originally produced by p2y).
The issue is that this gives the following correction (here GaN):
STEP1:
# K-point Band Eo E-Eo Sc|Eo
1.00000 36.00000 0.00000 0.27347 3.30133
1.00000 37.00000 1.93647 0.36405 -3.50613
STEP 2:
1.00000 36.00000 0.00000 0.18790 3.19005
1.00000 37.00000 2.02706 0.42061 -3.44100
STEP 3:
1.00000 36.00000 0.00000 0.09934 3.07721
1.00000 37.00000 2.25977 0.48511 -3.36805
STEP 4:
1.000 36.00 0.000 0.3390E-2 2.958
1.00000 37.00000 2.64555 0.56132 -3.28334

As you can see the VBM (band 36) correction correctly goes to 0. as the number of step increase. However the CBM correction always increase, therefore never converging.

What is the correct procedure to do this ?

Thanks,
Best,
Samuel
Samuel Poncé
Department of Materials, University of Oxford

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW self-consistency on eigenvalues

Post by Daniele Varsano » Thu Jan 11, 2018 12:41 pm

Dear Samuel,
What is happening to the states above the states explicitly considered in the GW calculations? Are the DFT value taken or are all the states above shifted by the last state correction ?
The states above (and below) the one explicitely considered are shifted accordingly to a fit done on the calculalted values. This is true when KfnQPdb is considered, i.e. when QP energies are inserted in the Bethe Salpeter equation.

When considering eigenvalue self-consistency (XfnQPdb and GfnQPdb) I'm afraid they are left as they are (KS energies). Anyway you can easily check if in the report you have or not the results of the fit. Probably should be not too hard to tell yambo to do the fit and apply it also in that case. I need to check.

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/

sponce
Posts: 12
Joined: Tue Oct 04, 2011 9:52 am

Re: GW self-consistency on eigenvalues

Post by sponce » Thu Jan 11, 2018 3:00 pm

Dear Daniele,

Thanks for the answer.

In any case, what I'm most interested in, is to do a k-resolved scissor shift of all the CB (based on the shift of the CBM) and of all the VB (based on the shift of the VBM).

This is what I tried to do with my script. However it seems the GW correction increase indefinitely (see point 2) of my original post). Any idea how to solve this?

Thanks,
Samuel
Samuel Poncé
Department of Materials, University of Oxford

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW self-consistency on eigenvalues

Post by Daniele Varsano » Thu Jan 11, 2018 3:53 pm

Dear Samuel,
I used to do eigenvalues self consistently but only for molecules, it worked provided that I included all the qp correction (all occupied and empty states included in the screening).
I think that if you have a mix of KS and QP eigenvalues you have a problem in the calculations of the screening. This is just a conjecture.
As I told in the other post, Yambo interpolates the QP corrections for the states not explicitly calculated, this is done for the BS case, but may also for the XfnQPdb and GfnQPdb
but then they are not stored in databases.

A possibility it comes me to mind is the following:
1) generate by hand the QP database assigning the desired scissor for each kpoints for all the bands. This can be done by using ypp.
a) ypp -q g
b) fill the ypp.in file assiging the qp corrections (the syntax should be clear from the legend in the input file)
c) run ypp
This will generate the desired ndb.QP

If you have many k points/bands the ypp.in file could be too long and not properly read by the ypp parser (around 200 lines).
In this case, you can split it into different inputs generating different databases and then merge them by using again ypp
a) ypp -q merge
b) fill ypp.in using this syntax:
QPDBs # [R] Quasi-particle databases
QPDB_merge # [R] Mergering
%Actions_and_names # [QPDB] Format is "what"|"OP"|"prefactor"|"DB"|. OP can be +/-/x(ony for Z).
"E" | "+" | "1" | "./qp1/ndb.QP" |
"E" | "+" | "1" | "./qp2/ndb.QP" |
"E" | "+" | "1" | "./qp3/ndb.QP" |
"E" | "+" | "1" | "./qp4/ndb.QP" |
...
%
c) run ypp
This will produce a unique database containing the correction for all the bands and kpoint.

