The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

> Hello users:
>    I am trying to get familiar with doing electron-phonon calculations
> with density functional perturbation theory using ABINIT.
You're a courageous man, Prasenjit.

> the Eliashberg Tc does not converge, though MacMillan Tc seems to
> converge.
McMillan is calculated straightforwardly from lambda and so on: there is 
nothing to converge. The Eliashberg code is not tested at all: that's why 
it is in the log file. I added it on as a feature, but mainly in view of 
doing anisotropic Eliashberg. It usually converges ok, but may depend 
sensitively on your parameters (not necessarily better for more converged 
results). I use a standard method to iterate the Eliashberg isotropic 
s-wave equations: multiplying a matrix by itself over and over, and trying 
to find T such that the matrix's highest eigenvalue is 1 (Tc) There are 
definitely improvements to be brought to this bit of code.

> 1. Can one trust the el-ph coupling results (*_A2F file, say) when
> Eliashberg Tc has not converged?
Yes, at least for Al we have done extensive tests and compared with 
literature and experiment. Several bugs have been found and fixed in the 
latest versions of the code, and affect the linewidths and a2f function. 
Please keep up to date for all of this - the code is still under 
development.

> 2. Which parameters should I focus on if I need to converge Eliashberg Tc?
Absolutely no idea :) I suspect it's more a matter of putting in a more 
robust algorithm for the Eliashberg case: If you're interested it's in the 
file 07ddb/eliashberg_1d.F90

If the Tc is converging (ie the column of numbers in the log is going to 
1) you may just want to increase the number of iterations of the equations 
(hard coded) but this shouldn't be necessary.

Matthieu

--
================================================================
Dr. Matthieu Verstraete                 mailto:mjv500@york.ac.uk
Dept. of Physics, University of York,     tel: +44 1904 43 22 08
Heslington, YO10 5DD York, United Kingdom fax: +44 1904 43 22 14