AGN feedback is an essential mechanism in cosmological simulations in order to reproduce the quenching of star formation. Observationally, the actual interplay of BH accretion, star formation, and AGN outflows is far less well understood. In large AGN samples, a correlation between star formation rate (SFR) and AGN luminosity (BH accretion rate) is often observed (e.g. Rosario+12, A\&A, 545, A45). This has increased the confusion about the efficiency of the radiative AGN feedback scenario, its timescale and how the energy couples to the host galaxy. New simulations of galaxy-mergers (Volonteri+15, MNRAS in press) show that accretion rate and SFR (averaged over 100\,Myr) are correlated at nuclear scales while it is uncorrelated on galaxy-wide scales. This could linked to the different time-scales of SF and AGN activity (Hayward+14, MNRAS, 445, 598; Hickox+14, ApJ, 782, 9).

With CARS we will investigate the relation between BH accretion and the SFR of the host galaxies in more detail than possible before. Due to the large FoV of MUSE, we map regions of star formation over the entire galaxy. Thus, we can correlate the SFR on various scale from <500pc to the entire galaxy with the instantaneous BH accretion rate. Since our AGN sample contains host galaxies of various morphological types from discs to major mergers, we can compare those results with the predications from the numerical simulations for the different relations seen fo those types. Furthermore, we can go a step further and connect the BH accretion to the total gas content (and also spatially resolved with ALMA data), because it is the gas that drive BH accretion and star formation.