Re: [ESPResSo-users] Hard Dimers in 2D

[Rudolf Weeber]

Hi,
On Fri, Jan 10, 2014 at 10:00:51AM -0600, Salvador H-V wrote:
> I am interested in the patch to set the Langevin Gamma for rotation, that
> you mentioned in your email.
> 
> I would like to decrease the rotation of the dimers. In my current
> simulations the dumbbells rotate very fast; this is an unphysical scenario
> when particles are supposed to be immersed in a solvent.
If you just want to reduce the average rotational velocity, you can also 
increase the inertial tensor (rinertia). Use three equal values for a sphere.
The thermostat will keep the expectation value of E_kin at 1/2k_B T per degree 
of freedom, so increasing the inertial tensor will decrease the mean square 
velocity. 
(the same goes for mass and linear velocity)

The Langevin gamma is used to set the friction coefficient/viscosity of the 
solvent or the "relaxation time", i.e., after what time the velocity of the 
particle becomes random. 

> Could you please send the patch to me. Also, some guidance of how to
> implement the patch would be great.
I put it on my github repo in the branch gamma-rotation:
https://github.com/RudolfWeeber/espresso/tree/gamma-rotation
You can download it via git.
It is used like this
thermostat langevin <temperature> <gamma_translation> <gamma_rotation>
But be really careful with it. It has not been tested much.
In particular, I'm not at all sure how this interacts with the barostat (NpT 
ensemble).


One other thing:
Restricting the dimers to 2D. 
I ignored the 2D problem in my last email. Using virtual sites to build the 
dimers, you would have to hack the code to achieve this.
I have no idea, how the "rigid bond" (rattle algorithm) interacts with 
part <n> fix 0 0 1 (constraining the particle positions to a plain)
Maybe, try that out first. If it works together (as in giving the correct 
thermodynamic properties for a single dimer and not violating the constant 
distance constraint in the process),  use rigid bonds.
If it doesn't, I can help with the virtual sites version, but it will need some 
programming.


Hope, that helps.
Regards, Rudolf





Regards, Rudolf