[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ESPResSo-users] Fwd: Question of NVT ensemble
Re: [ESPResSo-users] Fwd: Question of NVT ensemble
Sat, 08 Nov 2014 22:43:56 +0000
Mozilla/5.0 (X11; Linux i686; rv:31.0) Gecko/20100101 Thunderbird/31.2.0
note that it is essential to apply the thermalization to the peculiar
velocities, i.e., excluding any component along external degrees of
freedom. Otherwise one may get the flying ice cube effect that Florian
mentioned even for Langevin-like thermostats. Please have a look at this
which discusses various thermostats in detail.
Also keep in mind that polymer dynamics can be very different depending
on whether you take hydrodynamic correlations into account or not -
think Rouse vs Zimm scaling. I don't think you will get meaningful
answers for your problem without hydrodynamics.
Ulf D. Schiller
Centre for Computational Science
University College London
20 Gordon Street
London WC1H 0AJ
On 11/08/2014 10:13 PM, Florian Weik wrote:
> ---------- Forwarded message ----------
> From: *Joe* <address@hidden <mailto:address@hidden>>
> Date: Tue, Nov 4, 2014 at 5:17 PM
> Subject: Re: [ESPResSo-users] Question of NVT ensemble
> To: Florian Weik <address@hidden
> Hi Florian and Xikai and others,
> Thanks for the information. The system I would like to simulate is a
> long, linear, charged polymer chain moving under inhomogeneous electric
> field in salt solution and untangle by hitting obstacles. To make this
> cheap, I don't think I can apply any solvent, including LB method.
> So far it has L-J potential, linear elasticity, bending elasticity as
> well as Coulomb force. I would like to see how long does it take the
> chain to untangle as well as how it untangles. It is a system not close
> to equilibrium. But I do need some random motion of each bead so that
> the polymer chain is physical.
> The options I think I have are: 1. scale the velocity of the bead so
> that the temperature is a constant. 2. scale the relative velocity to
> the center of the mass, so that the vibrational energy is conservative.
> 3. apply a Langevin thermostat so that the beads feel friction between
> each other and have the some random impulse from the solvent.
> The option 2 and 3 seem to be the two best candidates. But will the
> option 3 be better for this?
> On Nov 3, 2014, at 5:39 AM, Florian Weik wrote:
>> Hello Joe,
>> as Xikai Jiang wrote ESPResSo does not implement velocity rescaling
>> thermostats (one of the reasons for that is that there are well known
>> difficulties with this approach, if you don't now about that you
>> should google for "Flying ice cube" effect.). The default thermostat
>> in ESPResSo is a Langevin thermostat. DPD can also be used, but this
>> is not the default approach and should only be used if the other
>> features of DPD are needed. Langevin dynamics introduces a friction
>> term into the equations of motion for the particles which is
>> proportional to the velocity of the particle. The translational
>> velocity you get is the one at which the electric force on your chain
>> and the friction force on its particles are in balance. As such the
>> stationary velocity you get depends on the friction you choose. This
>> somewhat models the interaction between molecules and the solvent
>> which in typical physical situations is present in reality. The DPD
>> thermostat can be configured in such a way that it only has friction
>> on the particle-particle interactions, so that the center of mass
>> movement hast no dampening on it. In this case the translational
>> velocity will diverge with time, because the external field keeps
>> pumping energy into the system but no energy is dissipated.
>> Typical (coarse grained) simulations of charged polymers use the
>> Langevin thermostat, or something more elaborate like a
>> Lattice-Boltzmann fluid if hydrodynamic interactions should be
>> explicitly modeled.
>> The choice of the thermostat is not a merely technical one, but should
>> correspond to the physics of you system. Please read the relevant
>> sections of the ESPResSo User's guide before you start simulations. If
>> you want further help, you should describe more closely to what end
>> you want to do your simulation and what physics you want to capture.
>> On Thu, Oct 30, 2014 at 6:00 PM, Joe <address@hidden
>> <mailto:address@hidden>> wrote:
>> Hi Guys,
>> If I would like to simulate a charged chain move under the
>> external electric field and using NVT ensemble, will it cause a
>> reduction of the translational velocity? From the text book I know
>> that the way NVT works is keep rescaling the velocities to keep
>> the temperature constant. But in this scenario, will the drift
>> velocity also be rescaled?
>> How to avoid this problem?
>> Florian Weik, Dipl.-Phys.,
>> Institut für Computerphysik, Allmandring 3, D-70569 Stuttgart
>> Phone: +49-711-685-67703 <tel:%2B49-711-685-67703>
>> Public Key 0x0562F11D Fingerprint 3294 6360 EC93 37A3 BD40 F597 0BAD
>> 3AF8 0562 F11D
> Florian Weik, Dipl.-Phys.,
> Institut für Computerphysik, Allmandring 3, D-70569 Stuttgart
> Phone: +49-711-685-67703
> Public Key 0x0562F11D Fingerprint 3294 6360 EC93 37A3 BD40 F597 0BAD
> 3AF8 0562 F11D