I know that the procedure it is cumbersome, but it can possibly be automatized by some scripts.
Hope it helps,

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/

martinspenke
Posts: 149
Joined: Tue Apr 08, 2014 6:05 am

Re: GW self-consistency on eigenvalues

Post by martinspenke » Thu Jan 11, 2018 8:58 pm

Dear Samuel,
Ok, sorry.
Best,
Martin
Last edited by martinspenke on Fri Jan 12, 2018 1:46 pm, edited 2 times in total.
Martin Spenke, PhD Student
Theoretisch-Physikalisches Institut
Universität Hamburg, Germany

sponce
Posts: 12
Joined: Tue Oct 04, 2011 9:52 am

Re: GW self-consistency on eigenvalues

Post by sponce » Fri Jan 12, 2018 1:32 pm

Dear Daniele,

Thank you very much for the reply.
I have try your suggestion:
1) ypp -q g
2) Filled ypp.in as follow:

Code: Select all

QPDBs                          # [R] Quasi-particle databases
QPDB_edit                      # [R] Generation/editing
%QPkrange            # [QPDB] Correction( kp1| kp2| bnd1| bnd2| E-Eo[eV]| Img(E)[eV]| Re[Z] |)
  1|  1|  1| 36| 0.27 | 0.0 | 0.79 |
  1|  1| 37| 40| 0.36 | 0.0 | 0.81 |
  2|  2|  1| 36| 0.22 | 0.0 | 0.79 |
  2|  2| 37| 40| 0.59 | 0.0 | 0.81 |
  3|  3|  1| 36| 0.09 | 0.0 | 0.79 |
  3|  3| 37| 40| 0.59 | 0.0 | 0.81 |
  4|  4|  1| 36| 0.10 | 0.0 | 0.79 |
  4|  4| 37| 40| 0.64 | 0.0 | 0.80 |
  5|  5|  1| 36| 0.03 | 0.0 | 0.79 |
  5|  5| 37| 40| 0.54 | 0.0 | 0.81 |
  6|  6|  1| 36|-0.13 | 0.0 | 0.78 |
  6|  6| 37| 40| 0.60 | 0.0 | 0.81 |
  7|  7|  1| 36|-0.07 | 0.0 | 0.78 |
  7|  7| 37| 40| 0.49 | 0.0 | 0.83 |
  8|  8|  1| 36|-0.15 | 0.0 | 0.78 |
  8|  8| 37| 40| 0.49 | 0.0 | 0.82 |
  9|  9|  1| 36|-0.28 | 0.0 | 0.78 |
  9|  9| 37| 40| 0.43 | 0.0 | 0.82 |
 10| 10|  1| 36|-0.28 | 0.0 | 0.79 |
 10| 10| 37| 40| 0.54 | 0.0 | 0.82 |
 11| 11|  1| 36|-0.20 | 0.0 | 0.79 |
 11| 11| 37| 40| 0.70 | 0.0 | 0.82 |
 12| 12|  1| 36|-0.26 | 0.0 | 0.79 |
 12| 12| 37| 40| 0.66 | 0.0 | 0.81 |
 13| 13|  1| 36|-0.02 | 0.0 | 0.79 |
 13| 13| 37| 40| 0.55 | 0.0 | 0.81 |
%
Question: Should I put 40 (max GW corrected band) or 200 (max nb of band in my DFT calculation) ?

3) Run ypp with yambo-4.2.0/bin/ypp -F ypp.in
I'm getting

Code: Select all

 <---> [01] CPU structure, Files & I/O Directories
 <---> [02] Y(ambo) P(ost)/(re) P(rocessor)
 <---> [03] Core DB
 <---> :: Electrons             : 36.00000
 <---> :: Temperature       [ev]: 0.000000
 <---> :: Lattice factors [a.u.]: 6.026000  5.218669  9.799999
 <---> :: K points              : 13
 <---> :: Bands                 : 200
 <---> :: Symmetries            : 12
 <---> :: RL vectors            :  68977
 <---> [04] K-point grid
 <---> :: Q-points   (IBZ): 13
 <---> :: X K-points (IBZ): 13
 <---> [05] CORE Variables Setup
 <---> [05.01] Unit cells
 <---> [05.02] Symmetries
 <---> [05.03] RL shells
 <---> [05.04] K-grid lattice
 <---> [05.05] Energies [ev] & Occupations          40          13
However no ndb.QP is produced or modify. Note that my current ndb.QP file is in the directory where I run ypp. This file does not get modify and or no new file is produced.

