Galactic winds regulate star formation in disk galaxies and help to enrich the circum-galactic medium. They are therefore crucial for galaxy formation, but their driving mechanism is still poorly understood. Recent studies have demonstrated that cosmic rays (CRs) can drive outflows if active CR transport is taken into account. Using hydrodynamical simulations of isolated galaxies with virial masses between $10^{10}$ and $10^{13}\mathrm{~M_\odot}$, we study how the properties of CR-driven winds depend on halo mass. CRs are treated in a two-fluid approximation and their transport is modelled through isotropic or anisotropic diffusion. We find that CRs are only able to drive mass-loaded winds beyond the virial radius in haloes with masses below $10^{12}\mathrm{~M_\odot}$. For our lowest examined halo mass, the wind is roughly spherical and has velocities of $\sim20\mathrm{~km\;s^{-1}}$. With increasing halo mass, the wind becomes biconical and can reach ten times higher velocities. The mass loading factor drops rapidly with virial mass, a dependence that approximately follows a power-law with a slope between $-1$ and $-2$. This scaling is slightly steeper than observational inferences, and also steeper than commonly used prescriptions for wind feedback in cosmological simulations. The slope is quite robust to variations of the CR injection efficiency or the CR diffusion coefficient. In contrast to the mass loading, the energy loading shows no significant dependence on halo mass. While these scalings are close to successful heuristic models of wind feedback, the CR-driven winds in our present models are not yet powerful enough to fully account for the required feedback strength.

The absence of large cooling flows in cool core clusters appears to require self-regulated energy feedback by active galactic nuclei (AGNs) but the exact heating mechanism has not yet been identified. Here, we analyse whether a combination of cosmic ray (CR) heating and thermal conduction can offset radiative cooling. To this end, we compile a large sample of 39 cool core clusters and determine steady state solutions of the hydrodynamic equations that are coupled to the CR energy equation. We find solutions that match the observed density and temperature profiles for all our clusters well. Radiative cooling is balanced by CR heating in the cluster centres and by thermal conduction on larger scales, thus demonstrating the relevance of both heating mechanisms. Our mass deposition rates vary by three orders of magnitude and are linearly correlated to the observed star formation rates. Clusters with large mass deposition rates show larger cooling radii and require a larger radial extent of the CR injection function. Interestingly, our sample shows a continuous sequence in cooling properties: clusters hosting radio mini halos are characterised by the largest cooling radii, star formation and mass deposition rates in our sample and thus signal the presence of a higher cooling activity. The steady state solutions support the structural differences between clusters hosting a radio mini halo and those that do not.

Self-regulated feedback by active galactic nuclei (AGNs) appears to be critical in balancing radiative cooling of the low-entropy gas at the centres of galaxy clusters and in regulating star formation in central galaxies. In a companion paper, we found steady-state solutions of the hydrodynamic equations that are coupled to the CR energy equation for a large cluster sample. In those solutions, radiative cooling in the central region is balanced by streaming CRs through the generation and dissipation of resonantly generated Alfv{\'e}n waves and by thermal conduction at large radii. Here we demonstrate that the predicted non-thermal emission resulting from hadronic CR interactions in the intra-cluster medium exceeds observational radio (and gamma-ray) data in a subsample of clusters that host radio mini halos (RMHs). In contrast, the predicted non-thermal emission is well below observational data in cooling galaxy clusters without RMHs. These are characterised by exceptionally large AGN radio fluxes, indicating high CR yields and associated CR heating rates. We suggest a self-regulation cycle of AGN feedback in which non-RMH clusters are heated by streaming CRs homogeneously throughout the central cooling region. We predict {\em radio micro halos} surrounding the AGNs of these CR-heated clusters in which the primary emission may predominate the hadronically generated emission. Once the CR population has streamed sufficiently far and lost enough energy, the cooling rate increases, which explains the increased star formation rates in clusters hosting RMHs. Those could be powered hadronically by CRs that have previously heated the cluster core.