Re: [ESPResSo-users] Porting to 3.1.0

[Axel Arnold]

On Monday 09 April 2012, Michael Winokur wrote:
> Hi All,
> 
> So it has been a while but I've looking into running Gay-Berne
> simulations on the 3.1.0 release as opposed to the 2.1.2 version I
> have used in the past.  There are a lot of changes to be sure and so
> it is a bit difficult to track down the underlying cause for
> differences in the numerical values of a tcl test code.
 
> One difference, which I believe may be a simple error, has to do with
> the definition of GB_chi2 now located in gb.c.  The original
> definition in 2.1.2  (in gb.h) was
>   data->GB_chi2 =
> (1-pow(data->GB_k2,(1/data->GB_mu)))/(pow(data->GB_k2,(1/data->GB_mu))+1);

No, it wasn't, at least not according to git. It was since 2004 the same 
formula, the one that is below.

> but in 3.1.0 there is a sign change to specify
>   data->GB_chi2 =
> (pow(data->GB_k2,(1/data->GB_mu))-1)/(pow(data->GB_k2,(1/data->GB_mu))+1);
>
> The values of energy and torques that I see in gb.h, as far as I can
> tell, seem to give consistent numbers using the 2.1.2 expression.

chi2 is used consistently in energy and forces, therefore, no matter which 
value we put there, force and energy are consistent. The two forms simply 
correspond to two different ways of defining the potential. If you take a 
quick look at the formula, just replacing k2 with 1/k2 does the conversion 
between the two formulas.

We were discussing about this form of the Gay-Berne potential long ago, and as 
far as I remember, you were using a patched form to match your definition, 
which is probably why you have the changed sign in your 2.1.0.

Many regards,
Axel

-- 
JP Dr. Axel Arnold      Tel: +49 711 685 67609
ICP, Universität Stuttgart      Email: address@hidden
Pfaffenwaldring 27
70569 Stuttgart, Germany