I guess that after this I should then run Yambo with:
XfnQPdb = 'E < ./ndb.QP'
GfnQPdb = 'E < ./ndb.QP'

PS: Note that the self-consistent procedure in Yambo work fine for me. What I want to do is a scissor shift that is k dependent based on VBM and CBM shift.

Thank you,
Best,
Samuel
Samuel Poncé
Department of Materials, University of Oxford

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW self-consistency on eigenvalues

Post by Daniele Varsano » Fri Jan 12, 2018 1:52 pm

Dear Samuel,
Question: Should I put 40 (max GW corrected band) or 200 (max nb of band in my DFT calculation) ?
Well, I would say it depends on how many bands you include in the calculation of the response X, in general the number of the bands you are actually using.
However no ndb.QP is produced or modify.
It looks the run did not finished properly, did you get any error message?

Actually, when I used it I used to specify the correction band by bands
1| 1| 1| 1| 0.27 | 0.0 | 0.79
1| 1| 2| 2| 0.27 | 0.0 | 0.79
...
even if it is supposed to be working also in the why you used it.
Note that my current ndb.QP file is in the directory where I run ypp.
Yambo reads the databases in the SAVE directory and the one in the directory indicated by the -J option. In this case the ndb.QP you have should be ignored and a new database created,
but it looks ypp stopped before starting iìhis job.
I guess that after this I should then run Yambo with:
XfnQPdb = 'E < ./ndb.QP'
GfnQPdb = 'E < ./ndb.QP'
For a GW calculation yes, for BSE you need to use KfnQPdb
What I want to do is a scissor shift that is k dependent based on VBM and CBM shift.
If I well understood what you want to do I think you need to assign a zero correction for all the valence bands and a k dependent shift to the conduction bands right? This should do the job, but you would need to repeat the ypp stuff at each iteration.
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/

sponce
Posts: 12
Joined: Tue Oct 04, 2011 9:52 am

Re: GW self-consistency on eigenvalues

Post by sponce » Fri Jan 12, 2018 3:27 pm

Hello Daniele,

I do not get any error.

By doing

Code: Select all

~/program/yambo-4.2.0/bin/ypp -q g -J gw1.out
I get an ypp.in file that is automatically generated:

