User-defined variable in replica exchange simulations

Deal all,

I want to run a replica exchange molecular dynamics simulation for a finite size system using spherical boundary conditions (wall/region sphere for example). However I would like to have a ``temperature dependent’’ radius.
I understand that in the LAAMPS implementation it is the temperatures that are exchanged and not the actual replicas.
With the current implementation, is it possible to exchange the value of a user defined variable (in my case here the radius of the sphere) at the same time as the value for the temperature or do I need to modify the code to do that ?

Thanks for your help,

Cyril

Deal all,

I want to run a replica exchange molecular dynamics simulation for a
finite size system using spherical boundary conditions (wall/region sphere
for example). However I would like to have a ``temperature dependent’’
radius.
I understand that in the LAAMPS implementation it is the temperatures that
are exchanged and not the actual replicas.

​that is true for any REMD implementation i have seen. it is far too
convenient to swap temperatures instead of coordinates.

With the current implementation, is it possible to exchange the value of a
user defined variable (in my case here the radius of the sphere) at the
same time as the value for the temperature or do I need to modify the code
to do that ?

​when you are doing replica exchanges with volume change, you need to
include the work due to volume change under pressure, p*deltaV, in the
acceptance criterion computation. that is one of the reasons why the temper
command only accepts thermostats/integrators without volume change.

​axel.​

Hi, Cyril,

I wrote a VT-REMD module for LAMMPS before. Yet the volume exchange only applies to Cartesian coordinate. You need to code specifically for the spherical wall in your system.

Just for your inspiration, paper and source code is as followed:

http://www.sciencedirect.com/science/article/pii/S0010465514004044

LC Liu