1201.2082 (Chris Orban)
Chris Orban
In setting up initial conditions for cosmological N-body simulations there
are, fundamentally, two choices: either maximizing the correspondence of the
initial density field to the assumed fourier-space clustering or, instead,
matching to the real-space clustering. As a stringent test of both approaches,
I perform ensembles of simulations using power law models and exploit the
self-similarity of these initial conditions to quantify the accuracy of the
results. Originally proposed by Pen 1997 and implemented by Sirko 2005, I show
that the real-space motivated approach, which allows the DC mode to vary,
performs well in exhibiting the expected self-similar behavior in the mean
xi(r) and P(k) and in both methods this behavior extends below the scale of the
initial mean interparticle spacing. I also test the real-space method with
simulations of a simplified, powerlaw model for baryon acoustic oscillations,
again with success, and mindful of the need to generate mock catalogs using
simulations I show extensive powerlaw tests for the halo mass function and halo
bias in our simulations. Although requiring a few to many times more
simulations than the standard, fourier-space method to reach a given certainty
on the correlation function or power spectrum, I find that the real-space
method is more reliable for modeling P(k) when the clustering level becomes
significant on the scale of the simulation box. As such a carefully-constructed
real-space approach could be optimal for simulating extremely red power spectra
(n_eff < -2), as in excessively small box simulations to model the "end" of the
CDM hierarchy. I conclude by discussing some possibilities for optimizing the
real-space method for more general use and an appendix demonstrates the
potential for using perturbation theory to model the effect of the box scale on
the simulated growth of structure.
View original:
http://arxiv.org/abs/1201.2082
No comments:
Post a Comment