The ABINIT Users Mailing List ( CLOSED )

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

> For example, in the tutorial telphon (with obviously a crude set
> of parameters), lambda for bulk Al works out to be ~9.6. By increasing
> kpints, q-points and ecut it turns out to be ~0.3 while the expt number
> is 0.44.
> Can I hope to do better?
Yes - should be able to. I got to 0.45 or so.

> Which other parameters are important for this?
well, how high in kpt convergence did you go? The pseudopotential can be 
crucial (no one has studied this in detail, and many others use all- 
electron methods). The unit cell parameter can change the phonons 
considerably as well. To get good results Savrasov does something ugly and 
calculates some things from experimental lattice parameters and others 
relaxed. Formally you should use everything at the relaxed lattice 
parameter (ie. you should start from the fully relaxed structure), but 
many people use this kind of trick if they know what the physical picture 
should be.

> How best to set the parameter mustar?
mustar is usually fit to get the experimental Tc. For Al in my case I used 
0.136 (a bit higher than others). Otherwise many materials have a value 
between 0.12 - 0.17, but this is horribly qualitative. The treatment of 
the Coulomb interaction in Eliashberg theory is very crude, but seems to 
work in many cases beyond the expected regime (fcc metals with low el-ph 
coupling and "reasonable" DOS at E_F).

bye

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