Speaker
Mr
Devon Powell
(SLAC/Stanford)
Description
Traditional N-body simulations discretize the dark matter
distribution into an ensemble of point particles. However, estimating
the local density for a set of point particles is difficult due to
Poisson noise. Abel et al. (2012) instead describe the phase-space
distribution of dark matter as a 3D manifold tessellated into tetrahedra.
This has the advantage of giving an unambiguous value for the density
everywhere in configuration-space. Analyzing such a collection of
tetrahedra requires a method for projecting a tetrahedron onto a
uniform grid (voxelization). Various schemes have been tried (e.g.
Angulo et al. 2013, Hahn et al. 2013), though each has its advantages
and drawbacks.
I present here a new method for voxelizing polytopes. This
method computes the exact intersection volume between the polytope and
each voxel, so it is noiseless and exactly mass-conserving. In
addition, polynomial functions defined over the polytope can be
exactly voxelized, giving the ability to apply mass interpolation schemes
over the phase-space sheet.
My implementation of this method yields an unprecedentedly smooth
and continuous dark matter density field, with exciting prospects for
studying the phase-space structure of dark matter haloes, WIMP
annihilations, gravitational lensing, and more.
Primary author
Mr
Devon Powell
(SLAC/Stanford)
Co-author
Prof.
Tom Abel
(SLAC)
