[Wien] use of in1new

[L. D. Marks]

As Stefaan says, with a converged calculation use N=1. In general I would 
suggest something like:

run_lapw -cc 0.01 -ec 0.01 -in1orig
run_lapw -cc XX -ec YY -in1new 1

i.e. first approximately converge, and only then turn -in1new on. If you 
turn -in1new on too early you may run into problems.

On Mon, 12 Dec 2005, Stefaan Cottenier wrote:

>
>> i have performed a calculation with all defaulted parameters and the
>> calculation successfully converged. but i want to know if the default
>> linearization energy and local orbit are correct or good enough, so i
>> want to perform a calculation with a -in1new switch. i dont what
>> number should i set for N, i.e. how many iterations should be taken
>> before rewriting the case.in1 file. from the user mannual page 52 i
>> know that if the N is not set correctly, a bad result would be listed.
>> for my calculation was converged,should i set the N larger, for
>> example just before the end of the calculation(that is about 19)?
>
> Because you start already from a converged case, you can set N=1.
> Usually I set N=5-10 when starting from scratch, and switch to smaller
> values (N=2-3) once moderate convergence has been reached. The large
> values in the beginning prevent getting crazy linearization energies
> (and subsequent divergence) obtained from the still very much
> oscillating solutions in the beginning of the scf cycle. But once
> convergence has set in, the linearization energies will not change very
> much any more, and it does not harm to update them every 2-3 iterations.
>
> If you don't want to interrupt your scf-cycle to change N, then start
> with N=5 from the beginning. It works almost always.
>
> Stefaan
>
>
> Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

Note: if you have an old email address for me, please note that "nwu" has
been changed to "northwestern".
-----------------------------------------------
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60201, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
http://www.numis.northwestern.edu
-----------------------------------------------