error in running bse

[samaneh]

Dear Developers,

I've run yambo -d -k hartree -g n -p p
without any problem
but I do not know when I run BSE calculations
I get this error
"<36s> [x,Vnl] computed using 248 projectors
<36s> [WARNING] [x,Vnl] slows the Dipoles computation. To neglect it rename the ns.kb_pp file
<36s> [M 0.937 Gb] Alloc KBV ( 0.259)
<36s> [X] Using Kohn-Sham energies for the dipoles
Error STOP signal received while in :[07.01] Main loop
Error in parallel wf distribution"
in log or report files!!
Would you please help me with this problem
Thank you in advance
Best regards
Samaneh

[Daniele Varsano]

Dear Samaneh,
please post input/report and logs files otherwise it is impossible to help you.
Next, remember also to fill your signature with your affiliation. You can do once for all by filling the signature form on your forum profile.

Best,
Daniele

[samaneh]

Dear Daniele,
Thank you for your quick reply
I've attached the needed files
Best regards
S. Ataei, PhD student, University of Tehran, Iran

[Daniele Varsano]

Dear Samaneh,
the input file is not present, the yambo.in file you posted contain a report file.
Please post it, and I will have a look.
Anyway looking at the report it looks there is some confusion with databases of previous calculations of dipoles:

*ERR* Metallic bands : 98

Check what you did in previous runs and what you need to do with the present run.

Best,
Daniele

[samaneh]

Dear Daniele
I've attached the input file.
My pwscf input file contains
"occupations='smearing', smearing='mv', degauss=0.005,"
I do not know if the error :
*ERR* Metallic bands : 98
is because of that?
Thank you
Best regards
Samaneh

[Daniele Varsano]

Dear Samaneh,
please fill your signature with your affiliation. If you do it in your user profile then you can forget about it. Thanks.

I suggest you to clean your SAVE directory as there are files of previous runs in conflict whit the present runs, at least the dipoles file.
Next: why you need to calculate both stsatic (em1s) dynamic (em1d) and plasmon pole screening? It is not needed, BS kernel is build using the static dielectric matrix, or it can be read from the pp file f you have calculated it before. Anyway I suggest you to do a clean calculations from scratch and calculating the static one.
Finally, you are including mostly metallic bands in the BSE, as the Fermi energy is modified by the effect of the QP corrections you are including, have you verified that you still have electron-hole pairs?
From the report file:

Code: Select all

[QP] Fermi level variation [ev]: 1.227149
 [QP] Last Filled/Metallic band :  93  115

I think that this is causing problems. You can try to reduce your electronic temperature.

Best,
Daniele

[samaneh]

Dear Daniele,
Thank you for your suggestions
I will remove elms from input file.
About the SAVE file I should note that
first I did yambo -g n -p p
then ypp -s b to do band calculations
and then since I need the SAVE file for bse calculation
I did yambo -o b -k sex -y h
So, what do you mean by
doing calculation from scratch?
I think I can only remove doing band calculations from the mentioned steps!!!
I do not understand what do you mean by
"have you verified that you still have electron-hole pairs?"
and "You can try to reduce your electronic temperature "
how can I reduce it (I do not have it in my input file?)
and how much should I reduce it?
Best regards
S. Ataei, PhD student, University of Tehran, Iran

[Daniele Varsano]

Dear Samaneh,
I mean to delete the files you have in the SAVE directory from previous runs that are causing problems as you can see by inspecting your report files. In particular the dipole database.

I suggest you to maintain the em1s and remove the em1d and pp (also I would delete the pp database form your SAVE directory).

In order to build the excitonic matrix you need electron-hole pairs, or id you want pairs of occupied and unoccupied states, are you sure you have it after applying the QP corrections, taking into account that the Fermi level shift (please inspect the report file carefully).

You can reduce the electronic temperature by adding in input file the variable:

Code: Select all

ElecTemp= 0.000000     eV   

or activate it by adding the string "-V gen" when building up the input file.

Best,
Daniele

[samaneh]

Dear Daniele,
Many thanks for your help.
I read the report file and I think that my G0W0 calculation with this temperature (300K)
is wrong because all of the width[mev] data in it's output are positive while I have both positive and negative
Eqp or E0 (for holes and electrons)!!!
so I should reduce the electronic temperature in the input file
and recalculate the G0W0 (-g n -p p)?
You mentioned to do bse from scratch just by using emls
(without any database of GW QP corrections ( KfnQPdb= "E < SAVE/db.QP"))
but if I want to use the dynamic response function for bse calculations
I can do 2 jobs
1- doing G0W0 and then use its database for bse (removing emls, emld and ppa from input of bse)
2-doing bse from scratch with emld or ppa in the input of bse
I think I should do G0W0 and then use its database for bse
am I right?
Thanks in advance
Best regards
S. Ataei, University of Tehran, Iran

[Daniele Varsano]

Dear Samaneh,

s wrong because all of the width[mev] data in it's output are positive while I have both positive and negative
Eqp or E0 (for holes and electrons)!!!

This is not really clear to me what do you mean.I do not know which width you are talking about. If you are doing PP, qp widths do not have physical meaning.

I should reduce the electronic temperature in the input file
and recalculate the G0W0 (-g n -p p)?

No, what electronic temperature does , is to change the occupation numbers, you can do it at BSE level.

You mentioned to do bse from scratch just by using emls
(without any database of GW QP corrections ( KfnQPdb= "E < SAVE/db.QP"))

No, I have not said that, you can use QP corrections, I was just suggesting you to calculate the screening from scratch, just an advise.

but if I want to use the dynamic response function for bse calculations
I can do 2 jobs

No, you cannot do that, as the BSE kernel is calculated using the static screening, always.

1- doing G0W0 and then use its database for bse (removing emls, emld and ppa from input of bse)

Not clear what do you mean. which database?

2-doing bse from scratch with emld or ppa in the input of bse

Useless, as the code for building the kernel reads the 0 frequency screening, ie the stati one.

I think I should do G0W0 and then use its database for bse am I right?

Again not clear to me which databases are you talking about. QP sure, GW is done in order to include the QP corrections to the energies. You can the db.PP calculated for GW, then the code will read the 0 frequency screening. You can do that, I was just suggesting to recalculate the screening (em1s) because I saw in your report there was files conflicting with your input.