Code: Select all

#                                                           
# :::   :::   :::     ::::    ::::  :::::::::   ::::::::    
# :+:   :+: :+: :+:   +:+:+: :+:+:+ :+:    :+: :+:    :+:   
#  +:+ +:+ +:+   +:+  +:+ +:+:+ +:+ +:+    +:+ +:+    +:+   
#   +#++: +#++:++#++: +#+  +:+  +#+ +#++:++#+  +#+    +:+   
#    +#+  +#+     +#+ +#+       +#+ +#+    +#+ +#+    +#+   
#    #+#  #+#     #+# #+#       #+# #+#    #+# #+#    #+#   
#    ###  ###     ### ###       ### #########   ########    
#                                                           
#                                                           
# GPL Version 4.2.0 Revision 109. (Based on r.10030 h.c8f2c8
#                        MPI Build                          
#                http://www.yambo-code.org                  
#
QPDBs                          # [R] Quasi-particle databases
QPDB_edit                      # [R] Generation/editing
%QPkrange            # [QPDB] Correction( kp1| kp2| bnd1| bnd2| E-Eo[eV]| Img(E)[eV]| Re[Z] |)
  1|  1| 24| 24|    1.213652|   -0.000982|    0.767763|
  1|  1| 25| 25|   -0.916251|    0.019443|    0.739247|
  1|  1| 26| 26|   -0.916310|    0.019444|    0.739289|
  1|  1| 27| 27|   -0.071164|    0.014375|    0.784223|
  1|  1| 28| 28|   -0.071160|    0.014375|    0.784218|
  1|  1| 29| 29|   -0.072318|    0.014312|    0.784258|
  1|  1| 30| 30|   -0.072304|    0.014314|    0.784331|
  1|  1| 31| 31|    0.252207|    0.011801|    0.789587|
  1|  1| 32| 32|    0.252207|    0.011802|    0.789583|
  1|  1| 33| 33|    0.275638|    0.012679|    0.786816|
  1|  1| 34| 34|    0.275638|    0.012679|    0.786802|
  1|  1| 35| 35|    0.273447|    0.012673|    0.786670|
  1|  1| 36| 36|    0.273466|    0.012674|    0.786724|
  1|  1| 37| 37|    0.364054|   -0.011288|    0.813454|
  1|  1| 38| 38|    0.364045|   -0.011288|    0.813430|
  1|  1| 39| 39|    0.492504|   -0.013537|    0.818497|
  1|  1| 40| 40|    0.492510|   -0.013537|    0.818508|
  2|  2| 24| 24|    1.239105|   -0.016847|    0.712162|
  2|  2| 25| 25|   -0.721479|    0.013748|    0.759928|
  2|  2| 26| 26|   -0.721447|    0.013747|    0.759893|
  2|  2| 27| 27|   -0.057842|    0.012484|    0.789652|
  2|  2| 28| 28|   -0.057842|    0.012484|    0.789637|
  2|  2| 29| 29|   -0.025972|    0.014152|    0.784603|
  2|  2| 30| 30|   -0.025987|    0.014152|    0.784605|
  2|  2| 31| 31|   -0.026755|    0.014088|    0.784543|
  2|  2| 32| 32|   -0.026805|    0.014091|    0.784712|
  2|  2| 33| 33|    0.219379|    0.012986|    0.786329|
  2|  2| 34| 34|    0.219389|    0.012986|    0.786334|
  2|  2| 35| 35|    0.217263|    0.012935|    0.786373|
  2|  2| 36| 36|    0.217295|    0.012936|    0.786408|
  2|  2| 37| 37|    0.587602|   -0.011505|    0.806302|
  2|  2| 38| 38|    0.587582|   -0.011505|    0.806281|
  2|  2| 39| 39|    0.492533|   -0.013628|    0.816111|
  2|  2| 40| 40|    0.492477|   -0.013626|    0.816011|
  3|  3| 24| 24|    1.896223|    0.014410|    0.740088|
  3|  3| 25| 25|   -0.397915|    0.016330|    0.778374|
  3|  3| 26| 26|   -0.397882|    0.016329|    0.778315|
  3|  3| 27| 27|   -0.409100|    0.016408|    0.778408|
  3|  3| 28| 28|   -0.409081|    0.016407|    0.778376|
  3|  3| 29| 29|    0.091023|    0.013589|    0.785577|
  3|  3| 30| 30|    0.091022|    0.013588|    0.785520|
  3|  3| 31| 31|    0.090360|    0.013590|    0.785585|
