Abstract
Radio galaxies are intensively discussed as the sources of cosmic rays observed above about 3 × 1018 eV, called ultra-high energy cosmic rays (UHECRs). We present a first, systematic approach that takes the individual characteristics of these sources into account, as well as the impact of the extragalactic magnetic-field structures up to a distance of 120 Mpc. We use a mixed simulation setup, based on 3D simulations of UHECRs ejected by observed, individual radio galaxies taken out to a distance of 120 Mpc, and on 1D simulations over a continuous source distribution contributing from beyond 120 Mpc. Additionally, we include the ultra-luminous radio galaxy Cygnus A at a distance of about 250 Mpc, as its contribution is so strong that it must be considered as an individual point source. The implementation of the UHECR ejection in our simulation setup, both that of individual radio galaxies and the continuous source function, is based on a detailed consideration of the physics of radio jets and standard first-order Fermi acceleration. This allows to derive the spectrum of ejected UHECR as a function of radio luminosity, and at the same time provides an absolute normalization of the problem involving only a small set of parameters adjustable within narrow constraints. We show that the average contribution of radio galaxies taken over a very large volume cannot explain the observed features of UHECRs measured at Earth. However, we obtain excellent agreement with the spectrum, composition, and arrival-direction distribution of UHECRs measured by the Pierre Auger Observatory, if we assume that most UHECRs observed arise from only two sources: the ultra-luminous radio galaxy Cygnus A, providing a mostly light composition of nuclear species dominating up to about 6 × 1019 eV, and the nearest radio galaxy Centaurus A, providing a heavy composition dominating above 6 × 1019 eV . Here we have to assume that extragalactic magnetic fields out to 250 Mpc, which we did not include in the simulation, are able to isotropize the UHECR events {at about 8 EeV} arriving from Cygnus A. Even in this case, significant anisotropy correlated with Cygnus A and Centaurus A could be present at higher energies, and thus allow for differences in UHECR spectrum and composition between the northern and southern hemispheres. If this scenario can be confirmed, it would also imply that the UHECR flux in our local cosmic environment is significantly above the average throughout the universe.