[FLASH-USERS] Destroying particles

[Seyit Hocuk]

Dear Anshu,

The problem persists, but now it shows up in Grid_sortParticles

line 131: particles(:,m)=gr_ptSourceBuf(:,i)
segmentation fault occurred

I think the problem is the same as using nonexistent positions. I'm 
gonna do a few more tests.

Thanks,
Seyit





Anshu Dubey wrote:
>
> On Wed, Sep 1, 2010 at 4:57 AM, Seyit Hocuk <seyit at astro.rug.nl 
> <mailto:seyit at astro.rug.nl>> wrote:
>
>     Thanks Anshu,
>
>     What exactly did you mean by time history?
>
>  
> By time history I mean the trajectory of the particle. Once you set 
> its block number to NONEXISTENT, that particle will no longer be 
> output. I am not entirely sure how the post processing will behave if 
> a particle tag disappears mid-way through a run. I can think of ways 
> of getting around it, but you may not need them.
>
>     For sorting the partiles, do you mean re-tag them?
>
> There is a routine called Grid_sortPartilcles, it moves the particles 
> with invalid block number to the end of the array and updates the 
> local count of particles to those with valid block numbers.
>  
>
>     I will try to change the blocknumber instead and see if it helps.
>
>     Kind regards,
>     Seyit
>
>
>
>     Anshu Dubey wrote:
>
>         One possibility if you don't need the time history is to set
>         the particles' block number to NONEXISTENT, and then sort the
>         particles data structure before you continue processing.
>
>         On Tue, Aug 31, 2010 at 9:22 AM, Seyit Hocuk
>         <seyit at astro.rug.nl <mailto:seyit at astro.rug.nl>
>         <mailto:seyit at astro.rug.nl <mailto:seyit at astro.rug.nl>>> wrote:
>
>            Dear Community,
>
>            I have a simple question, one that we discussed before, I
>         think.
>
>            Here goes:
>            If you have a certain amount of particles, and want to
>            reduce/destroy some of them during simulation, what is the best
>            approach to do this. I alwas was placing them outside the box
>            (negative x,y,z-positions), however, this causes occasional
>            crashes (coming from "pt_mapFromMeshWeighted" line 146), and
>            running in debug mode already was showing this. I think that
>            putting 0 mass is also not a good solution.
>
>            Is there any other way?
>
>            Kind regards,
>            Seyit
>
>
>
>
>
>
>         -- 
>         Anshu Dubey
>         Associate Director and CS/Applications Group Leader
>         ASC/Flash Center
>         Fellow, Computation Institute
>         University of Chicago and Argonne National Laboratory
>         773 834 2999 (office)
>         312 420 0033 (mobile)
>         773 834 3230 (fax)
>
>
>
>
>
> -- 
> Anshu Dubey
> Associate Director and CS/Applications Group Leader
> ASC/Flash Center
> Fellow, Computation Institute
> University of Chicago and Argonne National Laboratory
> 773 834 2999 (office)
> 312 420 0033 (mobile)
> 773 834 3230 (fax)