Yambo Community Forum

Dear all,
I get an error importing the pw.x output (ver 7.0) into yambo (ver 5.1.1.) for a 1D system treated with spin polarization and with a transversal electric field (a sawtooth_potential). The premises are that without the electric field it behaves correctly, with the electric field and a spin unpolarized approach it behaves correctly as well.
I get an error in the qexsd_read_planewaves routine while reading the RL vectors

<---> K-points mesh... done
<---> RL vectors...
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Error in routine qexsd_read_planewaves (39):
fmt error II
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

stopping ...
Do you have suggestions to cure the error?
Thank you
Giuseppe

Giuseppe Zollo
Sapienza University of Rome
Dept. SBAI

Dear Giuseppe,

no idea on what is going on in this case, can you please post your nscf input file (along with pseudos if not easily findable) so that we can try to reproduce the error.
Ideally, can you provide an example with minimal parameters (e.g. energy cutoff, k points etc) showing the error?

Many thanks,

Daniele

Dear Daniele, thank you very much for your assistance.
I attach the scf, the nscf and the pseudo files.
If you need I could also attach the xml

here the hydrogen one

Dear Giuseppe,
up to now, I've not been able to reproduce your error,

I used qe-7.1 and yambo-5.1 and modified your input in order to have it run in a few minutes:
*I removed ecutrho as it is not needed for NC pseudo ( I suggest you to do the same).
*Lowered ecutwfc to a ridiculous 5 Ry
*Reduced k points to 1 1 10.

Note, I just run scf, as the nscf was not provided (what's named nscf it is an output and not an input).
So, most probably the problem arise when considered a large number of gvectors.

Nevertheless, as it is a just few minutes test, can you try to reproduce the error with the parameters above just to be sure that is not something depending on the releases (you can run the p2y on top of the scf)?
If it works, can you adjust the size of the problem in order to have a test showing the error but easy/fast to run?

Many thanks,

Daniele

Dear Daniele,
thank you for your answer.
Well, that's very strange. I've employed exactly the same parameters (including ecutrho) for both the spin-polarized and the unpolarized+static electric field and it worked without problems.
Now I've followed your suggestion for the spin-polarized+electric field case: I've removed ecutrho and I've found-out that it woks even with larger ecutwfc and the same kpoint grid I was employing.
I do not understand the reason (not sure is the memory allocation because it should have shown the same problem for the spin-polarized case with zero electric field) but the point is that without ecutrho is works.
Many thanks for your suggestion.
Giuseppe