The ABINIT Users Mailing List ( CLOSED )

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

>    In PRB 53,R7575, the scatting matrix element of e-ph Hamiltonian 
> seems have no relation to the phonon amplitude?  In the article, 
> equation (1), the Hamiltonian in the middle is just the phonon 
> polarization vector times the gradient of the self-consistent potential. 
> This is quite different compared with the equations in Mahan's "Many 
> particle Physics".
the perturbed hamiltonian should be dotted into the phonon displacement 
vector. The difference with the polarization vector is the 1/sqrt(M) which 
is present in eq (1). In Amy we trust!

> Another question: Also in the PRB article,equation (2), there are two 
> delta function delta(Enk-Ef)*delta(En'k'-Ef). Does this mean before and 
> after scattering with phonon,the electron states should both have the 
> same energy,which equals to Ef?
yes.

> This is strange. I'm interesting in the 
> scatting of phonons and non-equilibrium "hot" electrons, If the ABINIT 
> code could only handle the e-ph coupling just at the fermi level, it 
> will be a great depression to me.
Well... valium? Or check the publications by Mauri and Ferrari on hot 
phonons. PWSCF may be able to do something. Or you can code in abinit. It 
shouldn't be a huge change, but I haven't looked at it in detail. For the 
moment the FS is taken into account with either the tetrahedron method or 
a smearing width. With the latter you can increase the temperature and get 
a wider contribution to the FS. That's a start, but there should also be 
something about the finite T of the phonons themselves somewhere. Dunno.

>    The third question: dose the GKK file contain the scattering matrix 
> elements? I can't understand its format.Can someone tell me how the bulk 
> S-matrix data organised in the GKK file? Can I find a detailed 
> explaination of the format,and the meaning of date in such *.ep_* files?
The gkk file contains the raw perturbations generated by abinit from 
phonon runs: perturbation of 1 atom in 1 reduced direction for 1 qpoint, 
for all atoms and directions but for the irreducible qpoints only. If you 
look at the input for mrggkk you can set the output to ascii text if you 
really want to look at them (huge file though). In the latest version 
(released in a few days) you'll have access to flags which let you write 
the actual gkk matrix used in the code to disk instead of keeping 
everything in memory. However these matrices contain an intermediate 
quantity used for interpolation

gamma=<psi_k+q | H_q,a,idir | psi_k > <psi_k | H_q,b,jdir | psi_k+q>

or the same thing dotted with the displacement vectors then rotated to the 
cartesian reference frame (q dropped, sum over everything):

e^j_a',idir' d^j_a,idir gamma d^j_b,jdir e^j_b',jdir'

Anyhow, this may not answer your question, but if you want the standard 
g_kk' matrices out you will have to write a few lines of code in 
src/07ddb/completeperts.F90 and src/07ddb/nmsq_gam_sumfs.F90

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