Yambo Community Forum

Dear all,

I am trying to do BSE calculations with spin-orbit.
When I run yambo I get this message:

<---> [01] Files & I/O Directories
<---> [02] CORE Variables Setup
<---> [02.01] Unit cells
<---> [02.02] Symmetries
<---> [WARNING]Spatial Inversion not found among the given symmetry list
<---> [02.03] RL shells
<---> Shells finder |####################| [100%] --(E) --(X)
<01s> [02.04] K-grid lattice
<01s> [02.05] Energies [ev] & Occupations
<01s> [03] Transferred momenta grid
<01s> [RL indx] X grid is not uniform. Gamma point only.
<01s> [04] Game Over & Game summary

I am using a 6x6x1 k-grid but yambo is able of reading just the Gamma point.

Has anyone found a similar problem?

Thanks,

Alejandro.

Dear alejandro,
indeed yambo recognizes that your k-points grid is 6x6x1

[02.04] K-grid lattice
======================
Compatible Grid is 2D
B1 [rlu]=0.167 -.793E-09 0.00
B2 =-.7451E-8 0.1667 0.000
Grid dimensions : 6 6
K lattice UC volume [au]: 0.0076

However according to yambo the k-points grid is not uniform, thus it can perform calculations only at the gamma point of the transferred momenta grid (q=k-k'). This is the grid derived from the k-points grid.
[03] Transferred momenta grid
=============================
[RL indx] X grid is not uniform. Gamma point only.

To understand why this happens in your case, we would need the input files and the pseudo-potentials you used to generate the ground-state and then the KSS with abinit. In this way we can try to reproduce your problem.

Regards,
Davide

Dear Davide,

you have here the files to generate the kss file.

Best,
Alejandro.

Dear Alejandro,
from what I see the grid you are using is indeed not uniform.

Try to work with shiftk 0.0 0.0 0.5, which should provide a uniform grid in case of hexagonal systems.
http://www.abinit.org/documentation/hel ... tml#shiftk

Davide

I have tried this but it only changes the ERROR to that:

[ERROR] STOP signal received while in :[03] Transferred momenta grid
[ERROR]Error in Q-grid search [qindx_X]

At least, now there is a list o k-points, but still yambo crashes...

Ok, I think I found your problem, but I need to prepare a patch for that.

As a workaround for the moment, you can choose one of the two following options.
1) run abinit with the option nspden=4. Indeed the experimental interface a2y for nspinor=2 was implemented only for this case.
2) compile a2y replacing in the source the file I'll send you separately.
(Remember that in this case a2y would not correctly work for nspden=4!)

Davide

P.S.: Please also forget my previous post. As your system is a 2D system the shift 0.0 0.0 0.0 is the correct one.

Hi,
the new patch works well (you can see in the r_setup.txt file) after doing:
a2y -F kss-file
yambo_sc -N

The problem is to generate the yambo.in file. If I do (for example)
yambo_sc -b -o b -y h -V all
nothing happens, the yambo.in file doesn't contains any variable to perform a BSE calculation... (see yambo.txt)

Dear Alejandro,
what happens is that your system now includes the time_reversal symmetry.
As I told you the n_spinor=2 case was implemented only for the case of magnetic systems and, to avoid problems, by default all the run-levels are "switched off" when time-reversal symmetry is off.

Thus there are two options you can follow:
1 - go in the file "src/interfaces/barriers.F" and comment the line number 103, i.e. replace
if(i_time_rev==1.and.n_spinor>1) call switch_off_runlevel('all',on_name=trim(always_runlevels))
with
! if(i_time_rev==1.and.n_spinor>1) call switch_off_runlevel('all',on_name=trim(always_runlevels))
and then compile yambo again

2 - compile ypp (make ypp) and use it to remove the time-reversal symmetry.
ypp -n -F ypp_input.in
uncomment the RmTimeRev flag (i.e. remove the # in front of it)
run ypp -F ypp_input.in
it should generate a new SAVE directory without the time_reversal symmetry.

In both case you should be able to perform standard calculations. It would be a good idea to check that the two give the same results.
Also check that you are using the last yambo revision doing "svn up"

Davide

Thanks for the help. I am testing now but it seems it works.