[gmx-users] Re: WCA and free energy calculations

[David Mobley]

Hi,

I'm CCing the users list on this since it might be of general interest.

>  I
> can't begin to tell you already how much help you've given me with your gromacs
> user-list posts and free energy tutorial.

Glad I can help!

> I was searching the Gromacs user list and came across a thread by you titled:
> tabulated nonbonded interactions and free energy calculations.
>
> I'm trying to do exactly the same thing that you described, using the WCA LJ
> potential and free energy calculations.  No one ever responded to this thread,
> but I was wondering if you figured out the answer?  Also, just to confirm that
> I'm not spinning my wheels did you get the WCA LJ potential to work using the
> user specified potential function?

So, no one ever responded to that thread, as you noticed. But my end
conclusion was that the tabulated nonbonded interactions know nothing
about lambda. So while one could implement the WCA separation using
tabulated nonbonded interactions, one couldn't use it to do free
energy calculations in a straightforward way, at least not
simultaneously with the soft core potentials I wanted to use (where
the effective radius is modified as a function of lambda). So I've put
it on the back burner for now. I guess you could in principle get WCA
to work with linear lambda scaing, but there is a lot of work in the
literature indicating that linear scaling is bad for transformations
involving insertion or deletion of LJ sites, which is presumably the
point of using the WCA separation in the first place.

Anyway -- I think that all means that one would actually need to
implement WCA into GROMACS to be able to use it for free energy
calculations.

I hope this makes sense. Please let me know if any of it needs further
explanation. Developers are welcome to weigh in if I'm going wrong
here anywhere...

Thanks,
David