E-ph mobility calculation error in optdriver

Running Abinit Response calculations
1

Hi!
I am performing a carrier mobility calculation considering eph interaction. I am following this tutorial eph_mobilities . I want to perform these calculations using PAW method. So far the calculations went well, but, in the section where we perform calculation for mobility, it’s failing showing this error below -

chkint_prt: ERROR -
Context: the value of the variable usepaw is 1.
The value of the input variable optdriver is 7, while it must be
equal to one of the following: 0 1 3 4 99 8 5 9
Action: you should change the input variables optdriver or usepaw.

— !COMMENT
src_file: m_chkinp.F90
src_line: 4055
message: |
dipdip can be set to 0 in case of metals whereas dipdip 1 should be used in polar materials.

But, I want to go on with this PAW method, so, usepaw will be 1 but with that optdriver 7 is not fitting well. If it doesn’t fit in How am I supposed to calculate Mobilities?
Also, I haven’t use this tag dipdip , why is this referring to dipdip tag when this tag is used in anaddb calculations instead of abinit calculations?

Please guide me how am I supposed to resolve this error?

2

The EPH driver does not work with PAW yet. Almost though, it will come soon, and should enable LDA+U+EPC as well.

The message about dipdip is a comment, unrelated to the PAW issue. It is also set by default to 1, but is not needed in metals. You can set it to 0 explicitly. If you calculate mobilities, normally they are for lightly (or un-) doped semiconductors, so dipdip makes sense

best

Matthieu

3

Would you like to suggest any workaround for the PAW issue to keep moving on?

4

  1. wait

  2. some things can be run with the old work flow using GKK files made with PAW (see legacy tutorials), but not mobility, only LOVA transport

  3. use ncpp potentials without the +U

5

Thanks!

Apparently, the system I’m dealing with it will require a hubbard correction to capture a decent ground state. Although if I perform ZPR calculations on it, I’d be able to get my required observable. Could you please tell me if ZPR is implemented with PAW method or ncpp+U?

Thanks

6

No - not yet. I am also waiting impatiently for the EPC + U possibility