[gmx-users] constraint pull code / gromacs 5.0.4

[Alex]

>
>
>>
> ...which is correct if one is sampling independent windows and collecting
> the forces to reconstruct the PMF.
>

That is nominally true, but if there are weird COM motions (due to
something seriously wrong in the system), the puller could further shake
things up -- I have seen this in my own simulations. I am suggesting taking
out the pull code completely, one step at a time.