Fwd: Particles

[Andreas Ipp]

Regarding particles, it should not be necessary to copy all particles
from the neighboring cell - only the ones that are "close enough".

I think, as you said, it is necessary to think through carefully which
cell needs what information at what timestep. For interpolation, one
may need particles that are not too far away (e.g. within the region
of one cell spacing), but probably not beyond that.

The point is that neighboring fields and particles are not calculated
by the current grid. One only needs the minimum information that is
required to update the particles and fields in the current grid, so
currently it would be one extra rows of grid, and particles that lie
(partly) within this one extra row of grid (taking into account their
radius).

Duplicated particles slightly outside of the current grid are *not*
updated by the current grid, but they are updated by the grid to which
they belong. At every time step, the updated information (position,
velocity) is copied from the neighboring grid to the current grid.

Does this make sense? Maybe we need more drawings, as describing this
in words (and I think understanding it) becomes quite complicated.

---------- Forwarded message ----------
From: Jan Kis <jano...@gmail.com>
Date: 2012/6/2
Subject: Re: Particles
To: Andreas Ipp <i...@hep.itp.tuwien.ac.at>


> I don't know whether you also want to think about particles, or whether you first want to concentrate on the grid.

I want to think about them as well. It is better to have it more or
less all thought out because we can not ignore them and when we do not
think about them right from he beginning we might end up with a lot of
rewriting later

> So there is not only the operation of moving one particle from one grid to another grid, but also the operation of duplicating a part of the particles (that lie within some range that also depends on the size of the particle).

I agree but what is more, I thought that everything from the border
cells has to be duplicated (also the particles - for interpolation).
So, from the picture I think we can not ignore the gray particles as
we might need them in this cell for interpolation.

I will maybe draw a more detailed picture on data distribution when I
will write down exactly what data needs to be distributed and when.

Jan

On Thu, May 31, 2012 at 10:03 AM, Andreas Ipp <i...@hep.itp.tuwien.ac.at> wrote:
>
> I don't know whether you also want to think about particles, or whether you first want to concentrate on the grid.
>
> Just like for grids where there is the "1 row of ghost cells" that needs to be copied constantly from one grid to a neighboring grid, also for particles there is a range where particles need to be copied if they have a finite radius.
>
> So there is not only the operation of moving one particle from one grid to another grid, but also the operation of duplicating a part of the particles (that lie within some range that also depends on the size of the particle).
>
> Cheers,
> Andreas
>
> --
> Andreas Ipp
> Institute for Theoretical Physics
> Vienna University of Technology
> Wiedner Hauptstr. 8-10/136
> A-1040 Vienna, Austria
>
> phone: +43-1-58801-13635
> fax:   +43-1-58801-13699
> http://hep.itp.tuwien.ac.at/~ipp/
>
>

[Jan Kis]

Hi Andreas,

Duplicated particles slightly outside of the current grid are *not*
updated by the current grid, but they are updated by the grid to which
they belong. At every time step, the updated information (position,
velocity) is copied from the neighboring grid to the current grid.

This is clear, I agree :) 

Regarding particles, it should not be necessary to copy all particles
from the neighboring cell - only the ones that are "close enough".

I am a little slow I still do not get this. (By cell I understand one cell of the grid - eg. grid 10 X 10 has 100 cells) As far as I understand we do the interpolation as depicted in the picture interpolation.svg (ie. p1 interpolates to A1, A2, B1, B2). Consequently, looking at the picture sharedData.svg in the current grid we need both particle "top 1" and also particle "top 2" to correctly interpolate to cell A1. Am I mistaken here?

Cheers,

Jan

[Andreas Ipp]

Yes, sorry I meant grid instead of cell.

Yes, you need to copy both particles shown in your picture.

If there was a particle 3 in the top cell row of the neighbouring grid, then you would not need to copy that.

[Jan Kis]

Cool, cool, then I think we understand each other :) 

Right now I am writing a text on which data needs to be shared and at what time. Hopefully, I will have something meaningful by tomorrow.