2D kpoint grid being interpreted as 3D

[arb83@cam.ac.uk]

Dear Yambo Community Forum

I am experiencing some errors when I initialise yambo in specific cases of 2D systems. Specifically, the 'r_setup' file interprets my kpoint grid as 3D, while in fact it is 2D. As an example, I use the unit cell (defined in quantum espresso):
"
CELL_PARAMETERS angstrom
7.6779370 -0.0001288 -0.1924281
0.4494135 6.0244098 -0.0270052
-6.9265044 -2.2903289 33.8392186
"
and kpoint grid
"
K_POINTS automatic
5 5 1 0 0 0
".

When I try to initialise yambo, the 'K'grid lattice' section states

"
[02.04] K-grid lattice
======================

Compatible Grid is 3D
B1 [rlu]=-.2000 -.3070E-8 0.1490E-7
B2 =-.262E-09 0.200 0.00
B3 =-0.400000 -0.400000 1.000000
Grid dimensions : 5 5 1
K lattice UC volume [au]:0.9444E-3
"

while the compatible grid should be a 2D grid. Is there a way of ensuring that yambo interprets a k grid as 2D? I am receiving error messages later in the initialisation sequence, but I believe this is the route cause of these errors.

With best wishes,

Alan

[Daniele Varsano]

Dear Alan,
you can add this variable in your set up input file:

Code: Select all

setup
NoDiagSC 

rerun yambo and see if it solves the problem?
Actually this extra-variable appears in the input by adding versbosity as

Code: Select all

>yambo -i -V gen

and you need to uncomment it.

If this does not solve the problem please post your report file.
Best,
Daniele

[arb83@cam.ac.uk]

Dear Daniele

Many thanks for the suggestion. Unfortunately this didn't seem to solve the problem (the line you mentioned wasn't commented out, but rather wasn't present, so I added it as an extra line). I'm attaching the r_setup file I obtain. I appreciate there are subsequent errors in the file, but believe they all stem from the kpoints being interpreted incorrectly. If I do an equivalent calculation with (for example) a 5 5 2 or 6 6 2 kpoint grid then the initialisation runs fine.

With best wishes,

Alan

[Daniele Varsano]

Dear Alan,
in order to spot the problem, we would need to reproduce your error.
Can you provide us your sci/nscf input file together with the used pseudo potential?

Best,

Daniele

[arb83@cam.ac.uk]

Dear Daniele

Of course, please find attached. I found I could reproduce the error if I replaced all the atoms in my unit cell with simply a PbI2 molecule (but kept unit cell parameters fixed), so I attach the PbI2 model as it's easier to run.

I have added '.txt' to the end of all uploaded files as otherwise I was unable to upload them. I am using yambo GPL version 4.4.1.

Best wishes,

Alan

[arb83@cam.ac.uk]

Final pseudopotential here...

[Daniele Varsano]

Dear Alan,
I was not able to reproduce your error, e.g. a simple setup does not work as you experienced:

Code: Select all

 [WARNING][RL indx] 2 equivalent points in the rlu grid found

but the problem is solved by using the NoDiagSC variable in the input setup.

Note: I have tested the 4.5: most probably the NoDiagSC algorithm was not present in the 4.4.1, so it does not take effects.
My advise is to upgrade your source to the latest release and run the following input file:

Code: Select all

setup                            # [R] Initialization
Nelectro= 48.00000               # Electrons number
ElecTemp= 0.000000         eV    # Electronic Temperature
BoseTemp=-1.000000         eV    # Bosonic Temperature
OccTresh= 0.100000E-4            # Occupation treshold (metallic bands)
NoDiagSC                      # New setup for non-diagonal supercells

Note, the grid is still recognised as 3D (you have sampling in all the directions from the qe calculations), anyway this is not a problem at all.

Best,
Daniele

[arb83@cam.ac.uk]

Dear Daniele

That seems to have worked, thank you. Could I ask what the line NoDiagSC does please? Does it introduce extra errors in the calculation?

Best wishes,

Alan

[Daniele Varsano]

Dear Alan,
it is a different algorithm to identify the k/q grid.
It has been developed by @claudio so he may provide you more details.
No extra errors, it just recognise what the default algorithm do not.

Best,
Daniele

[claudio]

Dear Alan

the NoDiagSC option activates an alternative algorithm to identify k and q points.
It does not introduce any additional errors in the calculations, so use it without problems.
It is not the default in Yambo, because at present it is not compatible with multiple-shifted grids (not your case).
In the future, I will generalize it and make it default algorithm.

best
Claudio