LJ to semi-empirical

[ivan]

Hi Axel,

On szept. 17, 13:40, Axel <akoh... at gmail.com> wrote:
> ivan,
>
> On Sep 17, 7:15 am, ivan <solt... at yahoo.com> wrote:
>
> > Hi Axel,
>
> > thanks for your help. i would prefer a pure semi-empirical simulation,
>
> that is understandable, but as juerg pointed out, without
> support for periodic boundary conditions, you can do only
> cluster calculations.

OK, I see.

>
> > rather than a hybrid run. it's not clear to me if AM1 or PM3 can be
> > used with some kind of stochastic boundary conditions (or anything
> > else, that simulates a thermostat), or not. Are they only suitable for
>
> i don't quite get what you mean with 'simulate a thermostat'.
> the way i understand juergs post is, that cp2k currently does not
> account for the interactions between periodic images. you can use
> thermostat just fine, but with a cluster calculation the problem is
> how to control the density etc.
> hence my suggestion for doing a qm/mm calculation. this would
> give you almost ideal boundary conditions for the SE cluster.

So this means, that Semi-Empirical methods can be used even
periodically, if they're embedded in a calssical environment? Of
course, one neglects the SE-SE interactions between periodic images,
but if the box around the QM part is big enough it does not change
things too much.. Is that correct? Or would it be simply another
cluster calculation (with significantly more atoms due to reduced
computational intensitiy of the classical model)?

> and for that i'd have a couple more ideas on how you'd get
> the optimal correction to the interactions with minimal additional
> programming by mostly using features already present in cp2k.
>

many thanks in advance,
i