Re: [ESPResSo] Some Espresso questions and comments

[Axel Arnold]

On Friday 20 November 2009 12:19:21 am Michael Winokur wrote:
> Hi All,
>
> I am slowly learning more about C and it isn't quite as daunting as I
> thought it would be.  I've now added a subtracted Gay-Berne pair
> interaction for bound Gay-Berne pairs and I'm wondering how one goes about
> adding these modifications to the development area.  In addition it is my
> impression that the Gay-Berne potential wasn't fully implemented. 
> Basically there was no way to enter or modify the default rotational
> inertias I_xx, I_yy and I_zz from their default values of "1".  This gave
> some interesting results.  The particle properties have been modified to
> incorporate a rotational inertia and I've made some modest "fixes" to the
> Gay-Berne subroutines.  I've tried hard to follow syntax of existing
> particle parameters so that reading, writing and interprocess communication
> is done according to the established protocol.  There is still more to do
> but the current subroutines are in reasonable shape.

Hi Michael!

Currently we are still using CVS over ssh, therefore it is a bit difficult to 
directly access the repository for you. However, if you can post the files 
that you modified here on the mailing list, I guess I can try to incorporate 
them.

> Mapping of Gay-Berne particles:  I would like to use the interactive imd
> subroutine to allow me to watch the simulation in real time.  Each GB
> particle represents a benzene type structure and it is relatively easy to
> use the GB particle position and quaternion to reconstitute the fine grain
> picture.  To integrate this into Espresso it would be nice to have a
> "sub-particle" mapping which would generate the fine grain picture for the
> vmd program.   If anyone has a good idea as to how this could be most
> easily implemented I would like to hear their suggestions.  I've started to
> clone the particle_data.h and particle_date.c subroutines for generating
> subparticles but I think this is really a difficult task.

Yes, this is not an easy task, since somewhere, you need to specify how to map 
a single GB particle to N sub-particles. I would suggest not to put this into 
the particle, but rather to make a separate translation-table that contains 
for some particle types a) to how many particles this type translates and b) 
what are the positions of these particles in some coordinate system relative 
to the GB axes. This table you only need on the master node, so there is no 
communication involved in this, which makes things easier. You can then use 
this table in imd_parse_pos to replace these particles with the 
sub-particles. imd_parse_pos has one loop that copies the positions to a send 
buffer; here you first would have to calculate the effective total number of 
particles, and then write the routine such that the particles are still 
copied into the buffer according to their identity, so that they do not hop 
in VMD. Well, not too easy, but as far as I know, IMD, does not directly 
allow sub-particles or so.

Axel

-- 
Axel Arnold
Ludwigstr. 51
53721 Siegburg, Germany
Tel. +49 160 9166 7532