Re: Help in FEP backward/reverse calculation

From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Fri Sep 23 2011 - 03:22:52 CDT

Hi Bruno,

> For the bacward transformation I need to use the files otbained from the
> forward transformation (forward_0.coor, forward.coor, forward.vel, etc...)
> Am i right???

Yes, that's correct, because that configuration is equilibrated at lambda=1.

> Another doubt:
> My backward input should be:
>
> alchEquilSteps           8000
> set nSteps               40000
>
> set dLambda             0.1
>
> set init                   {0.9 0.95 0.99 0.999 0.9999 0.99999 0.999999
> 0.9999999 1.0}
> set end                  {0.0 0.0000001 0.000001 0.00001 0.0001 0.001 0.01
> 0.05 0.1}
>
> runFEPlist              $init  $nSteps
> runFEP                  0.9    0.1      $dLambda   $nSteps
> runFEPlist              $end   $nSteps
>
>
> Than the backward simulation followed the same setup with the exception of
> the fep.tcl file where the fags are 1 for the Re(CO)3. That´s it??

No. You have two choices here.
a) reverse the FEP flags and keep the "forward" lambda schedule (going
from 0 to 1), or
b) keep the same FEP flags and reverse the lambda schedule to go from
1 to 0, like this:

set init {1.0 <intermediate values here> 0.9 }
set end {0.1 <intermediate values here> 0.0 }

runFEPlist $init $nSteps
runFEP 0.9 0.1 $dLambda $nSteps
runFEPlist $end $nSteps

Both options should give exactly the same results. The common practice
is option b), maybe because it is closer to the intuitive idea of
doing one transformation forward then backward. Instead, in option a),
you would change the identity of the group from "vanishing" to
"appearing", and always go forward.

Cheers,
Jerome

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:24:25 CST