GW correction in generic k-point (ypp) - Segmentation Fault

[jmmorbec]

Hello.

I am trying to calculate GW correction in a generic k-point as explained in the tutorial (http://www.yambo-code.org/input_file/ypp/ypp_kk.php).

1) I did a GW calculation in a regular k-point grid (please, see o.qp and r_setup);
2) I did "ypp -k k" and got ypp.in; then I edited ypp.in to include the new k-point (see ypp.in);
3) I runned "ypp" and I got a new set of k-points (see new-kpoints.txt) - Here there are 28 k-points;
4) I copied the new set of k-points for the nscf.in, and I runned a scf calculation with regular k-points (3x3x3 000) and nscf calculation with new set of k-points (28 k-ponits) - see Ta3N5-scf.in and Ta3N5-nscf.in in pwscf folder - In Ta3N5-nscf.out there are 44 k-points, although I just included 28 k-points in Ta3N5-nscf.in - I don't understand what is the problem here;
5) I generated yambo database using "p2y -N";
6) I runned the setup using "yambo -i -V kpt", and then I changed the IkXLim from 44 to 10.

setup # [R INI] Initialization
% IkSigLim
1 | 44 | # [KPT] QP K-points indices range
%
IkXLim= 44 # [KPT] X grid last k-point index
% QptCoord # [KPT] [iku] Q-points coordinates (compatibility)
%

7) I runned the setup (yambo -N);
8) I copied ndb.pp from SAVE of my previous calculation to the new SAVE;
9) I copied the input (yambo.in) from my previous GW calculation to the current folder (see yambo.in);
10) In yambo.in I changed

%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 27| 57|72 |
%
%QPerange # [GW] QP generalized Kpoint/Energy indices
1| 27|10.0|-10.0|

to

%QPkrange # [GW] QP generalized Kpoint/Band indices
28| 28| 57|72 |
%
%QPerange # [GW] QP generalized Kpoint/Energy indices
28| 28|10.0|-10.0|

11) Finally, I runned yambo and I got

" <44s> EXS^M <44s> EXS | | [000%] --(E) --(X)_pmiu_daemon(SIGCHLD): [NID 01102] [c4-1c0s7n0] [Thu Feb 20 01:18:30 2014] PE RANK 0 exit signal Segmentation fault"

(see yambo.out).

Could anyone help me with this, please? I have no idea what I am doing wrong.

I will really appreciate any help.

[jmmorbec]

I forgot to attached the files. Here they are!

[Daniele Varsano]

Dear Juliana,
well, not easy to spot the problem. The error you get could be related to the fact that your q-grid is not uniform, but it is just a possibility.
May be you can look to this information looking at the r_setup of the new run (check: X grid is uniform ).
Anyway the fact the qe generates more k-points than you asked it is due to the fact the symmetry of the k-point grid supplied is lower than the full symmetry of the lattice (see Variables: xk_x, xk_y, xk_z, wk in the qe manual), and I think that this should not occur. Please check the correctness of the input/output units:

Code: Select all

cooIn= "iku"                 # Points coordinates (in) cc/rlu/iku/alat
cooOut= "rlu"                # Points coordinates (out) cc/rlu/iku/alat

May be the k-point grid you provide to the qe, it is not the right one.

Unfortunately I cannot help you more on this issue, as I never used this utility, may be someone else can gives you better ideas to solve the problem.
Anyway, the file yambo.out it is ok, I think you posted the output calculation on the regular grid, may be could be useful if you post the output and report calculations on the new grids (setup etc.).

Best,
Daniele

[jmmorbec]

Thank you very much Daniele for your reply. I will check those units.