[Yt-dev] minimum halo mass

Stephen Skory stephenskory at yahoo.com
Mon Mar 16 15:14:47 PDT 2009


Yo,

> We never came to an agreement.

To recap, here is what Matt proposed last:

rho = total_mass / product(root_grid_dimensions)
padding = B * root_dx * minimum_halo_mass / rho

Where B is some safety factor (integer).

I have a small cosmo run I use often: 64^3 root grid, 64^3 part, HOP at default threshold gives the smallest halo at 4.02e12 Msol with 39 particles, so a DM particle is 1.03e11 Msol, which is also rho. Here root_dx = 1/64, so if we were to use 4.02e12 as our minimum_halo_mass: 

padding = B*39/64 = B*0.6

which is a lot of padding. I find that at and above a padding of about 0.05 for this dataset things don't change.

Maybe I'm just dumb, but it's the biggest object we care about, right? So that we can ensure that even the biggest halo (by spatial dimensions) exists fully in at least one padded sub-box. So we'd want something like this:

radius = f(maximum_expected_halo_size) (gives a number in kpc/h)
cells = floor(radius/pf['kpch']/root_dx)
padding = B * cells

where f() is some reasonable mapping of halo mass to halo radius. maximum_expected_halo_size can be calculated from the cosmology and a rough halo mass function. It's not neccessary to do a high-quality estimate, I think.

So... comments? Am I confused? Any ideas of what is a reasonable f() and rough halo mass function should be?

Stephen

 _______________________________________________________
sskory at physics.ucsd.edu           o__  Stephen Skory
http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student
________________________________(_)_\(_)_______________



More information about the yt-dev mailing list