Yambo Community Forum

Dear developers,

I am currently running a self consistent run for an FePc molecule. I manage to run the G0W0 and the self consistent GW (updating only in G but not W) up to the third iteration, and after the 4th iteration, the job just stops perhaps due to memory issues (I am attaching the input, report and one of the log files). I was wondering why the memory usage in the self consistent GW seems to be larger than the G0W0. Could you please clarify this?. Also, is it correct to parallelize in the SE step with 4 cpus on qp and 16 for the gbands (I am working with only 1 kpoint, so the parallelization of q is 1)?

Best regards,
Kevin Lizarraga

Dear Kevin

the memory usage should be the same at all iterations.
Looking at your input I found a typos

_and_IO_CPU= "1.8.1.8.1" # [PARALLEL] CPUs for each role
X_and_IO_ROLEs= "q.g.k.c.v" # [PARALLEL] CPUs roles (q,g,k,c,v)

should be

X_and_IO_CPU= "1.8.1.8.1" # [PARALLEL] CPUs for each role
X_and_IO_ROLEs= "q.g.k.c.v" # [PARALLEL] CPUs roles (q,g,k,c,v)

Moreover in order to reduce memory I advice you to parallelize as:

X_and_IO_CPU= "1.4.1.8.2" # [PARALLEL] CPUs for each role
X_and_IO_ROLEs= "q.g.k.c.v" # [PARALLEL] CPUs roles (q,g,k,c,v)

regarding the parallelization of SE part in order to reduce memory you can put all cpus on the bands

SE_CPU= "1.1.64" # [PARALLEL] CPUs for each role
SE_ROLEs= "q.qp.b" # [PARALLEL] CPUs roles (q,qp,b)


best
claudio