Page 1 of 1
Random Integration Method (RIM)
Posted: Tue May 13, 2014 4:11 am
by jmmorbec
Hello.
Please, I would like to know if the Random Integration Method (ypp -r) works with Haydock solver.
Best wishes,
Juliana
Re: Random Integration Method (RIM)
Posted: Tue May 13, 2014 8:33 am
by Daniele Varsano
Dear Juliana,
the RIM is meant for calculating the coulomb integral in the Brillouin zone in the Self Energy calculations. The Haydock method is used to diagonalize the excitonic matrix. Thus, the two procedures are not related.
Best,
Daniele
Re: Random Integration Method (RIM)
Posted: Tue May 13, 2014 2:30 pm
by jmmorbec
Dear Daniele.
Thanks for your reply.
I am following the tutorial
http://www.yambo-code.org/input_file/ypp/ypp_rim.php to speed-up the convergence of the dielectric constant, and it says "The same approach works also for the Bethe-Salpeter equation using the inversion solver -y i". As I am doing BSE calculation using Haydock method (y -h) I was wondering if ypp-rim would work in my calculations.
Best wishes,
Juliana
Re: Random Integration Method (RIM)
Posted: Tue May 13, 2014 2:44 pm
by Daniele Varsano
Dear Juliana,
sorry for having misunderstood, I was thinking at the RIM applied to the q-point BZ integration used in the self energy calculation. I'm not expert in the RIM for the BSE, anyway looking at this
paper, it seems to me that it can be applied solving the BSE with the inversion technique only. In any case, may be other more expert than me, can add something or correct me.
Best,
Daniele
Re: Random Integration Method (RIM)
Posted: Tue May 13, 2014 2:55 pm
by jmmorbec
Thanks for your help, Daniele.
I will check this paper.
Re: Random Integration Method (RIM)
Posted: Tue May 13, 2014 2:59 pm
by Davide Sangalli
Dear Juliana,
yes, the RIM method, which indeed should better be called "double-grid method" for the sampling of the kpt in the BZ, is presently implemented only with the inversione solver.
Best regards,
Davide
Re: Random Integration Method (RIM)
Posted: Tue May 13, 2014 3:05 pm
by jmmorbec
Thanks, Davide.
I am doing BSE calculations, and I noticed the convergence is very slow with respect to the k-point mesh. I can not afford a denser grid (time and memory issues), so I was wondering if the RIM method could help me with this.
Re: Random Integration Method (RIM)
Posted: Thu May 15, 2014 4:51 am
by jmmorbec
Hello.
I am doing E-Rim to speed-up the convergence of my BSE calculations. I am following the instructions of
http://www.yambo-code.org/input_file/ypp/ypp_rim.php. However, I have a doubt about item 7 (calculate again the dielectric constant, using the E-RIM to average the denominators). Can I use the same yambo.in of my previous BSE calculation (with small grid)? How does yambo know that I am using E-Rim? Do I need to include any additional flag in yambo.in?
Thank you!
Juliana
Re: Random Integration Method (RIM)
Posted: Thu May 15, 2014 6:23 am
by Daniele Varsano
Dear Juliana,
please have a look at this post:
viewtopic.php?f=15&t=826
Best,
Daniele
Re: Random Integration Method (RIM)
Posted: Thu May 15, 2014 2:17 pm
by jmmorbec
Thanks, Daniele.