Page 1 of 1

NaN in GW calculation

Posted: Mon Dec 05, 2022 10:24 am
by liuyang_phy
Dear developers,

I'm using Yambo 5.1.1 to perform GW0 calculations on my 2D system. Everything is ok and no errors presented in my log files.
But, the values of (E-Eo) and Sc(Eo) are NaN in my o-xx.qp files for each K-point and each band. Moreover, i found this error appears
only when i do parallel computing on mutiple nodes. If i just run it on one node, the qp corrections seems normal. Even stranger,
sometimes the parallel computing may turns to be normal when i changes some parameters of my input file. I attached my input file.
What's the problem of my yambo code and How to fix it ? Thank you for your generous help!

Best,
Liu Yang
PhD student
Hua zhong university of science and technology, China

Re: NaN in GW calculation

Posted: Wed Dec 07, 2022 11:53 am
by Daniele Varsano
Dear Liu,
not easy to understand such a behaviour, probably a miscompilation in parallel.

can you try to run in parallel by assigning the parallel structure by hand e.g.:

X_and_IO_CPU= "1 1 1 36 2" # [PARALLEL] CPUs for each role
X_and_IO_ROLEs= "k g q c v" # [PARALLEL] CPUs roles (q,g,k,c,v)
DIP_CPU= "1 36 2" # [PARALLEL] CPUs for each role
DIP_ROLEs= "k c v"
SE_CPU= "1 2 36" # [PARALLEL] CPUs for each role
SE_ROLEs= "q qp b" # [PARALLEL] CPUs roles (q,qp,b)

Before rerunning the calculation, please delete all the previous databases.
Best,
Daniele