non-linear response

[claudio]

Dear Marzieh

convergence with the number of bands is quite fast for the Xhi2
instead regarding the number of k-points you have to do different test to get converged results
See for example in the X^3 case

http://www.attaccalite.com/lumen/thg_in_silicon.html

Finally to get the Xhi^2 at zero frequency I advise you to do something like

% NLEnRange
0.10000 | 0.500000 | eV # [NL] Energy range
NLEnSteps = 5

and try to fit X^2 in these 5 points to extrapolate at zero.
If you put 0.00001 simulation time becomes giant!

Finally if you want X^2 at 1064 nm just transform it in eV,
1064 nm = 1.16526 eV

and the do

% NLEnRange
1.16526 | 1.16526 | eV # [NL] Energy range
NLEnSteps = 1

best
Claudio

[Marzieh]

Dear Claudio,

I am trying to calculate nlo with increasing kpoint, nband and frequencies, but I have some errors like these:

slurmstepd: error: Detected 1 oom-kill event(s) in step 69771228.0 cgroup. Some of your processes may have been killed by the cgroup out-of-memory handler.
srun: error: lm3-w024: task 0: Out Of Memory
[mpiexec@lm3-w024.cluster] HYDT_bscu_wait_for_completion (../../tools/bootstrap/utils/bscu_wait.c:151): one of the processes terminated badly; aborting
[mpiexec@lm3-w024.cluster] HYDT_bsci_wait_for_completion (../../tools/bootstrap/src/bsci_wait.c:36): launcher returned error waiting for completion
[mpiexec@lm3-w024.cluster] HYD_pmci_wait_for_completion (../../pm/pmiserv/pmiserv_pmci.c:540): launcher returned error waiting for completion
[mpiexec@lm3-w024.cluster] main (../../ui/mpich/mpiexec.c:1157): process manager error waiting for completion
--
slurmstepd: error: Detected 1 oom-kill event(s) in step 69771228.batch cgroup. Some of your processes may have been killed by the cgroup out-of-memory handler.

or

srun: error: lm3-w076: task 1: Out Of Memory
slurmstepd: error: Detected 4 oom-kill event(s) in step 69771867.0 cgroup. Some of your processes may have been killed by the cgroup out-of-memory handler.
or

yambo_nl:51686 terminated with signal 11 at PC=50ee3f SP=7ffff7d28f70. Backtrace:
yambo_nl[0x50ee3f]
yambo_nl[0x50aad2]
yambo_nl[0x5074a1]
yambo_nl[0x40e2e1]
yambo_nl[0x4097ef]
/usr/lib64/libc.so.6(__libc_start_main+0xf5)[0x2aaab49a8555]
yambo_nl[0x4085e9]

I tried to increase memory and cpu but these errors were not yet fixed. Could you please explain how to set memory and cpu in nlo calculation job script or in yambo.in?
for example for kpoint 9 9 9 and nband 1|150 and NLEnSteps= 120 how to set?
I have attached
Best,
Marzieh

[claudio]

Dear Marzieh

you are using a huge number of bands, usually calculations converge with few bands around the gap.
Start with 3 valence and 3 conduction bands and then try to increase it.
You can have a look to the bands energy to understand the bands range you need

best
Claudio

[marsusi]

Dear Experts,

I am a beginner in Yambo and am trying to work with this code. I have initialized with the first tutorial of hBN. Everything is running well, the Hartree option in making the input file works as the tutorial, but the run level "yambo -x -g n -p p " gives an empty input file.
It would be appreciated if anyone would let me know the problem.

Farah Marsusi,
Department of Physics,
Amirkabir University of Technology.

[Daniele Varsano]

Dear Farah,
this is not the right subform (non linear response).
If you are using yambo v5.0 some command lines are changed.
type yambo -h for a full list of the options. Or you can have a look to the cheatsheet:
http://www.yambo-code.org/wiki/images/c ... et-5.0.pdf

the right command should be:
yambo -x -gw0 ppa -dyson n
or
yambo -x -p ppa -g n

Best,
Daniele

[Daniele Varsano]

Dear Farah,
sorry, but the command line
yambo -x -g n -p p
should be still working.
Have you run a setup first?

Best,
Daniele

[marsusi]

Dear Daniel,

Many thanks for your reply,

Yes! The r_setup file was created before. The command "yambo -o c -k hartree " generates an appropriate input file, but "yambo -x -g n -p p" gives an empty file.

Farah,
Department of Physics,
Amirkabir University of Technology.

[Daniele Varsano]

Dear Farah,

what happen if you just type yambo -x, or yambo -p p?
If they do not work, please post your r_setup file as attachment (you should rename it with a .txt suffix).
Please do not reply here, but open a topic in the right subform Running Yambo-> Initialisation as all this is not related to non-linear response.

Best,
Daniele

[Marzieh]

Dear Claudio,

I have several questions:
For smoothing spectra, Do I need to increase nband or change other parameters?
As you see the spectra are nor smooth.


Could you please tell me some words about these parametrs

ETStpsRt= 200 # Total Energy steps
% EnRngeRt
0.00000 | 10.00000 | eV # Energy range

in ypp.in and their difference with these parametrs

% NLEnRange
1.000000 | 5.000000 | eV # [NL] Energy range
%
NLEnSteps= 5 # [NL] Energy steps

in yambo.in?


I tried to calculate shg with increasing kpoint but I have this error.
[ERROR] STOP signal received while in :[06] NL Post Processing
[ERROR] File ./SAVE//ndb.Nonlinear_fragment_1; Variable NOT DEFINED; NetCDF: Unknown file format

I have attached one of log and yambo.in
(I could not attach any report files here because they are too large . So I will send you them by email).

I think I need to know how to set cpus and memory, right?

Could you please guide me about these issues/

Best,
Marzieh

[claudio]

Dear Marzieh

the fact that the spectra is not smooth can be due to the too small k-point grid,
try to increase it at least up to 9x9x9 or more
or to the integrator, try with

NLintegrator= "CRANKNIC"

Then if the problem of the smooth spectra is solved
and you should increase the number of bands to see if the spectra is converged, for example

27 | 37 | # [NL] Bands

because the valence bands have a very close energy, and so it is probable that they will contribute to the SHG.

best
Claudio