  3|  3| 32| 32|    0.090366|    0.013590|    0.785595|
  3|  3| 33| 33|    0.090361|    0.013532|    0.785647|
  3|  3| 34| 34|    0.090314|    0.013532|    0.785657|
  3|  3| 35| 35|    0.089710|    0.013532|    0.785610|
  3|  3| 36| 36|    0.089693|    0.013531|    0.785570|
  3|  3| 37| 37|    0.592048|   -0.013480|    0.807839|
  3|  3| 38| 38|    0.592037|   -0.013480|    0.807821|
  3|  3| 39| 39|    0.585604|   -0.013466|    0.807810|
  3|  3| 40| 40|    0.585582|   -0.013465|    0.807776|
  4|  4| 24| 24|    1.028255|    0.009962|    0.720737|
  4|  4| 25| 25|   -0.848255|    0.021613|    0.751260|
  4|  4| 26| 26|   -0.847524|    0.020922|    0.751235|
  4|  4| 27| 27|   -0.583741|    0.017291|    0.776726|
  4|  4| 28| 28|   -0.583353|    0.017241|    0.776868|
  4|  4| 29| 29|   -0.249662|    0.015056|    0.781767|
  4|  4| 30| 30|   -0.243247|    0.015032|    0.781911|
  4|  4| 31| 31|   -0.318750|    0.015544|    0.781856|
  4|  4| 32| 32|   -0.317437|    0.015541|    0.781960|
  4|  4| 33| 33|    0.011032|    0.013088|    0.788495|
  4|  4| 34| 34|    0.022375|    0.013044|    0.788554|
  4|  4| 35| 35|    0.098107|    0.013428|    0.785773|
  4|  4| 36| 36|    0.102303|    0.013417|    0.785899|
  4|  4| 37| 37|    0.639774|   -0.012869|    0.804978|
  4|  4| 38| 38|    0.648063|   -0.012893|    0.804857|
  4|  4| 39| 39|    0.630903|   -0.014480|    0.816591|
  4|  4| 40| 40|    0.633445|   -0.014479|    0.816521|
  5|  5| 24| 24|    1.082700|    0.024017|    0.729676|
  5|  5| 25| 25|   -0.652942|    0.017886|    0.759831|
  5|  5| 26| 26|   -0.653569|    0.017810|    0.759950|
  5|  5| 27| 27|   -0.453699|    0.016560|    0.778209|
  5|  5| 28| 28|   -0.446207|    0.016528|    0.778148|
  5|  5| 29| 29|   -0.413918|    0.016096|    0.780470|
  5|  5| 30| 30|   -0.415222|    0.016087|    0.780493|
  5|  5| 31| 31|   -0.267067|    0.015261|    0.782732|
  5|  5| 32| 32|   -0.265402|    0.015251|    0.782606|
  5|  5| 33| 33|   -0.026817|    0.013436|    0.787320|
  5|  5| 34| 34|   -0.025815|    0.013435|    0.787410|
  5|  5| 35| 35|    0.024972|    0.013767|    0.785515|
  5|  5| 36| 36|    0.028521|    0.013753|    0.785557|
  5|  5| 37| 37|    0.544842|   -0.013101|    0.810063|
  5|  5| 38| 38|    0.544831|   -0.013104|    0.810052|
  5|  5| 39| 39|    0.652955|   -0.014783|    0.811017|
  5|  5| 40| 40|    0.655474|   -0.014770|    0.810958|
  6|  6| 24| 24|    1.471603|    0.052630|    0.753462|
  6|  6| 25| 25|   -0.439787|    0.017636|    0.770504|
  6|  6| 26| 26|   -0.442536|    0.017419|    0.770538|
  6|  6| 27| 27|   -0.437722|    0.017447|    0.770711|
  6|  6| 28| 28|   -0.441456|    0.016821|    0.771514|
  6|  6| 29| 29|   -0.316606|    0.014892|    0.785105|
  6|  6| 30| 30|   -0.317783|    0.014894|    0.785137|
  6|  6| 31| 31|   -0.309273|    0.014853|    0.785112|
  6|  6| 32| 32|   -0.310042|    0.014853|    0.785120|
  6|  6| 33| 33|   -0.131548|    0.014537|    0.784231|
  6|  6| 34| 34|   -0.132305|    0.014540|    0.784303|
  6|  6| 35| 35|   -0.129039|    0.014529|    0.784363|
  6|  6| 36| 36|   -0.129735|    0.014529|    0.784283|
  6|  6| 37| 37|    0.596988|   -0.014090|    0.809970|
  6|  6| 38| 38|    0.595744|   -0.014086|    0.810002|
  6|  6| 39| 39|    0.601342|   -0.014102|    0.809956|
  6|  6| 40| 40|    0.599929|   -0.014096|    0.809938|
  7|  7| 24| 24|    0.778347|   -0.007106|    0.703223|
