Using the Zeldovich dynamics to test expansion schemes

TL;DR

This paper evaluates various expansion schemes for gravitational clustering using the exact solution of Zeldovich dynamics, revealing limitations in their ability to accurately model nonlinear regimes and response functions.

Contribution

It systematically compares perturbative and nonlinear expansion methods against the exact Zeldovich solution, highlighting their convergence issues and limitations in nonlinear regimes.

Findings

Most systematic expansions fail to capture decay of response in nonlinear regime.

Pade approximants provide bounded response at all orders.

Standard perturbation theory and certain nonlinear schemes are most effective.

Abstract

We apply various expansion schemes that may be used to study gravitational clustering to the simple case of the Zeldovich dynamics. Using the well-known exact solution of the Zeldovich dynamics we can compare the predictions of these various perturbative methods with the exact nonlinear result and study their convergence properties. We find that most systematic expansions fail to recover the decay of the response function in the highly nonlinear regime. ``Linear methods'' lead to increasingly fast growth in the nonlinear regime for higher orders, except for Pade approximants that give a bounded response at any order. ``Nonlinear methods'' manage to obtain some damping at one-loop order but they fail at higher orders. Although it recovers the exact Gaussian damping, a resummation in the high-k limit is not justified very well as the generation of nonlinear power does not originate from a finite range of wavenumbers (hence there is no simple separation of scales). No method is able to recover the relaxation of the matter power spectrum on highly nonlinear scales. It is possible to impose a Gaussian cutoff in a somewhat ad-hoc fashion to reproduce the behavior of the exact two-point functions for two different times. However, this cutoff is not directly related to the clustering of matter and disappears in exact equal-time statistics such as the matter power spectrum. On a quantitative level, the usual perturbation theory, and the nonlinear scheme to which one adds an ansatz for the response function with such a Gaussian cutoff, are the two most efficient methods. These results should hold for the gravitational dynamics as well (this has been checked at one-loop order), since the structure of the equations of motion is identical for both dynamics.