Inteqp warning message

Submitted by Eeshaa276 on Mon, 05/22/2017 - 09:40


User questions

I get the following warning message as per the tutorials:

WARNING: |Eqp0 - Eo| > finite_difference_spacing. Linear extrapolation for eqp1
may be inaccurate. You should test the validity of eqp1 by rerunning the
calculation with the self energy evaluated at the eqp0 energies. For that,
use the eqp_outer.dat file, created with script and point WFN_outer to
WFN_inner, if you were not already using WFN_outer.

In order to get rid of the error, I follow the steps as given in manual version 1.2

i.e. run
i) ln -s WFN_inner WFN_outer
ii) ln -s eqp1.dat eqp_outer.dat
and add eqp_outer_corrections to sigma.inp. If you get the same eqp1,
the linearization looks good; if not, you can continue iterations like
this to converge to the solution of the Dyson equation.

But I get the following error:
Missing k-points in file eqp_outer.dat

Leading to inaccurate interpolation in the next step. This just a 4 atomic unit cell. I need to switch to a larger system. So, I need to get rid off the error in the first step sigma calculation itself.

Also, WFN_outer is like a dummy file. Instead, is it possible to use actual WFN_outer that can be linked?
Plus, the is obsolete with version-1.2

Submitted by babarker on Mon, 05/22/2017 - 12:29


Linking WFN_inner to WFN_outer is correct. However, you should like eqp0.dat to eqp_outer.dat, not eqp1.dat. Make sure that eqp_outer.dat has matrix elements calculated for all k-points and bands for which new eqp1 matrix elements will be computed.

The python script can still be useful, for instance for generating easy-to-parse files for bare exchange matrix elements, Vxc matrix elements, etc. I will sometimes merge sigma_hp.log files from calculations trivially parallelized over k-points and then run to generate eqp1.dat, instead of concatenating the various smaller eqp1.dat files, as more information can then be extracted from sigma_hp.log, as needed.