Dear all
I am trying to simulate a Lennard-jones binary mixture at higher temp (T*=5) , For NVT simulation I am using nose-hoover thermostat . so my Tstart=5, Tstop=5. Tdamp=100 times of my timestep (according to manual). Timestep is 0.005tau. After simulation the average value of Temp is almost equal to 5, But the temp fluctuation range is little bit high . Can anyone suggest me any better thermostat for my system? The input file is given below. I have attached the temp fluctuation plot with MD step. I need less temp fluctuation for my calculation.
Thanks,
Atreyee
read_restart restart_5.dat
#mass 1 1
#mass 2 1
#velocity all create 5 11152 mom yes rot yes dist gaussian
pair_style lj/cut 1.122462048
pair_modify shift yes
pair_coeff 1 1 1.0 1.0 1.122462048
pair_coeff 1 2 1.5 0.8 0.897969638
pair_coeff 2 2 0.5 0.88 0.987766602
neighbor 0.3 bin
neigh_modify delay 0 every 5 check no
fix 1 all nvt temp 5 5 .5
thermo_style custom step temp press pe ke etotal vol enthalpy
thermo 100
thermo_modify flush yes
#dump 2 all custom 10 dump5.xyz id type xu yu zu
#dump_modify 2 sort 1
#restart 1000 restart_1.dat restart_5.dat
timestep 0.005
run 600000
Dear all
I am trying to simulate a Lennard-jones binary mixture at higher temp (T*=5)
, For NVT simulation I am using nose-hoover thermostat . so my Tstart=5,
Tstop=5. Tdamp=100 times of my timestep (according to manual). Timestep is
0.005tau. After simulation the average value of Temp is almost equal to 5,
But the temp fluctuation range is little bit high . Can anyone suggest me
any better thermostat for my system? The input file is given below. I have
this is the wrong question. do you *really* want to use a thermostat
that gives you unphysical dynamics?
the magnitude of the fluctuations depend (primarily) on the size of
your system. in the most extreme case - a harmonic - oscillator the
kinetic energy changes from zero to max and back.
attached the temp fluctuation plot with MD step. I need less temp
fluctuation for my calculation.
why so? it looks more like you need a little *more* statistical mechanics.
axel.
The damping parameter can control the temperature fluctuations to some extent, so try changing that.
Also, your pair coeffs look a bit funny...
Chris.
The damping parameter can control the temperature fluctuations to some extent, so try changing that.
i strongly recommend against that. the only place where you want to
control the fluctuations is if you are doing langevin or brownian
dynamics where you have an explicit damping term in your hamiltonian.
but not if you want a proper NVT ensemble. your just replacing doing
science with making a movie. now if you can prove that those
modifications (e.g. using the drag term) have no significant impact
(e.g. when you are simulating a very incompressible system like a
solid), then you may get away with it, but to tailor the proper output
of your simulations because you don't like it, is something else.
axel.
axel