Page 1 of 1

ypp_nl post-processing real-time simulation error

Posted: Thu Jul 27, 2023 10:11 am
by AmberLee123
Hi, I meet an error as ypp_nl post-processing real-time simulation based on COLL_SEX(the TD-SEX_nl finished correctly).

The error reports

Code: Select all

[06] NL Post Processing
 =======================


 [ERROR] STOP signal received while in[06] NL Post Processing

 [ERROR]LINEAR ALGEBRA driver [GETRF (SI)]WS%v_cmplxspace failed
It is very confusing... because the same ypp_nl works for nonlinear calculation based on IPA.

Please see the attachments for detail. Many thanks~

Re: ypp_nl post-processing real-time simulation error

Posted: Thu Jul 27, 2023 10:13 am
by AmberLee123
attachment2

Re: ypp_nl post-processing real-time simulation error

Posted: Thu Jul 27, 2023 10:14 am
by AmberLee123
attachment 3

Re: ypp_nl post-processing real-time simulation error

Posted: Fri Jul 28, 2023 10:23 am
by Davide Sangalli
Dear Guoyu Yang,
this might be due to an issue in the real-time simulaton with yambo_nl, e.g. too long time step, which makes the simulation numerically unstable.

Among the file attached, I do not see the o-TD-SEX_nl_2.* files.
If you plot the polarization, does it look reasonable ?

Best,
D.

Re: ypp_nl post-processing real-time simulation error

Posted: Sat Aug 05, 2023 8:50 am
by AmberLee123
The polarization file 1-3 out of 7

Re: ypp_nl post-processing real-time simulation error

Posted: Sat Aug 05, 2023 8:51 am
by AmberLee123
The polarization file 4-6 out of 7

Re: ypp_nl post-processing real-time simulation error

Posted: Sat Aug 05, 2023 9:15 am
by AmberLee123
The polarization file No. 7 out of 7.

Sorry for the later reply.

The plotting of polarization seems to be fuzzier than the tutorial.

And the polarization really gets worse as the simulation time goes on.
The bad news is the problem persists when I change the simulation time range.

Re: ypp_nl post-processing real-time simulation error

Posted: Sat Aug 05, 2023 10:01 am
by AmberLee123
I make the time step (NLstep) smaller from 0.0100 fs (10.000000 as) to 1.0 as , and redo the calculation from TD_nl.
Hoping that will make the polarization smoother

%%%%%%%%%%%%%%%%%%%

ypp_nl works perfectly on NLstep 1.0 as
The problem has been solved

Re: ypp_nl post-processing real-time simulation error

Posted: Thu Aug 24, 2023 9:37 am
by Daniele Varsano
Dear Guoyu,

great the problem is solved!

Daniele