  7|  7| 25| 25|   -0.786831|    0.020867|    0.757648|
  7|  7| 26| 26|   -0.786804|    0.020872|    0.757609|
  7|  7| 27| 27|   -0.702944|    0.025598|    0.766814|
  7|  7| 28| 28|   -0.703522|    0.025323|    0.767387|
  7|  7| 29| 29|   -0.683926|    0.020438|    0.771357|
  7|  7| 30| 30|   -0.683923|    0.020436|    0.771360|
  7|  7| 31| 31|   -0.520646|    0.016790|    0.778214|
  7|  7| 32| 32|   -0.520682|    0.016792|    0.778268|
  7|  7| 33| 33|   -0.347878|    0.014704|    0.787119|
  7|  7| 34| 34|   -0.347902|    0.014705|    0.787158|
  7|  7| 35| 35|   -0.070535|    0.014202|    0.784348|
  7|  7| 36| 36|   -0.070506|    0.014202|    0.784351|
  7|  7| 37| 37|    0.486636|   -0.013830|    0.826689|
  7|  7| 38| 38|    0.486633|   -0.013829|    0.826695|
  7|  7| 39| 39|    0.694194|   -0.014357|    0.811379|
  7|  7| 40| 40|    0.694172|   -0.014356|    0.811353|
  8|  8| 24| 24|    0.843046|    0.052489|    0.728388|
  8|  8| 25| 25|   -0.827676|    0.021805|    0.754382|
  8|  8| 26| 26|   -0.827604|    0.021777|    0.754322|
  8|  8| 27| 27|   -0.748766|    0.020917|    0.764240|
  8|  8| 28| 28|   -0.748767|    0.020993|    0.764225|
  8|  8| 29| 29|   -0.526713|    0.016564|    0.779484|
  8|  8| 30| 30|   -0.526791|    0.016561|    0.779513|
  8|  8| 31| 31|   -0.464765|    0.016480|    0.779173|
  8|  8| 32| 32|   -0.464835|    0.016482|    0.779286|
  8|  8| 33| 33|   -0.248991|    0.014392|    0.786395|
  8|  8| 34| 34|   -0.249000|    0.014393|    0.786433|
  8|  8| 35| 35|   -0.151184|    0.014594|    0.783937|
  8|  8| 36| 36|   -0.151179|    0.014593|    0.783929|
  8|  8| 37| 37|    0.490921|   -0.013062|    0.822850|
  8|  8| 38| 38|    0.491230|   -0.013063|    0.822905|
  8|  8| 39| 39|    0.470086|   -0.013860|    0.823685|
  8|  8| 40| 40|    0.469958|   -0.013860|    0.823746|
  9|  9| 24| 24|    1.061489|   -0.010680|    0.775675|
  9|  9| 25| 25|   -0.807649|    0.019958|    0.755974|
  9|  9| 26| 26|   -0.807647|    0.019937|    0.755967|
  9|  9| 27| 27|   -0.808809|    0.021035|    0.755307|
  9|  9| 28| 28|   -0.808750|    0.021042|    0.755263|
  9|  9| 29| 29|   -0.316559|    0.015447|    0.782064|
  9|  9| 30| 30|   -0.316586|    0.015447|    0.782052|
  9|  9| 31| 31|   -0.317188|    0.015446|    0.782027|
  9|  9| 32| 32|   -0.317199|    0.015446|    0.782027|
  9|  9| 33| 33|   -0.274410|    0.014834|    0.784283|
  9|  9| 34| 34|   -0.274335|    0.014834|    0.784310|
  9|  9| 35| 35|   -0.275329|    0.014837|    0.784226|
  9|  9| 36| 36|   -0.275295|    0.014837|    0.784231|
  9|  9| 37| 37|    0.425392|   -0.013243|    0.823042|
  9|  9| 38| 38|    0.425379|   -0.013242|    0.823020|
  9|  9| 39| 39|    0.425617|   -0.013193|    0.823127|
  9|  9| 40| 40|    0.425612|   -0.013194|    0.823135|
 10| 10| 24| 24|    0.940387|   -0.011150|    0.703700|
 10| 10| 25| 25|   -0.793680|    0.023024|    0.767342|
 10| 10| 26| 26|   -0.793489|    0.022878|    0.767305|
 10| 10| 27| 27|   -0.649636|    0.015457|    0.772421|
 10| 10| 28| 28|   -0.651500|    0.015482|    0.772971|
 10| 10| 29| 29|   -0.680160|    0.017198|    0.775340|
 10| 10| 30| 30|   -0.673637|    0.015812|    0.775836|
 10| 10| 31| 31|   -0.660216|    0.015086|    0.776422|
 10| 10| 32| 32|   -0.657289|    0.015001|    0.777141|
 10| 10| 33| 33|   -0.288328|    0.015070|    0.784136|
