[gmx-developers] Molecular vs atomic virial

[ramon at juguete.quim.ucm.es]

> Whether or not to use an molecular or atomic virial. The molecular
> virial is very unpractical and close to impossible to implement in
> combination with different pressure scaling algorithms and constraints.
> However it could be an option to scale molecular coordinates only when
> doing pressure scaling, so as not to induce constraint errors due to
> pressure scaling.

I can't see any problem with the atomic virial.
The constraint errors induced by pressure scaling
do not introduce any new problem because moving
the positions v dt also modifies constraints.
It seems logical to me to do both updates at
the same time.

I proposed an implementation in bug #18.

See also the DL_POLY manual, the description
of the barostat in 

http://www.cse.clrc.ac.uk/msi/software/DL_POLY/MANUALS/USRMAN2/node78.html

Ramon