Re: [ESPResSo] Position-dependent Langevoin thermostat

[Axel Arnold]

Hi!

> (1) I want to make each monomer in a cluster (i.e. individual particle
> I create in espresso) aware its first neighbors (at the very least he
> should know how many they are).
> I think this is doable since this is (to my understanding) already
> implemented in the DPD thermostat, where you can specifiy a cut-off
> distance the dissipative and random force. In other words, each
> particle knows how many other particles it finds within a distance
> r_cut.

Well, a particle never knows about all its neighbors; however, Espresso has 
Verlet lists which allow to loop over all particles that are closer than a 
certain distance. You could loop over all such particles pairs and increase 
for each particle in a pair the number of neighbors. Then, you can indeed get 
the number of neighbors.

> (2)Once I have implemented what I have described in (1), then the idea
> is to be able to express the friction felt by each individual monomer
> as a function of its number of (first) neighbors.The value of the
> corresponding noise term should be given again by the
> fluctuation-dissipation theorem. Or the other way around: given the
> noise term, that should fix the friction.

This should be doable; however, you need to fiddle around with the integration 
loop, since before assigning the random forces in the very beginning of the 
integration loop, you need to do the counting.

> (3) I would like (in order to be able to investigate the diffusion
> coefficient of a cluster) to track simultaneously many trajectories of
> copies of the same cluster. For instance: start a simulation with two
> clusters A and B. Clusters A and B should be "transparent" to each
> other (i.e. the do not feel any reciprocal interaction whatsoever and
> each of them can go through the other effortlessly) and for both A and
> B I would like to be able to do what described in (1) and (2).

Generally, the pair loops always run over all pairs of adjacent particles, 
however, interactions are based on particle types. If you use different types 
for all clusters, they can be transparent to each other; your counting 
routine would need to just count particles of the appropriate type. However, 
you cannot have more than a couple of 10 clusters at once; Espresso uses a 
matrix of interaction parameters, and that will quickly get big if you have a 
couple of 100 particle types.

Many regards,
Axel Arnold

-- 
Dr. Axel Arnold 
Fraunhofer SCAI
Schloss Birlinghoven, 53754 Sankt Augustin, Germany
Tel: +49 2241 14 2575