k-points parallelization while doing the dvscf calculation

[sitangshu]

Hi,

Can anyone give me an update on which version of Quantum Espresso does the k-points parallelization while doing the dvscf calculation is satisfied (including spin in nscf). Note this is required in el-ph.
I know upto 6.7Max no k-points parallelization on dvscf could be done.

Regards,
Sitangshu

[lyzhao]

Dear Sitangshu,
Have you tried QE7.2 ?

Best regards.

[sitangshu]

Dear lyzhao,

No I haven't. But I will now try. But I am not sure if the spin-orbit is also included there.

Thanks, I will write again here, if I get this.

Regards,
Sitangshu

[sitangshu]

Hi,

I tried with QE 7.2 with npool 2. It generates elph_dir and all the s.dbph files, but fails to generate gkkp matrices when ypp_ph is executed (note force_symmorphic=.true. warning is a bug in this case):

<---> [06] == Electron-Phonon Databases ==
<---> Inspecting databases ...
<---> ELPH databases: phonon frequencies and eigenvectors | | [000%] --(E) --(X)
<---> ELPH databases: phonon frequencies and eigenvectors |########################################| [100%] --(E) --(X)
<---> ELPH databases: K+Q-grid check | | [000%] --(E) --(X)
<---> ELPH databases: K+Q-grid check |########################################| [100%] --(E) --(X)
<---> [07] == Q-points list in the DBS (iku units) ==
<---> :: Code generator :PWscf
<---> :: DB Kind :dressed
<---> :: Expanded :no
<---> :: Q-points(read) : 7
<---> :: Q-points(written) : 7
<---> :: K-points : 4
<---> :: Bands (all) : 40
<---> :: Bands (loading) : 1 40
<---> :: Branches : 6
<---> :: Uniform sampling :yes
<---> :: Symmetry expanded :no
<---> :: Debye Energy : 193.5768 [meV]
<---> [08] == Q-points list in Yambo (iku units) ==


<---> ELPH databases: transfer and I/O | | [000%] --(E) --(X)
[ERROR] STOP signal received while in[08] == Q-points list in Yambo (iku units) ==
[ERROR] Error expanding GKKP: please set force_symmorphic=.true. in NSCF and nk=1 in DVSCF calculation


Regards,
Sitangshu

[claudio]

Dear Sitangshu

you are right, this is a problem, we are preparing a new interface from QE to Yambo
for the el-ph problem. The interface is almost ready, we are finishing the last tests,
I hope it will be released soon

best
Claudio

[claudio]

Dear Sitangshu

you can parallelize on g-vectors, just doing:

mpirun -np 8 ph.x -nk 1 -inp dvscf.in

and it works
let us know
best
Claudio