1. Tools for computing the AGN feedback: radio-loudness and the kinetic luminosity function Gabriele Melini Fabio La Franca Fabrizio Fiore Active Galactic Nuclei 9: Black Holes and Revelations Dipartimento di Fisica - Universit à degli Studi “ Roma Tre ” INAF - Osservatorio Astronomico di Roma Ferrara, 24-27/5/2010

2. Context Feedback from AGN is one of the main ingredients in AGN/galaxy co-evolutionary scenarios; heating is requested in order to inhibit further star formation at low redshift. Few direct observations; simulations depend on the triggering mechanism of the AGN and the feedback description. Quasar (radiative) mode: feedback is associated to the accretion episodes (during mergers); efficiency depends on the AGN fraction. Radio (kinetic) mode: feedback is related to a low (10 -5 M  /yr) continuous accretion of gas cooling from the halo. Little contribution to bolometric emission, related to the total accreted mass. SAMs (e.g. Croton et al. 2006; Bower et al. 2008) do not implement any direct relation between radio (feedback) and X-ray activity (accretion). A strong correlation between LX and LR is actually observed.

3. Work scheme This has been done by: measuring the probability distribution P(R|Lx,z) of the radio loudness ratio (defined as R=log L(1.4 GHz)/L(2-10 keV)), as well as its dependences from intrinsic X-ray luminosity and redshift; convolving this probability distribution with the AGN X-ray LF (La Franca+05 & Brusa+09) -> radio LF convolving the radio LF with a relation between kinetic and radio luminosity -> kinetic LF integrating this kinetic LF in luminosity -> kinetic power density We aimed to estimate the AGN kinetic power linking the AGN radio emission to the accretion rate related to the AGN activity, at variance with previous works -> useful for a detailed implementation in galaxy formation and evolution models. (see also kinetic estimates from Best+06; Merloni&Heinz07; Shankar+08; Kording+08; Cattaneo&Best09; Smolcic+09)

4. The sample data from various X-ray selected samples (also observed at 1.4 GHz) from literature; ~1600 AGN (Lx >10 42 erg/s), with z and NH measurements available; cross-correlation with radio data following a maximum likelihood technique (e.g. Sutherland & Saunders 1992; Ciliegi et al. 2003); correlation radii small enough to select (as much as possible) radio emission coeval to the X-ray one -> no extended FRII radio lobes; 375 objects radio detected radio and X-ray flux limits of the surveys

5. Probability distribution P(R) The shape of the P(R) distribution has been obtained by comparing the observed and expected (once a model is assumed) numbers of objects in each bin of the (Lx,z,R) space, using  2 estimators, thus taking into account selection effects. The average R value increases with decreasing X-ray luminosity and increasing redshift.

6. Checks: LR-LX relation dep L,z dep L only The LR-LX relation from the fundamental plane from Merloni et al. (2003) is well reproduced, assuming a fixed BH mass of 10 8 M .

7. Checks: 1.4 GHz LF and counts 1.4 GHz radio counts (with FRII contribution from Wilman et al. 2008) 1.4 GHz luminosity function our estimates of the radio LF and counts also are in good agreement with the observations

8. The kinetic luminosity function To derive the kinetic LF we have convolved our previously derived 1.4 GHz LF with a conversion relation between radio and kinetic power: LR < 10 25 W/Hz: Best et al. (2006) LR>10 25 W/Hz: Willott et al. (1999)

9. The kinetic luminosity function Kinetic LF Bolometric (radiative) LF (using the Marconi et al. 2004 bolometric correction) At the break luminosity (10 39 W) the KLF is two orders of magnitude smaller than the bol LF. Becomes comparable at LK~10 36 W, where most of the kinetic power density is produced.

10. Evolution of the kinetic power density kinetic power density good agreement with Croton et al. (2006) requirements at z>0.5, but sharp decrease at low redshift RQ AGN contributes by ~50% to the total kinetic energy  k /  rad increases at low z radiative power density adapted from Croton et al. (2006), assuming  = 0.1

11. Summary The knowledge of P(R) is necessary to estimate the AGN radio feedback, by allowing to couple it with the luminous (accreting) phases of the AGN -> self-consistent inclusion in SAMs is possible. The average value of R increases with decreasing luminosity and increasing redshift; absorbed and un-absorbed AGN have the same radio properties. The AGN kinetic luminosity density decreases by a factor ~5 below z<0.5; the ratio between kinetic and radiative densities increases at low redshifts. The exclusion of the radio quiet population would lead to an underestimation of the kinetic power of a factor ~2. details in La Franca, Melini, Fiore, 2010, ApJ, in press (next week on astro-ph!)