High fluctuation during NPT

Hello,

I’m trying to get the thermal expansion of a carbon nanotube. First the system was stabilized using fix langevin and nve for 35,000 time steps (dt=0.0001 ps). After this point, I applied npt ensemble to equilibrate the system at 50 K and 0 bar but there’s a big fluctuation in the pressure and I want to get of it. I tried different Pdamp values (0.01-1) and I also tried the drag keyword in npt but I couldn’t reduce the fluctuations. I’m attaching the figures in the ppt file that show the pressure and total energy behavior during the Langevin fix and npt. Any help to solve this problem would be appreciated. The code I used is copied below.

Fluctuation.pptx (1020 KB)

Hello,

I'm trying to get the thermal expansion of a carbon nanotube. First the
system was stabilized using fix langevin and nve for 35,000 time steps
(dt=0.0001 ps). After this point, I applied npt ensemble to equilibrate the
system at 50 K and 0 bar but there's a big fluctuation in the pressure and
I want to get of it. I tried different Pdamp values (0.01-1) and I also
tried the drag keyword in npt but I couldn't reduce the fluctuations. I'm
attaching the figures in the ppt file that show the pressure and total
energy behavior during the Langevin fix and npt. Any help to solve this
problem would be appreciated. The code I used is copied below.

​you are asking the wrong question here. it is not about making a property
of your system behave the way you want it, but rather you need to
understand why properties are the way they are. instantaneous pressure of
an atomic scale system fluctuates and it fluctuates more the less
compressible a system is and the smaller a system is. on what grounds do
you believe that the size of fluctuations that you are seeing are not
expected for your system. perhaps a closer look into a text book on
statistical thermodynamics might help.

what you should worry about *more* is to have a physically meaningful
simulation. and there i would be very concerned about your force model.
overriding mixed terms of a manybody potential the way you do almost always
leads to inconsistencies. also, i don't think that CNTs expand
isotropically.​

​axel.​