[gmx-users] REMD

[Eric Smoll]

Hello Gromacs Users,

I am interested in calculating the equilibrium distribution of molecular
structures at the vacuum-liquid interface of several different low vapor
pressure liquids. All of these liquids are very viscous at or near
room-temperature and I suspect that conformational barriers may inhibit
sampling at the vacuum-liquid interface. However, in NVT MD simulations,
these liquids increase fluidity at higher temperatures (400-500K) while
maintaining a fluid state and a reasonably well-defined vacuum-liquid
interface.

Can I use NVT REMD to efficiently overcome any kinetic trapping that might
be going on and obtain a true equilibrium distribution of molecular
structures at the vacuum-liquid interface? A superficial literature search
does not yield examples of NVT REMD on a liquid interface. I am curious if
there are issues or complications with this approach. Is there a better
alternative?

the manual states that "all possible pairs are tested for exchange" in
Gibbs REMD. Looking through the mdrun help output, it seems like this
option can be selected by setting the "-nex" flag. However, the comment for
this flag suggests using N^3. Isn't something like N*(N-1)/2 more
appropriate (where N is the number of replicas).

Thanks for the guidance!

Best,
Eric