Yambo run error

[will]

Hi all,
I try to test Yambo for calculating optic properties in parallelization machine.However, it is always aborted.The following is the snapshot of error.
More pieces of information are in my attachments.
Any help will be appreciated!
Best wishes,
Will

[Daniele Varsano]

Dear Will,
looking at your file it looks that yambo is not running in parallel, but in serial in several cpus.
Please note:

yambo -V par -F rpa.no.lf.in

The "-V par" option has to be set when building your input file and *NOT* when running yambo.
e.g.

Code: Select all

yambo -o c -V par 

In this way parallel variable appears and you can set them, otherwise yambo will use the default. In any case the instruction "yambo -V par -F rpa.no.lf.in" us not correct.

Try to run yambo simply as:

Code: Select all

yambo -F rpa.no.lf.in

or eventually setting the parallel variables in input.
If it does not work please attach also the report files (the files r_*) as it contains several information that can be useful to spot the problem, and
the config.log file in order to see if the parallel compilation was set set up properly.

Best,

Daniele

[will]

Dear Daniele
Thank you for your reply.I run Yambo according to your suggestion.1)don't set the parallel variables in input files;2)manually set parallel variables in input files.Both run erroneously...








Best wishes,
Will

[Daniele Varsano]

Dear Will,

at least now Yambo is running in parallel:

Code: Select all

 * CPU-Threads     :12(CPU)-1(threads)-1(threads@X)-1(threads@DIP)-1(threads@SE)-1(threads@RT)-1(threads@K)
 * CPU-Threads     :X_all_q(environment)-1 12 1 1(CPUs)-q k c v(ROLEs)
 * CPU-Threads     :SE(environment)-1 12 1(CPUs)-q qp b(ROLEs)
 * MPI CPU         : 12
 * THREADS    (max): 1
 * THREADS TOT(max): 12

The problem now is that you are not setting the right variables. As I wrote in the previous post add the flag "-V par" when building your input (or -V all), yambo will provide the appropriate variable for the runlevel. In this case you are calculating the q->o limit response function (optics) and you need the X_q_0_CPU variable. Yambo try to use the default but probably it fails as only very few conduction bands are present.

In your input for instance you have SE_CPU which is needed for the self energy evaluation and totally useless in your run. Next putting X_Threads=0 is quite dangerous, yambo set it to 1, in general anyway leave at least 1 thread for each runlevel.

I had a look to the report, you are treating a quite peculiar system, a crystal of 2 H atoms with a lot of k points:
please note that the system you are considering is metallic:

Code: Select all

  States summary         : Full        Metallic    Empty
                           0000-0000   0001-0002   0003-0012

May not the simplest system to start, I suggest you to practice a bit with simpler systems you can find in the tutorial yambo webpage.
Best,

Daniele