Dear Pascal,
> I was wondering how non-periodic simulations (ntb=0) work and
> whether there is
> some potential to keep a system together and avoid solvent
> molecules to drift
> off (which I have not seen in my trials..).
My first question is why you would want to try and run a non-periodic
simulation with solvent in place? If one can assume that you have good
reason for this, note performance is unlikely to justify it as once you have
increased the cutoff to something like 18 angstroms or so to make up for not
having a PME treatment of the long range electrostatics you will probably
find the 'pseudo' gas phase simulation to be slower.
Anyway, if you really need to run a non-periodic simulation without some
kind of restraint you will always see water molecules boiling off into
vacuum, as you would see in reality if you heated a droplet of water to 300K
in a vacuum. In Amber you can use a solvent cap to solvate your system, this
will add a restraint force around molecules at the edge of the cap that acts
to keep them inside the water sphere. The amber 8 QMMM tutorial
(
http://www.rosswalker.co.uk/tutorials/amber_workshop/Tutorial_Six/index.htm
) shows you how to create such a system. With classical simulations you can
also have an implicit solvent field around the outside of the solvent cap.
Note, however, that such a solvent cap method is likely to give you very
artificial pressures within the solvent cap unless you set it up very
carefully. The only reason it is recommended in the Amber 8 QMMM tutorial is
because periodic boundaries are not supported here. The Amber 9 update to
this tutorial will use regular PME periodic boundary conditions.
> If one were to start with an octahedric box, would there be
> anything to keep it octahedric during the simulation?
Nope. Although you might be able to create some elaborate restraint force
for a number of water molecules around the edge of the box that restrains
them to their original positions. You would have to do this with a
sufficiently large buffer of the water molecules though to ensure that any
of the waters deep inside the box that are not restrained cannot find their
way out into the vacuum.
> Last, and not very related, is there any input file keyword
> to determine a
> cutoff on the calculation of van-der-waals interactions?
This is the same as the value for the electrostatic cut off, you cannot set
it independently. Note even if you could set it independently there would be
very little performance benefit, actually it would probably run slower, as
you would have to build two non-bond lists, one for the electrostatics and
one for the van der Waals interactions. Most of the time is spent
constructing these lists and also calculating the value of 1/sqrt(r^2) which
you have to do for the electrostatics anyway. So once you have done the
electrostatic interaction between a pair of atoms you essentially get the
van der Waals interaction for free. Hence since it doesn't do any harm to
have 'too big' a cut off the VDW cut off might as well be the same as the
electrostatic cut off.
All the best
Ross
/\
\/
|\oss Walker
| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
|
http://www.rosswalker.co.uk | PGP Key available on request |
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.
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Apr 23 2006 - 06:07:07 PDT
