Increasing Fisher Information by Moving-Mesh Reconstruction

Reconstruction techniques are commonly used in cosmology to reduce complicated nonlinear behaviours to a more tractable linearized system. We study a new reconstruction technique that uses the Moving-Mesh algorithm to estimate the displacement field from nonlinear matter distribution. We show the performance of this new technique by quantifying its ability to reconstruct linear modes. We study the cumulative Fisher information $I(<k_n)$ about the initial matter power spectrum in the matter power spectra in 130 $N$-body simulations before and after reconstruction, and find that the nonlinear plateau of $I(<k_n)$ is increased by a factor of $\sim 50$ after reconstruction, from $I \simeq 2.5 \times 10^{-5} /({\rm Mpc}/h)^3$ to $I \simeq 1.3 \times 10^{-3}/({\rm Mpc}/h)^3$ at large $k$. This result includes the decorrelation between initial and final fields, which has been neglected in some previous studies. We expect this technique to be beneficial to problems such as baryonic acoustic oscillations, redshift space distortions and cosmic neutrinos that rely on accurately disentangling nonlinear evolution from underlying linear effects.