Page 2 of 2

Re: GW+BSE optics

Posted: Wed Feb 06, 2019 11:32 am
by burkzdemir
Dear Daniele,

I runned the same calculation "yambo -o b" with yambo 4.2.4 and it finishes without a problem. But the problem is my installation of 4.2.4 is not parallel and I need parallel yambo. I installed yambo 4.3.1 with parallel successfully and it does the GW calculation without a problem but it returns segmentation error for optics calculation. I suspect something happened in the compilation. Do you have any idea? I compiled yambo 4.3.1. with ./configure FC=ifort F77=ifort

Re: GW+BSE optics

Posted: Wed Feb 06, 2019 11:41 am
by Daniele Varsano
Dear Burak,

I cannot say anything with so few information. If you want, post your confg.log file and I will have a look.
By the way:
1. did you try to reduce your BSE bands.?
2. did you try to do an optic calculation from the tutorial?

Best,
Daniele

Re: GW+BSE optics

Posted: Thu Feb 07, 2019 10:21 am
by burkzdemir
I managed to install yambo 4.2.4 with parallel. I runned GW without a problem. However at BSE optics calculation the code stopped at Kernel loop. What could be the problem?

Best,
Burak

Re: GW+BSE optics

Posted: Thu Feb 07, 2019 10:23 am
by Daniele Varsano
Dear Burak,
1. did you try to reduce your BSE bands.?
2. did you try to do an optic calculation from the tutorial?

Daniele

Re: GW+BSE optics

Posted: Thu Feb 07, 2019 11:23 am
by burkzdemir
Dear Daniele,

I reduced the BSE bands and it worked now. Thank you! But what happens if I need more BSE bands and the code is stopping? What can I do then?

Best,
Burak

Re: GW+BSE optics

Posted: Thu Feb 07, 2019 2:27 pm
by Daniele Varsano
Dear Burak,
usually the needed BSE bands are not many, in any case you can use a parallelization strategy aimed to distributed the memory among progessor.
This is achieved by setting:
PAR_def_mode= "memory"

Best,
Daniele