[FLASH-USERS] Smoothing of Particles over Multiple Blocks

[Seyit Hocuk]

Hi Chalence,

Flash doesn't have shortrange forces (yet), doens't that work in favor 
in this case? The gravitational potential is calculated by mapping the 
particles onto mesh, on which the gravity is solved. I expect that the 
gravitational potential already smooths the particle contribution a 
little bit out. Anyone care to comment on that?

Kind regards,
Seyit




Chalence Safranek-Shrader wrote:
> Hi Community,
>
> I'm running a hydrodynamic simulation with FLASH's built in cosmology 
> module, adapted mainly from the pancake test problem. In this problem, 
> baryons are represented by a fluid while dark matter is represented by 
> discrete particles which only interact gravitationally.
>
> The goal of our simulations is to probe star formation in high 
> redshift protogalaxies - this requires extreme resoution (sub parsec) 
> in the baryons. The simulation is initialized so that each individual 
> zone has one particle in it. However, as the simulation refines, the 
> grid zones inevitably far outweigh the number of dark matter 
> particles. This is unphysical and causes the particles to act as high 
> gravity accretion centers which artificially funnel baryons to them. 
> This is a common problem in cosmological simulations and has routinely 
> been solved by smoothing the influence of the dark matter particle at 
> some level of refinement (or length scale). This is the solution we're 
> searching for but are having a difficult time implementing it in FLASH.
>
> Currently there is a particle mapping routine (pt_mapOneParticle, and 
> others) which has the capability to smooth a particle over its block, 
> but not over multiple blocks. We also considered mapping the particles 
> to a lower level of refinement and then prolong that mapping to all 
> its child blocks, but there appears to be no routine capable of that 
> either.
>
> Does anyone have any suggestions or solutions to this problem? Any 
> thoughts or insight would be extremely helpful.
>
> Thanks very much,
>
> Chalence Safranek