Re: [AMBER] How to turn off non-bonded interactions between non-bonded atoms?

From: Marc van der Kamp <marcvanderkamp.gmail.com>
Date: Thu, 8 Dec 2011 17:48:13 +0000

Hi Ross,

Thanks for your very useful comments.

I'd like to make one more clarification:
My aim is to turn off non-bonded interactions between SOME atom-atom pairs,
but not ALL.
(see also the related thread: http://archive.ambermd.org/201110/0200.html )
This is why I think changing the source code is not such an attractive
option, as I'd need to hard-code what type of atom-atom pairs I'd like to
exclude. This is obviously not very flexible. An exclusion list in the
prmtop is very useful for this particular purpose.

Thanks,
Marc

On 8 December 2011 16:59, Ross Walker <ross.rosswalker.co.uk> wrote:

> Hi Marc,
>
> > I can see one can argue that this is the danger of (hand) modification
> > of
> > prmtop files.
> > On the other hand, why does igb=0 (i.e. the ewald code for the EEL/VDW
> > calculations) recompute the exclusion list based on the bond list?
> > Is this for efficiency reasons?
>
> I have asked this question myself many times. It rebuilds the list to be
> able to deal with extra points. But WHY? I have been burnt by this myself
> many times - I still get burnt by it but to be honest it will make you grey
> trying to understand why things like this are done - Obviously doing it is
> fine BUT it should have been done completely, with the exclusion list being
> completely removed from the prmtop and the 'original' path in the code also
> being removed.
>
> > If not, wouldn't it be better to make the ewald code for the EEL/VDW
> > calculations use the exclusion list as defined in the prmtop?
> > First of all, this does make it very tricky (near impossible?) to turn
> > off
> > non-bonded interactions between non-bonded atoms without changing the
> > source code. (In principle, one could add bonds with 0 force constants
>
> The lesson here is one I try to teach to anyone coming into computational
> chemistry and molecular dynamics. You have the source code. The best and
> most powerful approach is to get yourself familiar with the code, at least
> some of it. What you want to do is VERY simple by just editing the code,
> far
> more simple than trying to mess with a fragile prmtop. A few lines is all
> you need to turn off non-bond interactions and would take 5 to 10 mins at
> most.
>
> > Second of all, it would be nice if simulations actually *do* use what
> > is
> > defined in the prmtop. This is what is nice about having a prmtop - it
> > defines, in principle, the basic energy function used for the system
> > (admittedly without things like boundary conditions, cut-offs, implicit
> > solvent models etc.).
>
> Agreed, in principal. But it needs everyone to appreciate this. The problem
> arrises from the fact that the original prmtops were not extensible. Hence
> it was difficult, near on impossible, to make modifications to the prmtop,
> add new features to the prmtop etc and so people took short cuts elsewhere.
> These work but ultimately come back and bite people.
>
> That said though, what you want to do, turn off the non-bond interactions
> should be a setting similar to the cutoff - not hard coded in the prmtop.
> Thus I believe the correct approach here is to edit the code and add a
> switch there rather than messing with the prmtop.
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> ---------------------------------------------------------
> | Assistant Research Professor |
> | San Diego Supercomputer Center |
> | Adjunct Assistant Professor |
> | Dept. of Chemistry and Biochemistry |
> | University of California San Diego |
> | NVIDIA Fellow |
> | http://www.rosswalker.co.uk | http://www.wmd-lab.org/ |
> | Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
> ---------------------------------------------------------
>
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.
>
>
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Dec 08 2011 - 10:00:02 PST
Custom Search