LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] Ncount bin sensitivity
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] Ncount bin sensitivity

From: "Roberts, Andrew" <A.P.Roberts@...5884...>
Date: Wed, 5 Jul 2017 15:18:25 +0000

Hi Steve,

Yes I just checked that the number of particles is 1002 as defined by the number of counts in each of the 20 bins across the profile and yes, this equals the total number of particles that were originally created. To prove it, this is the snapshot of the polydisperse system:

I have uploaded my profile in comparison with the literature:

There seems to be some kind of crystallization near the walls for the LAMMPS simulation, although the central flow region seems to be fairly linear. The daCruz case also contains no friction - presumably this means that the particles have no tangential force on them and if there is also no tangential damping, they cannot rotate. Could a tightly packed wall encourage crystallisation if the particles cannot rotate free of this zone - like raking the particles through the flow? I noticed that the literature ignores the zones close to the wall and has possibly rescaled the profile to show only the linear zone, but I'm not sure if that is the reason. Do you know what could cause this problem? Presumably increased polydispersity would cause more slippage at the wall and less polydispersity would cause more crystallisation?

But why would re-ordering the wall particles influence the velocity profile? - but it's true that the wall particles are added first in order, then the flow particles are randomly placed between them. Should the wall particles be added with random id numbers?

Also, why is there an excluded volume near the walls? - the velocity profile is being computed for all atoms in the simulation.

My input script looks like this:

Kind regards,


From: Steve Plimpton <sjplimp@...24...>
Sent: 04 July 2017 00:59:00
To: Axel Kohlmeyer
Cc: Roberts, Andrew;
Subject: Re: [lammps-users] Ncount bin sensitivity
Also, is the total count for all the bins equal
to the total # of particles you expect to be counting?

>From one snapshot in a dump file, you should
be able to manually count the particles in the
first/last bin and verify what LAMMPS is counting for the same.
I.e. use fix ave/time with no averaging.


On Mon, Jul 3, 2017 at 9:40 AM, Axel Kohlmeyer <akohlmey@...29....> wrote:
On Sat, Jul 1, 2017 at 7:41 PM, Roberts, Andrew
<A.P.Roberts@...5884...> wrote:
> Hello,
> I am new to LAMMPS, but I have generated a Couette flow and the profile is
> almost linear (attached), however the profile is slightly non-linear near
> x=0.1 and 0.9. I noticed that the number of counts (Ncount) is approximately
> half at x=0.1 and 0.9. Is there any reason why there should be a discrepancy
> at the ends of the profile? These are the lines I am using to generate the

my guess would be, that this is likely due to the ordering and
excluded volume near the walls. fix ave/chunk averages over points,
i.e. the position of the center of your particles. just make some
snapshots of your system and check manually.


> profile:
> run  100000
> compute cc1 flow chunk/atom bin/1d y center 0.05 units reduced
> fix 8 flow ave/chunk 2 50000 1100000 cc1 vx vz norm sample file vel.profile
> run  1000000
> Is there any way to make the number of counts the same in all bins? Or is
> there another reason for the difference - e.g. bin sensitivity? (I think I
> have enough timesteps)
> Kind regards,
> Andrew
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites,!
> _______________________________________________
> lammps-users mailing list

Dr. Axel Kohlmeyer  akohlmey@...43...4...
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.

Check out the vibrant tech community on one of the world's most
engaging tech sites,!
lammps-users mailing list