AMBER Archive (2004)

Subject: RE: AMBER: cut offs in amber?

From: Grace Li (w24li_at_student.cs.uwaterloo.ca)
Date: Wed Jul 21 2004 - 17:03:05 CDT

Dear Ross,

I'm sorry I forgotten to mention that I have a Periodic boundary system
with PME.

I thought that the atom based cutoff was on an atom by atom basis for all
phases not just for gas phase.

Also just to be sure, I should also set CUT < 1/2*boxLength just to make
sure that only one image is seen by each atom?

Thanks again,

Grace

On Wed, 21 Jul 2004, Ross Walker wrote:

> Dear Grace,
>
> > Thanks for your reply. I have a system that contains ions and it is
> > critical to get the electrostatic interactions as accurate as
> > possible. As
> > I understand it, atom based cutoff can cause problems if the
> > cut off is
> > too short. My system is approximately 24 X 24 X 24, would a
> > CUT = 9 be ok?
> > The ions are simple molecular ions.
>
> You don't say whether you are using periodic boundaries or not. With the
> particle mesh Ewald system implemented in Sander the cut off value is only
> used to limit the direct space sum and so it has a much smaller effect than
> in gas phase where the cut off value is used to truncate non-bonded pairs on
> an atom by atom basis. Thus if you are doing a periodic boundary simulation
> a cut off of 9 will probably be ok. If you doing gas phase then you should
> increase this to as large as you can computationally handle. Essentially the
> larger the better. Try running a single step of minimisation and plotting
> the energy as a function of cut off distance (see
> http://amber.scripps.edu/tutorial/polyA-polyT_New/minandmd3.html#5.1.2)
>
> You should choose a cut off distance larger than the point at which your
> energy starts to converge.
>
> > Wouldn't group based cutoff with shorter cut off be a better
> > alternative in this case rather than atombased?
>
> For gas phase yes it would as this would ensure that everything within the
> cutoff for a single atom has a charge sum of zero. This is not the case with
> an atom based cut off. However, this is not implemented within sander due to
> the problems with maintaining separate non-bonded lists and update
> mechanisms for gas phase vs non gas phase simulations. If you really need to
> do a gas phase simulation and need residue based cut-offs you could try
> using sander.QMMM (apply bugfix.13 first) and setting nqt=0. I have no idea
> if this will actually work though so you'd have to try it. I believe
> sander.classic that used to ship with older versions of amber also includes
> residue based cut-offs, although I have not tried this.
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> | Department of Molecular Biology TPC15 |
> | The Scripps Research Institute |
> | Tel:- +1 858 784 8889 | EMail:- ross_at_rosswalker.co.uk |
> | http://www.rosswalker.co.uk/ | PGP Key available on request |
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu