Langevin thermostat?

Hi,

I want to simulate a system of polymer in water, and I want choose a thermostat to start with equilibration.

I found that Langevin is suitable for solute and solvent systems. Meanwhile, in the manual I found:

“Apply a Langevin thermostat as described in (Schneider) to a group of atoms which models an interaction with a background implicit solvent.”

Does it mean that Langevin is not suitable for explicit solvents (my case)?

I want to use NVE+Langevin(thermostat) then NVE+Langevin(thermostat) + Berendsen(barostat) for the equilibration and simulation.

With Langevin dynamics, there is no explicit water, as in, there are not actually water molecules in your system. Instead, their effects on the system are coarse-grained into friction and collision forces acting on the particles in the system. Because of this model, there are for example no charges/electric dipoles (other than the ones present on your polymer, if any), no real hydrophobic interactions (although they can be modelled in other ways), etc…

It is up to you to judge whether or not this approximation will give you the results you want. If you do need to explicitly model the water molecules then you have to think long and hard about which water model to use and then add molecules of this model to your system.

2015-02-03 11:40 GMT+01:00 Xiaolin Xu <[email protected]>:

Hi,

I want to simulate a system of polymer in water, and I want choose a
thermostat to start with equilibration.

I found that Langevin is suitable for solute and solvent systems.
Meanwhile, in the manual I found:

*"Apply a Langevin thermostat as described in (Schneider) to a group of
atoms which models an interaction with a background implicit solvent."*

Does it mean that Langevin is not suitable for explicit solvents (my
case)?

there's no direct problem. It's just that a Langevin thermostat using
random kicks, which is "just like" solvent molecules colliding with the
thermostatted molecules. But you could also use it to thermostat a solvent
system. It's usually a good choice for equilibration purposes, but of
course it's your taks to find out whether that "usually" applies for your
particular system.

Kristof

Thank you for your comments.

I used this commands:

fix Eq1 all nve
fix Eq11 all langevin 290 290 50 48279

But the thermo style command shows the temp as ~261. Do you have any comment on this? Thanks in advance.

2015-02-03 12:54 GMT+01:00 Xiaolin Xu <[email protected]>:

Thank you for your comments.

I used this commands:

fix Eq1 all nve
fix Eq11 all langevin 290 290 50 48279

But the thermo style command shows the temp as ~261. Do you have any
comment on this? Thanks in advance.

There is no answer possible with such little information. How long did you
equilibrate? How long is your time constant (what is the "50", what
units?)? How does the temperature change during equilibration.
Equilibration and thermostating are standard MD procedures, so you should
try to learn about these before you try actual production work.

Kristof

Hi,

I want to simulate a system of polymer in water, and I want choose a
thermostat to start with equilibration.

I found that Langevin is suitable for solute and solvent systems. Meanwhile,
in the manual I found:

"Apply a Langevin thermostat as described in (Schneider) to a group of atoms
which models an interaction with a background implicit solvent."

Does it mean that Langevin is not suitable for explicit solvents (my case)?

you can use a langevin thermostat, but you have to be very careful
about the time constant. the shorter the time constant, the larger the
friction term and the more random noise is added and the less
realistic your simulation gets. for equilibration, it is usually not a
big problem. in fact, a dissipative thermostat like langevin is
usually a better choice at the beginning, as it helps to quickly
remove unphysical "hot spots". you may also consider fix temp/csvr
which has a similar philosophy, but avoids the less desirable friction
term for explicit solvent.

I want to use NVE+Langevin(thermostat) then NVE+Langevin(thermostat) +
Berendsen(barostat) for the equilibration and simulation.

i would advise against using langevin for production unless you use a
very long time constant. a nose-hoover thermostat via fix nvt is
usually the cleaner choice.