Reg_Fix SMD

Hi all,

I am trying to pull the polymer chain away from the fixed surface. I am using “fix smd” to pull the polymer chain along z direction. If am not wrong, I should use either nvt or npt in order to have updated position. In this case I have the surface fixed using “fix setforce” and one end of chain is fixed using “fix spring/self”.

I am not sure which ensemble to use either nvt or npt… Is there any specific conditions or applications under which these has to be used?

Also, is it possible to apply these ensembles only to the polymer chains, because the position of atoms in surface is not going to change over simulation.

If I use nvt ensemble for the entire system, after some run, the temperature is increasing drastically and I get an error “bond atom missing”.

If I use nvt ensemble (300K) for the polymer chain alone the temperature fluctuation is so high… It is between 70 to 600 K.

units real
atom_style full
bond_style harmonic
angle_style charmm
dihedral_style charmm
pair_style lj/cut/coul/long 10.5 10.5
kspace_style pppm 1.0e-4
neigh_modify every 1 delay 0 check yes one 10000

Here is my input script.

group HAP id > 2206
group CS id < 2206
group fixed id <> 2002 2206
group end id <> 1103 1127

dump 2 all dcd 10 dump_for_equi.dcd
compute 1 CS temp
fix 1 HAP setforce 0 0 0
fix 2 fixed spring/self 10
fix 3 CS nvt temp 300 300 1
fix 4 end smd cvel 7 -10 tether NULL NULL 150 0
thermo_style custom step temp press vol c_1

thermo 10
timestep 1
run 10000
unfix 1
unfix 3
unfix 4

Thank you,

Hi all,

I am trying to pull the polymer chain away from the fixed surface. I am
using "fix smd" to pull the polymer chain along z direction. If am not
wrong, I should use either nvt or npt in order to have updated position. In
this case I have the surface fixed using "fix setforce" and one end of chain
is fixed using "fix spring/self".

from your somewhat vague description it is not clear to me, why you
need fix spring/self.

I am not sure which ensemble to use either nvt or npt.. Is there any

please note that fix nvt or npt are *not* defining an ensemble, but
provide an integrator with thermostat (and barostat in case of npt).
when using smd, you are in neither ensemble; same goes if for setups
where you don't have fully periodic boundaries, immobile atoms or any
other system modification that impacts the dynamics of the system.

which to choose should be independent. in either time integration, you
have to make certain, that your time intergration is conserving energy
(without fix smd) properly. with your rather lax kspace convergence,
you'll likely run into a problem and unless your 1fs timestep my be
way too large if you have explicit hydrogen atoms in your system.

specific conditions or applications under which these has to be used?

Also, is it possible to apply these ensembles only to the polymer chains,
because the position of atoms in surface is not going to change over
simulation.

again, you can apply the integrator to a subset of atoms. please note
that for npt the application to the volume is always a global
operation. so that may not be a good choice for your simulation.

If I use nvt ensemble for the entire system, after some run, the temperature
is increasing drastically and I get an error "bond atom missing".

does this happen without fix smd? if yes, then you have to correct
your simulation to do reliable time integration first.
if it happens only with fix smd, you have to reconsider your pulling
speed and check how far you are stretching your polymer.
that is best done through visualizing the trajectory.

If I use nvt ensemble (300K) for the polymer chain alone the temperature
fluctuation is so high.. It is between 70 to 600 K.

temperature *will* fluctuate a lot for a limited number of atoms. in
fact, temperature as such is only rigorously defined in the limit of
an infinite number of atoms. the temperature that LAMMPS outputs is
based on the assumption of equipartitioning, which is getting worse
the smaller the number of atoms is.

axel.

Hi everyone,

I used the fix rigid/nve command to study the dynamics of a rigid body.

The key command is like :

fix 11 all rigid/nve molecule langevin 5.0 1.2 1.0 1114551
thermo 1000

thermo_style custom step temp epair emol etotal enthalpy

output of thermodynamics is like:

Step Temp E_pair E_mol TotEng Enthalpy
0 0 0.00021043267 0 0.00021043267 -0.0043702965
1000 0 0.00086645379 0 0.00086645379 0.67084094
2000 0 0.001281543 0 0.001281543 0.49383509
3000 0 0.00096547464 0 0.00096547464 -0.13033507

I can be found the temp is zero through my simulations. I have checked the trajectory via VMD, the dynamics seems reasonable.

I want to know How to output the right temperature in this situation?

Thanks in advance.

Wade

Fix rigid/nve can output a scalar temperature of the collection of rigid
bodies which accounts for the rigid DOFs. See the fix rigid doc page.

Steve

Hi everyone,

I used the fix rigid/nve command to study the dynamics of a rigid body.

The key command is like :

fix 11 all rigid/nve molecule langevin 5.0 1.2 1.0 1114551
thermo 1000
thermo_style custom step temp epair emol etotal enthalpy

output of thermodynamics is like:

Step Temp E_pair E_mol TotEng Enthalpy
0 0 0.00021043267 0 0.00021043267 -0.0043702965
1000 0 0.00086645379 0 0.00086645379 0.67084094
2000 0 0.001281543 0 0.001281543 0.49383509
3000 0 0.00096547464 0 0.00096547464 -0.13033507

I can be found the temp is zero through my simulations. I have checked the trajectory via VMD, the dynamics seems reasonable.‍

I want to know How to output the right temperature in this situation?

Impossible to say with such sparse information and no way to reproduce this. Most likely there is a mistake in your input or your procedure.

Hi, Axel and Steve,

Thanks for your kind reply.

Following the suggestion of Steve, I add the commands in the following, However, It does not work, the temperature is also zero.

Hi, Axel and Steve,

Thanks for your kind reply.

Following the suggestion of Steve, I add the commands in the following,
However, It does not work, the temperature is also zero.

sorry, it works just fine for me. i checked a test input that i have
with fix rigid/nvt and changed it to use fix rigid/nve with langevin
and it works as i would expect.

it must be something else that is messing up things.

Thanks for the reply,

Can we use “Fix setforce” to constrain one end of the polymer chain and use “fix SMD” to pull other end of the polymer chain along z direction perpendicular to the surface.

Thanks for the reply,

Can we use "Fix setforce" to constrain one end of the polymer chain and use
"fix SMD" to pull other end of the polymer chain along z direction
perpendicular to the surface.

fix setforce will do what it does. whether this is the right thing for
your system or not, i don't know because your description is far too
vague.
make some tests to try it out and see if it works as expected. that is
how research works.

axel.