[gmx-users] limitations in the umbrella sampling

[Seungho Choe]

Dear all,

In the manual, it says :
If a group contains part of a molecule of which the bondlengths are 
constrained, the results may be unreliable.

I'm a little confused.
Let's think of the following two cases (A: the reference group, B: a 
group which will be pulled) :

1) A(partially constrained)  + B (no constraints)
2) A(no constraints) + B(partially constrained)

Does it mean that both pmf calculations are not reliable ? Should I use 
the free energy code for both cases ?
Any comments would be appreciated.

Thank you.