[Wien] RMT reduction

[Laurence Marks]

Let me try and explain a little. In general you want to have the RMT's
as large as possible because then the calculations converge faster and
also this is a better model for the core electrons. The RMT's that you
need to have don't track with classic atomic radii but more what is
going on with the core electrons, for instance for transition metals
you generally want to have a larger RMT than for oxygen. What RMT you
use also depends upon which states you decide to treat as semi-core,
i.e. where you set the cutoff between core and valence in lstart (and
case.inc, case.in1, case.in2).

The Wien code cannot do a calculation if RMT's overlap. During
minimization, often atoms will move quite close to each other, for
instance if one wants to move to a very different position relative to
the first. In this case you need to have a much smaller RMT than you
would use for the optimized atomic positions.

The command "setrmt" will do this approximately for you. It is only a
computer code, so you should look at what it is recommending, think
about the values and run some tests (e.g. a simpler bulk structure) to
understand what different RMT's mean. You should also look at the
value of :NEC in case.scf and also the core integrals in case.outputm
to check that you have adequately constrained the core electrons
within the muffin-tins. No computer code is foolproof, the user has to
make sensible decisions.

When you are running a minimization, you want to set things up with
your "best guess" as to how close the atoms are going to come during
the minimization. If you think your starting point is very close to
the minimized positions, you can use a small reduction; if you think
you are far you may want to use a larger one. If you err with too much
reduction this is not too large a problem; if you use too little the
minimization won't run properly and you will have to adjust the RMT's
(see the clminter command).

The standard way I run a minimization is to make a guess at the
reduction, make the RMT's a bit smaller than this, use a smaller RKMAX
(e.g. RKMAX/min(RMT)=2.5 to 3) and a smaller oversampling in case.in0,
less k-points and less stringent convergence parameters (e.g. -fc 4
not -fc 1). Only when the forces have reduced down to some reasonable
level (e.g. 6-10 mRyd/au) do I change to more accurate values.

The value of "2" in nn is not particularly important, it is just how
far out the code prints out neighbors.

On Wed, Jul 2, 2008 at 6:21 AM, Santhy Jaiker <k_santhy79 at yahoo.co.in> wrote:
> Dear wien2k users,
>   i have some basic doubts in % reduction of RMT.  Any suggestion will be
> appreciated.
> ·        What is role of 'RMTreduction' in mini?
> ·         What are the factors decides the %reduction of RMT?
> ·     what is the minimum and maximum % reduction of RMT?
> ·         nn distance i.e distance factor is usually 2.  Is it applicable to
> all the cases?
> Thank is advance
>
> Regards,
> Santhy jaiker
>
> Regards,
> Santhy Jaiker
>
> Research Scholar,
> Materials Thermodynamics Lab,
> Department of Metallurgical and Materials Engineering,
> Indian Institute of Technology Madras,
> Chennai - 600 036,  India
> Tel: 91-044-22575763
> **********************************************
> A truly happy person is one
> who can enjoy the scenery on a detour
> **********************************************
>
> ________________________________
> Bring your gang together. Do your thing. Find your favourite Yahoo! Group.
> _______________________________________________
> 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
Web: www.numis.northwestern.edu
Commission on Electron Diffraction of IUCR
www.numis.northwestern.edu/IUCR_CED