[gmx-developers] Free energy: Thermodynamic integration

[deepak bapat]

Dear gmx-developers

I know these issues must be addressed at gmx-users but please pardon my
insistence/stubbornness on putting it on gmx-developers.

I was trying to do free energy calculations (V 5.1.2 ).

It is a simple molecule such as ethanol which is being decoupled from
water. In the first step I try to decouple charges from the system. So in
this setup VDW parameters are never modified; only electrostatics are
turned off. OPLS forcefield + tip4p water model.

I am using thermodynamic integration rather than gmx-bar. In order to do
the same I am making use of the dual topology (type B directive) .

In the .mdp file for production run I have tried two setups

1) I have persisted with sc-alpha directive (which is applicable with
gmx-bar and couple-moltype setup) and added -maxwarn option to my gmx
command.

2) In the second set I removed sc-alpha directive.

But sc-coul in both the cases is set to default value which happens to be
no.

I have observed that dh/dl values in both the cases differ significantly.

~ 511 kJ/mol and ~ 92 kJ/mol

Since while turning off elctrostaics VDW don't change I expected both set
of simulations to give similar values.

I have attached my working setup for both set of simulations along with
output of .xvg files as well.

https://drive.google.com/file/d/1P6gP2KYx3MYy_jL5OkJ6UUsZeMNI811F/view?usp=sharing

I am using customized work flow so atom name nomenclature may not
necessarily match with the one used by gromacs.

I kindly request you guys to look at the same. If I have blundered please
don't hesitate put all of this in trash! :)

Regards

Deepak



-- 
Deepak U. Bapat
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20180709/16649174/attachment.html>