virial contributions in pull.c
When looking to the code pull.c (gmx-3.3.1) I see e.g that in the umbrella
sampling part a virial contribution is calculated from the double sum of the
pull forces together with the pulled positions relative from equilibrium (the
well of the potential). The same holds for AFM and constraints.
However, instead of the expected factor -0.5 a factor of +0.5 is included while
in other parts of the code the -0.5 is included (calcvir.c).
The pressure equations implemented in gromacs is the one directly from
litterature which however also implies the -0.5 factor in the virial.
It looks that the virial contribution of pulling is now substracted instead of
added to the total_virial.
Physically I would expect the opposite, hence it's the virial contribution added
to form the virial which represents the state of the biased ensemble.