 10| 10| 34| 34|   -0.285346|    0.015059|    0.784186|
 10| 10| 35| 35|   -0.286871|    0.014819|    0.785462|
 10| 10| 36| 36|   -0.277031|    0.014777|    0.785570|
 10| 10| 37| 37|    0.540249|   -0.014028|    0.823988|
 10| 10| 38| 38|    0.548562|   -0.014043|    0.823861|
 10| 10| 39| 39|    0.978475|   -0.015999|    0.793897|
 10| 10| 40| 40|    0.985145|   -0.016007|    0.793856|
 11| 11| 24| 24|    1.002738|    0.026010|    0.753123|
 11| 11| 25| 25|   -0.766534|    0.022240|    0.759641|
 11| 11| 26| 26|   -0.767343|    0.022253|    0.759867|
 11| 11| 27| 27|   -0.627069|    0.020825|    0.766123|
 11| 11| 28| 28|   -0.627467|    0.020570|    0.765804|
 11| 11| 29| 29|   -0.678127|    0.021025|    0.778554|
 11| 11| 30| 30|   -0.674872|    0.021044|    0.778351|
 11| 11| 31| 31|   -0.543683|    0.016432|    0.780641|
 11| 11| 32| 32|   -0.534641|    0.016407|    0.780744|
 11| 11| 33| 33|   -0.375123|    0.015434|    0.783697|
 11| 11| 34| 34|   -0.374395|    0.015411|    0.783717|
 11| 11| 35| 35|   -0.203925|    0.014341|    0.785677|
 11| 11| 36| 36|   -0.198183|    0.014324|    0.785692|
 11| 11| 37| 37|    0.702662|   -0.014386|    0.815641|
 11| 11| 38| 38|    0.711623|   -0.014361|    0.816410|
 11| 11| 39| 39|    0.710287|   -0.014697|    0.808687|
 11| 11| 40| 40|    0.714609|   -0.014736|    0.807710|
 12| 12| 24| 24|    1.067823|    0.018668|    0.735573|
 12| 12| 25| 25|   -0.689551|    0.020982|    0.762945|
 12| 12| 26| 26|   -0.687154|    0.019909|    0.761145|
 12| 12| 27| 27|   -0.694604|    0.021093|    0.763109|
 12| 12| 28| 28|   -0.692163|    0.019883|    0.761242|
 12| 12| 29| 29|   -0.563277|    0.016529|    0.780921|
 12| 12| 30| 30|   -0.564599|    0.016541|    0.780787|
 12| 12| 31| 31|   -0.555571|    0.016901|    0.781211|
 12| 12| 32| 32|   -0.556927|    0.016838|    0.781115|
 12| 12| 33| 33|   -0.261147|    0.014594|    0.785080|
 12| 12| 34| 34|   -0.263081|    0.014597|    0.785015|
 12| 12| 35| 35|   -0.258890|    0.014575|    0.785130|
 12| 12| 36| 36|   -0.260660|    0.014576|    0.785035|
 12| 12| 37| 37|    0.656823|   -0.014239|    0.814340|
 12| 12| 38| 38|    0.664626|   -0.014248|    0.814270|
 12| 12| 39| 39|    0.658818|   -0.014262|    0.814257|
 12| 12| 40| 40|    0.666797|   -0.014272|    0.814240|
 13| 13| 24| 24|    1.088628|    0.019387|    0.733721|
 13| 13| 25| 25|   -0.655243|    0.016638|    0.762150|
 13| 13| 26| 26|   -0.655960|    0.016580|    0.762399|
 13| 13| 27| 27|   -0.451684|    0.016804|    0.778436|
 13| 13| 28| 28|   -0.444224|    0.016736|    0.778299|
 13| 13| 29| 29|   -0.412319|    0.016091|    0.780364|
 13| 13| 30| 30|   -0.413598|    0.016083|    0.780446|
 13| 13| 31| 31|   -0.270923|    0.015255|    0.782673|
 13| 13| 32| 32|   -0.269290|    0.015248|    0.782660|
 13| 13| 33| 33|   -0.025953|    0.013439|    0.787385|
 13| 13| 34| 34|   -0.024939|    0.013436|    0.787376|
 13| 13| 35| 35|    0.020586|    0.013762|    0.785580|
 13| 13| 36| 36|    0.024160|    0.013746|    0.785532|
 13| 13| 37| 37|    0.547755|   -0.013101|    0.809996|
 13| 13| 38| 38|    0.547652|   -0.013104|    0.809996|
 13| 13| 39| 39|    0.654725|   -0.014800|    0.811049|
 13| 13| 40| 40|    0.657339|   -0.014789|    0.811099|
%
This uses the GW correction from the first iteration. This makes perfect sense. In theory I could modify this myself.
Inside the code, I could follow from INIT_ypp.F that

