Re: [ESPResSo-devel] A bug in Espresso

[Ulf Schiller]

Hi,

On 07/03/2012 04:45 PM, Stefan Kesselheim wrote:
> Hi,
>
> On 07/03/2012 04:29 PM, Ivan Cimrak wrote:
> >  Hello,
> >
> >  I noticed some strange behaviour. I created an lbfluid, initiated it
> >  with some nonzero velocity and let it evolve (EXTERNAL_FORCES were
> >  switched off). I printed the velocity profile of the fluid
> >  throughout the simulation. Then I put a particle in the fluid and
> >  again printed the velocity of the fluid. I supposed, that the outcome
> >  should be different, since any particles influence the behaviour of
> >  the fluid. But the output was identical.
>
> >  Then I did the same with EXTERNAL_VELOCITY switched on and the
> >  outputs were different.
>
> What is EXTERNAL_VELOCITY? I have never heard of that and the Espresso
> sources show no sign of it.
>
> >  I think that in file lb.c, in procedure lb_viscous_coupling, after
> >  local_f is increased by
> >
> >  local_f[0] += delta[3*x+0]*delta[3*y+1]*delta[3*z+2]*delta_j[0];
> >  local_f[1] += delta[3*x+0]*delta[3*y+1]*delta[3*z+2]*delta_j[1];
> >  local_f[2] += delta[3*x+0]*delta[3*y+1]*delta[3*z+2]*delta_j[2];
> >
> >  also the flag lbfields[node_index[(z*2+y)*2+x]].hasforce shold be set
> >  to 1. Otherwise, in procedure lb_collide_stream, the procedure
> >  lb_apply_forces is not called
> >
> >  if (lbfields[index].has_force) lb_apply_forces(index, modes);
> >
> >  When EXTERNAL_FORCES are switched on, this problem does not occur.
>
> OK. I just realize that I never checked the LB without having
> EXTERNAL_FORCES switched on. Probably if it is deactivated the coupling
> just does not work, because the "has_force" flag is not set/read
> properly. I will fix that soon, but please switch on EXTERNAL_FORCES
> meanwhile. I tend to just remove the has_force flag, because the
> computational cost of calculating the influence of the force is not very
> large compared to the rest of the LB calculation.

Note that there is a slight difference between external forces and 
coupling forces. The external forces are applied once every LB update 
and enter the definition of the hydrodynamic velocity, while the 
coupling forces have to be applied to the fluid in every MD update and 
are neglected in the hydrodynamic velocity. Otherwise (total) momentum 
is not conserved during an MD time step. In principle, the coupling 
forces would enter the hydrodynamic velocity as well, but that brings up 
a self-consistency problem as the forces are themselves velocity 
dependent. To avoid an iterative update I usually neglect the coupling 
force when determining the momentum density.

Best,
Ulf

--
Dr. Ulf D. Schiller                        Building 04.16, Room 3006
Institute of Complex Systems (ICS-2)       Phone:   +49 2461 61-6144
Forschungszentrum Jülich, Germany          Fax:     +49 2461 61-3180

smime.p7s
Description: S/MIME Cryptographic Signature