[gmx-users] question on NPT ensemble

Lianqing Zheng lzheng at me.rochester.edu
Tue May 13 23:43:01 CEST 2003

Dear Gromacs pals:

As I understand (from Gromacs manual and codes), when
Nose-Hoover-Parrinello-Rahman NPT ensemble (temperature and pressure
coupling) is used, the atom coordinates are coupled with both xi (for
temperature) and b (for pressure), but xi and b are not coupled with each
other. Instead, in Nose JCP 81:511, 1984, s (xi=s_dot/s) and V (V=det(b)) 
are coupled in their equations of motion. s was interpreted as scaling of
time (s*dt' = dt, dt--scaled time, dt'--real time). One year later, Ray
and Rahman (JCP 82:4243, 1985) used this factor as a mass scaling (m=m'*s,
m--scaled atom mass, m'--real atom mass) and got un-coupled equations for
s and b. Actually their equations are equivalent to those under scaled
time in Nose's case. 

Generally, the equations of motion are deduced from the Lagrangian or
Hamiltonian of the system, however both in Gromacs and in Moldy, EOM for
NPT are simply EOM for NPH plus EOM for NVT. For me, it is ok to get EOM
for NPH or NVT from those of NPT, but this process may not be
mathematically reversible.

Any comments? Thanks!!


More information about the gromacs.org_gmx-users mailing list