Numerical and Perturbative Computations of the Fuzzy Dark Matter Model

TL;DR

This paper compares numerical and perturbative methods for modeling fuzzy dark matter, highlighting their strengths and limitations, and validates the models against Lyman-alpha forest data at high redshift.

Contribution

It provides a comprehensive comparison of Schrödinger-Poisson and fluid solvers, and develops perturbative calculations for the FDM power spectrum, enhancing understanding of nonlinear structure formation.

Findings

Schrödinger-Poisson solver accurately captures nonlinear dynamics.

Fluid solver fails in destructive interference regimes.

Perturbation theory agrees with simulations in mildly nonlinear regime.

Abstract

We investigate nonlinear structure formation in the fuzzy dark matter (FDM) model using both numerical and perturbative techniques. On the numerical side, we examine the virtues and limitations of a Schrodinger-Poisson solver (wave formulation) versus a fluid dynamics solver (Madelung formulation). We also carry out a perturbative computation of the one-loop mass power spectrum. We find that (1) in many cases, the fluid dynamics solver is capable of producing the expected interference patterns, but it fails where destructive interference causes the density to vanish which generally occurs in the nonlinear regime. (2) The Schrodinger-Poisson solver works well in all test cases, but it is demanding in resolution: one must resolve the small de Broglie scale to obtain the correct dynamics on large scales. (3) We compare the mass power spectrum from perturbation theory against that from the Schrodinger-Poisson solver, and find good agreement in the mildly nonlinear regime. Compared with fluid perturbation theory, wave perturbation theory has a more limited range of validity. (4) As an application, we compare the Lyman-alpha forest flux power spectrum obtained from the Schrodinger-Poisson solver versus one from an N-body simulation (which is often used as an approximate method to make predictions for FDM). At redshift 5, the two, starting from the same initial condition, agree to better than 10 % on observationally relevant scales as long as the FDM mass exceeds $2 \times 10^{-23}$ eV.