nosym=.true. and EPW

[The Pauli Principle]

When adding nosym=.true. to scf.in scf_epw.in and nscf_epw.in, example t04 no longer works. My guess is that epw finds the symmetries (as reported in epw.out) and tries to access files/data which is not there:

########################################################################################################################
# FROM IOTK LIBRARY, VERSION 1.2.0
# UNRECOVERABLE ERROR (ierr=-1)
# ERROR IN: iotk_scan_begin (iotk_scan.f90:182)
# CVS Revision: 1.23
# Tag not found
namel=DYNAMICAL_MAT_.3
########################################################################################################################

All pw.x and ph.x calculations report using no symmetry.

Without nosym=.true. all examples are working.

Is this intended behaviour and are you not allowed to set nosym to true or am I missing an input option for epw?
Note that I would like to use the nosym option.

[sponce]

Hello,

Why do you want to use the nosym option?

You should provide explicitly all the k-points in nscf_epw.in anyway.

Therefore not using symmetry will just make the scf calculation longer for no reason.

In principle, it should work though. Note however that this is not tested (see the espresso/test-suite/epw_* folders) so there might indeed be an issue.

Thanks for reporting it.

Best,

Samuel

[The Pauli Principle]

Thank you for your quick reply

Two reasons:
1) Reuse old dvscf data (these calculations can take very long for larger systems) which was generated before I could make use of symmetry
2) Not all configurations have the same symmetry and it can then be difficult to compare the different modes

Perhaps it is not the k-point symmetry but the q-point symmetry which is the problem?

[sponce]

Hello,

The code definitely expects the use of symmetry for the q-points (dvscf) calculations.

Do you mean that you have the q-point on a grid without symmetries? So for a 6x6x6 q-grid, you did 216 q-point calculations (so you have 216 dvscf files)?

The first part of EPW will try to reconstruct the q-grid from the IBZ to the full zone. If you explicitly computed those, you can hack the code and bypass the first part.

PS: If the dvscf are very old, you should not expect backwards compatibility. If you want to use them, you will need to use an old version of the code.


Best,
Samuel

[The Pauli Principle]

The only change I made was adding nosym=.true. to the pw.x input files (scf.in scf_epw.in nscf_epw.in) in example t04

There are indeed not 216 qpoint calculations.