[Ifeffit] Strategy to obtain cofigurationally averaged EXAFS from MD simulations

Matt Newville newville at cars.uchicago.edu
Tue Apr 12 15:27:05 CDT 2011


Francisco,

This isn't really a job that Artemis was designed for.   That is, you
do not want to do a path-by-path analysis.   I think what you want is
to write a script that runs Feff on your several hundreds of
configuration, and then extract and sum the chi.dat files from each of
the runs.

Feff itself is rather clumsy for this purpose, as it uses files with
fixed names ("feff.inp", "chi.dat"), and so each run needs to be done
in a separate directory.    As you say, it's mostly a book-keeping
problem.    For EXAFS and in these days of multi-core machines, I
would not worry too much about doing a "full feff calculation".  That
is, it might be possible to save some time by not recalculating Feff
potentials for each run, but this may not be a huge overhead.

You probably want to run with "configurational averaging" (that is,
using all Zn atoms in a cluster as the absorber).  I would recommend
carefully testing this -- I believe that several versions of Feff have
claimed to do such configurational averages and have not done this
correctly.     For your case, if you get a Zn-O coordination of 6.00
for every configuration, you'll know ithe averaging is wrong ;).  In
that case, you may need to run Feff separately for each Zn central
atom for each of your snapshots / clusters.

Perhaps the RMCProfile code can take care of all of this.  Sadly,
although this code was written for the US Government and so is
required to be placed in the Public Domain, and although this code
apparently makes use of the Open Source code of Ifeffit and Artemis, I
could not download the code, as following all links leads to asking
for a password to get the software.

Hope that helps,

--Matt

On Tue, Apr 12, 2011 at 1:51 PM, Francisco Garcia
<garcia.ff.000 at gmail.com> wrote:
> Dear Anatoly,
>
> Thank you very much for your reply. Unfortunately, your PRB references
> did not solve my problems. I am trying to reproduce the results of
> Cauet et al [J. Chem. Phys. 132, 194502 (2010)]. I basically have over
> 500 snapshots of a single Zn2+ ion (absorbing atom) in water (O atoms
> are the scattering atoms). I have the feff.inp file for each snapshot
> ready for Artemis. I basically want an efficient way to obtain the
> chi(k) of data for each snapshot and average the chi(k) as opposed to
> loading the feff.inp files one by one. Specifically a sample
> script/example will be of imense help. I could not find much
> information in the Artemis user manual. I must also state that I am an
> EXAFS novice so a step by step explanation will help me a lot.
>
> Thank you very much.
> _______________________________________________
> Ifeffit mailing list
> Ifeffit at millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>
>



More information about the Ifeffit mailing list