Code: Select all

 l_QP_DBs_create_and_modify
must be true. From that the code correctly goes into

Code: Select all

QP_DBs_initialize
This correctly produces the ypp.in file.
All great so far. Then I run:

Code: Select all

 ~/program/yambo-4.2.0/bin/ypp -F ypp.in -J gw1.out
This gives me:

Code: Select all

 __   __  _______  __   __  _______  _______
|  | |  ||   _   ||  |_|  ||  _    ||       |
|  |_|  ||  |_|  ||       || |_|   ||   _   |
|       ||       ||       ||       ||  | |  |
|_     _||       ||       ||  _   | |  |_|  |
  |   |  |   _   || ||_|| || |_|   ||       |
  |___|  |__| |__||_|   |_||_______||_______|


 <17s> [01] CPU structure, Files & I/O Directories
 <17s> [02] Y(ambo) P(ost)/(re) P(rocessor)
 <17s> [03] Core DB
 <17s> :: Electrons             : 36.00000
 <17s> :: Temperature       [ev]: 0.000000
 <17s> :: Lattice factors [a.u.]: 6.026000  5.218669  9.799999
 <17s> :: K points              : 13
 <17s> :: Bands                 : 200
 <17s> :: Symmetries            : 12
 <17s> :: RL vectors            :  68977
 <17s> [04] K-point grid
 <17s> :: Q-points   (IBZ): 13
 <17s> :: X K-points (IBZ): 13
 <17s> [05] CORE Variables Setup
 <17s> [05.01] Unit cells
 <17s> [05.02] Symmetries
 <17s> [05.03] RL shells
 <17s> [05.04] K-grid lattice
 <17s> [05.05] Energies [ev] & Occupations          40          13
This produces no ndb.QP files.

However, from the code I cannot see what the code is suppose to do? In which routine ypp should go to create the new database ?

Best,
Samuel
Samuel Poncé
Department of Materials, University of Oxford

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: GW self-consistency on eigenvalues

Post by Daniele Varsano » Fri Jan 12, 2018 4:07 pm

I get an ypp.in file that is automatically generated:
Ok it is already filled as I imagine you have an ndb.QP in the SAVE or in the gw1.out directory.

From the standard output it looks that ypp is hanging, have you checked if it is still running?
This should take few seconds, somehow it takes more if the input file is very large, but this is a problem of the parses.
Actually, your input file is rather long, more than 500 lines.

Can you try to repeat the same removing the ndb.QP from the SAVE directory with a simple input file containing just few kpt and bands?
In which routine ypp should go to create the new database ?
The part of the code doing the job is:
ypp/qp/QP_DBs_create_and_modify.F

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/

sponce
Posts: 12
Joined: Tue Oct 04, 2011 9:52 am

Re: GW self-consistency on eigenvalues

Post by sponce » Fri Jan 12, 2018 6:14 pm

The code is no hanging. I'm running interactively.

There is a barbaric "stop" inside ypp/qp/QP_DBs_create_and_modify.F though ...

Code: Select all

 call QP_state_table_setup(en)
 !
 qp%n_states=QP_n_states
 qp%nb      =QP_nb
 qp%nk      =QP_nk
 ! 
 call QP_alloc(qp)
 write (*,*) QP%nb,QP%nk
 stop
This explains why it does not exist properly.

I'm using Yambo 4.2.0

You might want to remove this if it is still in your trunk.
It seems to be working now.
I will let you know if it solve the problem of self-consistency exploding.

Thanks a lot for your very rapid help !

Best,
Samuel
Samuel Poncé
Department of Materials, University of Oxford

Post Reply