[gmx-users] Shift parameter

[B. Nick]

Dear gromacs users,

I have also changed my input parameters of mdrun
for vdwtype and coulombtype from "cut-off" to "shift"
to decrease the potential smoothly to zero. 
Similar to what was reported by Cristian some days
ago I observe an increase in the box fluctuations
during a variable shape MD run on cubic boxes
of PE. I do not have an explanation for this.
 From my point of view the opposite should happend:
smooth reduction to zero potential should decrease
the fluctuations (as written in the manual,p.53) 
Can somebody give any explanation for this?

My system: PE box of 4.5nm edge length (no charges)
Parrinello Rahman anisotropic pressure control
dt          = 0.0005
rlist       = 1.2
vdwtype     = shift
rvdw_switch = 1.1
rvdw        = 1.2

Thanks for any suggestions,
Gitta