FENE bond simulation

Hi Jan:

Thanks for replying. I used 0.005, shorter.


Dear Shaorui,

I would have thought that if your simulation runs for some time and crashes only later, the problem might be due to the dynamics properties of your system; are you using other potentials, for instance, that allow your atoms to move too close to each other over the course of your simulation?

With best wishes,

Hi Anna:

I re-checked the initial configuration. Bond lengths are indeed all close to 0.97. During the energy minimization the warnings that FENE bond is too long is probably due to some very close non-bonded pairs. With LJ potential this will cause large repulsive forces and may cause some bonds to be very long. So I switched to soft potential (with linearly increasing magnitude of parameter A) plus FENE for the initial equilibration and then transits to LJ potential. Things seemed better but still during the soft equilibration or during the subsequent LJ simulation some bonds occasionally get too long. Don’t know why


I am attaching my input script below:

boundary p p p
units lj
atom_style molecular

read_data MCCG_chainF.data

neighbor 0.3 bin
neigh_modify delay 0 every 1 check yes

thermo_style custom step temp pe epair ebond eangle ke etotal press vol lx ly lz pxx pyy pzz pxy pxz pyz
thermo 100

compute bondinfo all property/local btype batom1 batom2
compute bondvalue all bond/local dist eng

Don't know. However, I do know that most people
use a langevin thermostat with FENE, not NVT.
Because it's a coarse-grain model that matches
the Langevin assumption of there being a background
heat bath, e.g. of implicit solvent.