Page 1 of 1
HXC collisions not found
Posted: Sun Mar 17, 2024 4:05 am
by sdwang
Dear all,
I got a problem when executing the yambo_nl (v5.2.1) as:
Code: Select all
<01m-34s> [WARNING] HXC collisions not found. Potential is computed on the fly.
[ERROR] STOP signal received while in[05.01] Messaging
[ERROR] HF / SEX need Collisions. Cannot be computed on the fly.
I have already calculated the collision in the previous step and ndb.COLLISIONS_HXC_header and ndb.COLLISIONS_HXC are in the SAVE fold.
Attached are the input, log and report files.
Thanks!
Best
Shudong
Re: HXC collisions not found
Posted: Tue Mar 19, 2024 1:27 pm
by sdwang
Dear all,
Can you provide some suggestions?
Thanks!
SD
Re: HXC collisions not found
Posted: Tue Mar 19, 2024 4:40 pm
by Davide Sangalli
Dear Shudong,
I do not see any file attached.
Usually this happens because there is some inconsistency between the input variables in the generation of the collisions and the input variables for the real time simulation.
Best,
D.
Re: HXC collisions not found
Posted: Wed Mar 20, 2024 3:00 am
by sdwang
Dear Davide,
Thanks for your suggestions and I have already fixed the problem through unifying the parameters of parallel.
I have another question about the option in the probe/pump input file:
#UseDipoles
When I perform the example of hBN, this option has been neglected and everything is OK. But for production calculation of my own system, I have to use this option or else there exist segmentation fault and Yambo will stop. I compared this with hBN but can not find the difference between with and without "Usedipoles" except there is no ndb.Overlap file in the output directory with usedipoles.
Thanks!
Shudong
Re: HXC collisions not found
Posted: Wed Mar 20, 2024 6:35 pm
by Davide Sangalli
The key
mkes the code switching from the berry phase formulation for the polarization and the coupling with the external field, to a formulation based on the dipoles.
The latter is in principle correct only to linear order in the external field. So, if you are looking into some non linear effect, you might get some error in your results.
Remember that any pump and probe signal is a non linear process in the sum of pump plus probe.
However, it is not easy to predict a priory how big such error would be. In some cases it is negligible, in others it is important.
To give you an example, in this work
https://journals.aps.org/prb/abstract/1 ... 107.205203 I could not get any signal in the transient spectrum between 0 and 3 eV (e.g. in the region of exciton - exciton transitions) by using the coupling with the fields based on the dipoles. On the other hand, in this work
https://journals.aps.org/prmaterials/ab ... s.5.083803 I could hardly see any difference in the transient signal. In the two works I'm doing simulations on the same material, just looking into the transient signal at very (!) different energy ranges.
Hope it helps.
Best,
D.
Re: HXC collisions not found
Posted: Thu Mar 21, 2024 2:07 am
by sdwang
Dear Davide,
Thanks for your advices! It is absolutely helpful to me.
Best
SD