[gmx-developers] Reaction field questions (David van der Spoel)

[Shirts, Michael R. (mrs5pt)]

> Given that some people are rather strict when it comes to accuracy of
>integrators and thermostats (thinking of you Michael ;)) the subject of
>simplifying the potential should not be disregarded completely.

If the integrators and the thermostats are incorrect, then statistical
mechanics breaks down if we don't have a Boltzmann distribution. Hence we
need first and foremost physically consistent methods, whatever them ode.
Bad physics has consequences that are uncontrollable.  But a models are
different.  "All models are wrong, but some are useful".

> In fact, the hamiltonian contains potential energy V(r) and a PMF like
>term G(r) which by definition includes the entropy and is an average over
>time and/or coordinates. In other words we get a new potential energy
>V'(r) = V(r) + G(r)

But from the point of view of the simulation, it's just a V(r).  A single
configuration gives rise to a single potential energy.  The INTERPRETATION
of that potential energy in physical terms is now problematic, but it's a
perfectly reasonable model to simulate.  If it existed, the simulation
would give the correct properties of it.



> different from dV(r)/dr, since it contains dH/dr - TdS/dr due to e.g.

But it doesn't -- it contains a V(r).  One can _interpret_ it as a dH/dr -
T dS/dr for a different model with more conformational freedom, but that
breaks down when the temperature changes.


> For sure one loses the entropy due to correlated motions between e.g.
>protein and solvent (for an implicit solvent).

I agree 100% with this statement.

> As a result, I think, all thermodynamic quantities including the
>temperature dependence (heat capacity) will be systematically wrong. This
>even goes for united atom models (or indeed when using constraints).

They will be systematically different from another model that includes
those degrees of freedom, yes.

> Now the question is, HOW bad is this?

Great question!  I have very little idea and would love to know in
different situations (gas vs liquid vs crystal, etc).

> That I'm not completely sure off (but working on it). For entropy of
>liquids the effect of using constraints on the entropy is an entropy
loss of a few %. I suspect that the effect of a reaction field may be
measurable as well.

Interesting! I suspect here that entropy by temperature difference will be
more accurate than entropy by (H-G)/T, FWIW.


Best,

~~~~~~~~~~~~
Michael Shirts
Associate Professor
Department of Chemical Engineering
University of Virginia
michael.shirts at virginia.edu
(434) 243-1821