[gmx-users] potential under constrain application

[Erik Lindahl]

> Dear gmx,
>   I'm wondering if application of LINCS algorithm changes the 
> potential surface.
>   In other words, if I calculate potential energy value U(X) of a 
> particular set of coordinates X during an unconstrained simulation, 
> and the potential energy U*(X) of the same set of coordinates X during 
> the simulation of the same system, with the same parameters but adding 
> LINCS constrains, will U(X) and U*(X) be the same?
>

No, not in general - unless your unconstrained coordinates just 
happened to be exactly at the zero value of the harmonic potentials, 
which isn't very likely.

If constraints did not change the potential surface (by removing some 
degrees of freedom) they simply wouldn't have any effect :-)

It gets a bit more complicated with simulations, but the thermodynamics 
of constraints is well worked out - consult e.g. Allen & Tildesley for 
a good (although somewhat equation-heavy) description.

Cheers,

Erik