[Wien] PORT

[Laurence Marks]

On 10/28/06, John Appleton <banger_deep at yahoo.com> wrote:
> Dear L. Marks,
>
> after carefully reading your notes on port,
> I find myself asking the following questions which
> has not been approached on the mailing list:
>
> It is stated that for surfaces -fc 5.0 and 5-10 unique k-points should
> be okay for a start. what will be a "good" value for temp in case.in2?

Yes, in other words it depends from case to case. I would try
something like 0.0035 to 0.005. It normally does not matter too much
unless you have a very small bandgap material or a low work-function
surface.

>
> It is also stated that after a run is done and Rkmax is increased,
>      .min_hess should be copied to .minrestart and case.tmpM and
>       case.finM should be removed. What about case.tmpM1? Should
>       it also be removed?

You can remove it, it does not matter.

>
>  In the userguide it states that when you more 4000 matrix elements
>      and "few" electrons, the -it N switch can help. This was not addressed
>      in your notes. How does it work?

That's because this is not directly related to minimization. The -it
switch does iterative matrix diagonalization, someone else should
answer (I've never managed to get it to be really useful, probably
because I've used it wrong). I suspect that it has to be used very
carefully in a minimization.

>
> Just like newt in the FAQ, can the values of delta be adjusted in
> case.inM be adjusted while min is running using the port method?
> E.g. oscillations or small  changes in f

NO. DO NOT CHANGE THEM, particularly if you used "x pairhess" to
generate the initial case.inM and .minrestart. An occaisional step
upwards in energy is NOT a problem, and is to be expected. (Similarly,
an occaisional step where :DIS & :PLA increase in the mixer is NOT a
problem.) Any minimization algorithm (mixer is related) will take some
steps uphill, but if it is well safegaurded this is not a problem and
can in fact help. PORT is very well safegaurded -- not my work, this
was done by some of the biggest names in minimization some years ago.

>
> Thank you
>
> J. Appleton
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>
>


-- 
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
http://www.numis.northwestern.edu