A simple sub-grid model for cosmic ray effects on galactic scales

Many recent numerical studies have argued that cosmic rays (CRs) from supernovae (SNe) or active galactic nuclei (AGNs) could play a crucial role in galaxy formation, in particular by establishing a CR-pressure-dominated circumgalactic medium (CGM). But explicit CR-magnetohydrodynamics (CR-MHD) remains computationally expensive, and it is not clear whether those results can be applied to simulations that do not explicitly treat magnetic fields or resolved interstellar medium phase structure. We therefore present an intentionally extremely simplified 'sub-grid' model for CRs, which attempts to capture the key qualitative behaviors of greatest interest for those interested in simulations or semi-analytical models including some approximate CR effects on galactic (greater than or similar to kpc) scales, while imposing negligible computational overhead. The model is numerically akin to some recently developed sub-grid models for radiative feedback, and allows for a simple constant parametrization of the CR diffusivity and/or streaming speed; it allows for an arbitrary distribution of sources (proportional to black hole accretion rates or star-particle SNe rates or gas/galaxy star formation rates), and interpolates between the limits where CRs escape the galaxies with negligible losses and those where CRs lose most of their energy catastrophically before escape (relevant in e.g. starburst galaxies). The numerical equations are solved trivially alongside gravity in most codes. We compare this to explicit CR-MHD simulations and discuss where the (many) sub-grid approximations break down, and what drives the major sources of uncertainty.

Automated summary

Recent numerical studies suggest that cosmic rays (CRs) from supernovae (SNe) or active galactic nuclei (AGNs) may have a significant impact on galaxy formation by establishing a CR-pressure-dominated circumgalactic medium (CGM). However, explicit CR-magnetohydrodynamics (CR-MHD) simulations are computationally expensive and it is unclear whether these results apply to simulations without explicit treatment of magnetic fields or resolved interstellar medium phase structure. In this study, we propose a simplified 'sub-grid' model for CRs that captures the key qualitative behaviors of interest for simulations or semi-analytical models, while imposing minimal computational overhead.