Table of Contents
Fetching ...

Resolving Cosmic Neutrino Structure: A Hybrid Neutrino N-body Scheme

Jacob Brandbyge, Steen Hannestad

TL;DR

This work introduces a hybrid grid- and N-body neutrino simulation framework that resolves non-linear neutrino clustering on Mpc scales. By sampling neutrinos in 15 momentum bins and converting selected bins from a linear grid to N-body particles at an optimally chosen redshift, the method preserves momentum-dependent transfer functions while suppressing noise and enabling non-linear structure formation to be captured efficiently. The approach reproduces the non-linear total matter power spectrum with high accuracy (≈0.2% for the matter power at k ≲ 1 h Mpc^{-1} and 2–4% for neutrinos at the same scales) and demonstrates effective control of leakage across momentum boundaries by converting multiple bins simultaneously. This hybrid scheme significantly reduces computational demands compared to full particle-based neutrino simulations and enables probing of small-scale neutrino features within halos, with plans to extend to sub-Mpc scales in follow-up work.

Abstract

We present the first simulation capable of resolving the structure of neutrino clustering on Mpc scales. The method combines grid- and particle-based methods and achieves very good accuracy on both small and large scales, while keeping CPU consumption under control. Such simulations are not only ideal for calculating the non-linear matter power spectrum but also particularly relevant for studies of how neutrinos cluster in galaxy- or cluster-sized halos. We perform the largest neutrino N-body simulation to date, effectively containing 10 different neutrino hot dark matter components with different thermal properties.

Resolving Cosmic Neutrino Structure: A Hybrid Neutrino N-body Scheme

TL;DR

This work introduces a hybrid grid- and N-body neutrino simulation framework that resolves non-linear neutrino clustering on Mpc scales. By sampling neutrinos in 15 momentum bins and converting selected bins from a linear grid to N-body particles at an optimally chosen redshift, the method preserves momentum-dependent transfer functions while suppressing noise and enabling non-linear structure formation to be captured efficiently. The approach reproduces the non-linear total matter power spectrum with high accuracy (≈0.2% for the matter power at k ≲ 1 h Mpc^{-1} and 2–4% for neutrinos at the same scales) and demonstrates effective control of leakage across momentum boundaries by converting multiple bins simultaneously. This hybrid scheme significantly reduces computational demands compared to full particle-based neutrino simulations and enables probing of small-scale neutrino features within halos, with plans to extend to sub-Mpc scales in follow-up work.

Abstract

We present the first simulation capable of resolving the structure of neutrino clustering on Mpc scales. The method combines grid- and particle-based methods and achieves very good accuracy on both small and large scales, while keeping CPU consumption under control. Such simulations are not only ideal for calculating the non-linear matter power spectrum but also particularly relevant for studies of how neutrinos cluster in galaxy- or cluster-sized halos. We perform the largest neutrino N-body simulation to date, effectively containing 10 different neutrino hot dark matter components with different thermal properties.

Paper Structure

This paper contains 11 sections, 8 equations, 6 figures, 1 table.

Figures (6)

  • Figure 1: Momentum dependent neutrino TFs at $z=0$ for $\sum m_\nu = 1.2 \, {\rm eV}$.
  • Figure 2: Density grids from the hybrid simulation $A_1$ with $\sum m_\nu = 1.2 \, {\rm eV}$ found with the adaptive smoothing length kernel from monaghan. Top mosaic: $z = 4$. Bottom mosaic: $z = 0$. Top row: CDM, neutrino particles and neutrino grid. Middle row: $q/T =$ 1, 2 and 3. Bottom row: $q/T =$ 5, 7 and 10. Bottom mosaic: The CDM, neutrino particle, $q/T =$ 1, 2 and 3 density grids have been raised to the power of $0.25$ to enhance the dynamical contrast. The density slices have a thickness of $20 h^{-1} \, {\rm Mpc}$ and are $512 h^{-1} \, {\rm Mpc}$ on a side.
  • Figure 3: Left: Difference between the linear simulation $A_3$ (green) and the full non-linear $A_2$ (red) and hybrid $A_1$ (blue) methods at various redshifts. Right: Absolute neutrino power spectra at $z = 0$.
  • Figure 4: $z = 0$. Left: Hybrid simulations $B_i$ with $q_{\rm cut}/T$ ranging from 4 to 15 compared to $q_{\rm cut}/T = 15$ ($B_6$). Right: Corresponding neutrino power spectra.
  • Figure 5: $\sum m_\nu = 1.2 \, {\rm eV}$. Left: Matter power spectra from simulations $C_i$ at $z = 0$ with grid-to-particle conversion factors of $f_{\rm flow} = 2$, 4, 8 and 16 relative to $f_{\rm flow} = 8$ ($C_3$). Right: Corresponding neutrino power spectra.
  • ...and 1 more figures