Exciton dispersion

[DavidPolito93]

Dear Developers,

I am trying to extract the exciton dispersion. I have seen that in the tutorial it is possible to do so via:

%BSEQptR
iq_start | iq_end | # [BSK] Transferred momenta range
%

My problem is the following: I have 51 k-points in the IBZ, so in the code I wrote

%BSEQptR
1 | 51 | # [BSK] Transferred momenta range
%

However the code works well up to the point 33 and then stops without any message.

I thought it might be related to some memory issue (?). So my question is: is it possible to reduce the number of points in the computation of the exciton dispersion?

For example, is there a way to compute only 26 homogeneously distributed points in the IBZ instead of all 51 and then interpolate them for obtaining a dispersion relation?

Best,

Davide Romanin
-----------------------------------------------------
Post-doctoral fellow
Institut des NanoSciences de Paris
Sorbonne Université, CNRS
4 place Jussieu,
75252 PARIS Cedex 05

[claudio]

Dear Davide

I don't know why there is a problem in the BSE if you calculate many q-points.... we will investigate.

Then two comments on your calculations:
1) you are not obliged to calculate the BSE for all q-points altogether.
You can do many jobs, one for each q-point if you want.

...
%BSEQptR
1 | 1 | # [BSK] Transferred momenta range
%
...

and then another
...
%BSEQptR
2 | 2 | # [BSK] Transferred momenta range
%
...

and so on.
Yambo saves the results on different files, and then you can use them for the interpolation.

2) unfortunately for the moment you have to calculate BSE on all q-points in the IBZ in order to interpolate it

best
Claudio

[DavidPolito93]

Dear Claudio,

Thank you for your answer! So the best strategy in my case would be to perform many BSE computations at finite q.

Then, if I wanted to plot a dispersion relation based on those computations (i.e. without interpolation) where can I find the energy associated to a particular state?
Do I look at the peak in the absorbance or do I have to do ypp -e s ?

Best,
Davide

[Daniele Varsano]

Dear Davide,
you can run ypp:


ypp -e s -b #iq

Best,
Daniele

[DavidPolito93]

Dear all,

I tried to perform a BSE computation at q = 2 alone.
However, the program stops without any warning during the creation of the BSE kernel.
I am attaching the output.

What is strange is that if I do a continuous computation from 1 to 2 it doesn't stop and it ends normally.

Best,

Davide

[DavidPolito93]

Sorry,

You probably need also the input.

Best,

Davide

[Daniele Varsano]

Dear Davide,
thanks for reporting, do you have any error message from the machine in the submission output?

Daniele

[DavidPolito93]

Dear Daniele,

The only error message that I get from the machine is:

===================================================================================
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= PID 45677 RUNNING AT r1i2n1.ib0.xa.dsi.upmc.fr
= EXIT CODE: 9
= CLEANING UP REMAINING PROCESSES
= YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
===================================================================================
Intel(R) MPI Library troubleshooting guide:
https://software.intel.com/node/561764
===================================================================================

By the way, I am using Yambo-5.0.0.
Should I try with the latest?

Best,

Davide

[Daniele Varsano]

Dear Davide,
yes please try to repeat your calculation with the latest release. If the problem persists we will have to reproduce your error, so it is better to work with the latest release.

Many thanks,
Daniele

[DavidPolito93]

Dear Daniele,

I performed the computation with the 5.0.2 and it broke down again when generating the kernel

Best,

Davide