Testing common approximations of neutrino fast flavor conversion

TL;DR

This work tackles how coarse-grained neutrino flavor models handle fast flavor conversion (FFC) in extreme astrophysical environments. Using the Emu quantum kinetic framework with two flavors, it compares driving schemes (sudden, discrete, continuous), periodic subgrid homogenization, and phase-randomization to test effective classical transport and BGK subgrid models. The findings show that sequential-instability driving, spatial homogeneity assumptions, and reliance on instability-driven driving can significantly bias flavor evolution, and that FFC can occur without instabilities or in the presence of subgrid inhomogeneities and coherent phases. The results argue for developing inhomogeneous subgrid asymptotic states or fully self-consistent QKE-based transport to reliably predict neutrino flavor effects in CCSNe and neutron star mergers, with direct implications for observable signals and dynamics.

Abstract

A new chapter is opening in the theory of core-collapse supernovae and neutron star mergers as simulations of these events begin to incorporate fast flavor conversion (FFC) and other forms of neutrino flavor mixing. Using numerical experiments, we show that the approximations of FFC that have been implemented so far are limited by at least two of three factors: (1) approximating continuous evolution as a discrete sequence of instabilities, (2) using spatially homogeneous asymptotic states, and (3) assuming that FFC must be accompanied by instability. The factors we identify in this work will be important considerations as the research area progresses from initial exploratory studies to more quantitatively precise assessments.

Figures (4)

• Figure 1: We test effective classical transport subgrid models by driving fast flavor instabilities (FFIs) through three different mechanisms: sudden, discrete, and continuous emission. In the sudden case, the system is initialized directly in an unstable configuration. In the discrete case, the instability is triggered by injecting eight heavy-neutrino packets, while in the continuous case, it develops through sustained heavy-neutrino emission. In all cases, the emission occurs exclusively in the dashed beam propagating along the $-z$ direction. Simulations in the lower panel experience eight times more emission than those in the upper panel. The distinct driving schemes lead to different final flavor compositions, confirming that the way the instability is driven influences the outcome of flavor evolution. The observed discrepancies highlight the limitations of the two-step approach, where instabilities arise abruptly and subsequently decay. Such simplified procedures fail to fully capture the nonlinear dynamics of flavor conversion. The simulation legends apply to both upper and lower panels.
• Figure 2: Comparison of homogenous flavor conversion paths (BGK and QKE + periodic homogenization) versus inhomogeneous (QKE). The relaxation time $\tau(t)$ is shown for BGK, QKE, and sudden emission (blues in Fig. \ref{['fig:seqins']}). Simulations in the lower panels experience eight times more emission than those in the upper panel. See Sec. \ref{['sec:homog']} for more simulation details. Imposing subgrid homogeneity alters the dynamics, producing large deviations from the behavior predicted by the QKE evolution. Fast flavor conversion can occur without instability or ELN-XLN angular crossing (see QKE bottom panels for $t>1200\,\mu^{-1}$). The emergent relaxation timescale $\tau$ for flavor conversion is larger when the system is driven quasi-statically compare to the sudden cases where $\tau\sim\mu^{-1}$. The simulation legends apply to all panels.
• Figure 3: Reversibility properties of a kinetic neutrino gas undergoing FFC. ELN–XLN crossing is induced by continuous emission heavy neutrinos in the dashed beam followed by absorption the same amount (or until there is not more heavy neutrinos). Simulations in the lower panel experience eight times more emission than those in the upper panel. See emission and absorption term in Eq. \ref{['eq:cont_col_term_1xemission_rever']} for upper panel and Eq. \ref{['eq:cont_col_term_8xemission_rever']} for lower panel. In the absence of flavor conversion, the system is reservable since it returns to its initial state. QKE + homogenization at $t = 2400\,\mu^{-1}$ reset $\mathbf{P}$ to its domain-averaged seeding a random perturbation of amplitude $10^{-4} P_z$ in $P_\bot$. QKE + randomization at $t = 2400\,\mu^{-1}$ randomly rotates $P_\bot$ around $P_z$. The upper ($t\in[2400,4800]\,\mu^{-1}$) and lower ($t\in[1200,2400]\,\mu^{-1}$ and $t\in[3600,4800]\,\mu^{-1}$) panels show that FFC can occur without being accompanied by flavor instability. The simulation legends apply to both upper and lower panels.
• Figure 4: Flavor evolution when subgrid phase information at each location and direction is randomized by rotating the polarization vector $\mathbf{P}$ by a random phase around the $P_z$ axis, while keeping $|\mathbf{P}|$ and $|P_{\perp}|$ fixed. Simulations in the lower panel experience eight times more emission than those in the upper panel. See Sec. \ref{['sec:random']} for more simulation details. The randomization is applied periodically (8 and 16 times). Within the first $2400\,\mu^{-1}$, an ELN--XLN crossing is induced by emitting $n^{+z}_{0}/2$ (top) and $4n^{+z}_{0}$ (bottom) heavy neutrinos in the dashed $-z$ beam. See emission term in Eq. \ref{['eq:cont_col_term_1xemission_rever']} for upper panel and Eq. \ref{['eq:cont_col_term_8xemission_rever']} for lower panel. When subgrid phase information is lost, flavor conversion without ELN-XLN crossing and FFI still occurs but deviates from the QKE evolution, indicating that quantum-coherent phases provide essential feedback in the subsequent flavor dynamics. The simulation legends apply to both upper and lower panels.