AMBER Archive (2004)

Subject: Re: AMBER: TI with electrostatic decoupling

From: David A. Case (case_at_scripps.edu)
Date: Tue Nov 23 2004 - 10:42:16 CST


On Tue, Nov 23, 2004, Nelson Fonseca wrote:
>
> In order to run a TI free energy calculation with sander,
> with electrostatic decoupling, I should run 2 separate
> jobs. In the first one, the cahrges are perturbated and in
> the last one, the atom type parameters.
> My question is if starting, for example from state A to
> reach state B, I should pass throug an intermediary state
> A?, in such a way that from A --> A? the charges are
> perturbated and from A? --> B only the atom type are
> changed or if I can reach state B from another way, A
> -->B, where the charges are perturbated and starting again
> from the same point A --> B, where only the atom types are
> perturbated.

You should use the former option, not the latter.

> Should the inicial geometry from the path A? --> B
> correspond to the last geometry A --> A? where the charges
> were changed?

Results from TI ideally should be independent of the starting geometry. If
they are not, you are not really converging the calculation. One way of
checking the internal consistency of your simulations is to try different
options for the starting geometry and see what happens.

...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu