Heating & cooling cycles in cool cluster cores Prateek Sharma (IISc) Collaborators: Deovrat Prasad, Arif Babul The Astrophysical Journal, 811:108 (21pp), 2015 October 1 doi:10.1088/0004-637X/811/2/108 © 2015. The American Astronomical Society. All rights reserved. COOL CORE CYCLES: COLD GAS AND AGN JET FEEDBACK IN CLUSTER CORES The Astrophysical Journal, 811:108 (21pp), 2015 October 1 doi:10.1088/0004-637X/811/2/108 1 1 2 Deovrat Prasad , Prateek Sharma , and Arif Babul © 2015. The American Astronomical Society. All rights reserved. 1 Joint Astronomy Program and Department of Physics, Indian Institute of Science, Bangalore, 560012, India; [email protected], [email protected] 2 Department of Physics and Astronomy, University of Victoria, Victoria, BC V8P 1A1, Canada; [email protected] CORecOeivLed C201O5 RApEril 1C2;YacCcepLteEd 2S01:5CJuOly L28D; pubGlisAhedS20A15NSeDptemAbeGr 2N8 JET FEEDBACK IN CLUSTER CORES ABSTRACT 1 1 2 Deovrat Prasad , Prateek Sharma , and Arif Babul 1 Using high-resolution 3D and 2DJoi(natxiAsystmromnoemtriyc)PhroygdrraomdyannadmDicepsiamrtmuleantitoonfs Pinhysspichse,riIcnadliagneoInmsetittruyt,ewofe Sstcuiednyceth, eBangalore, 560012, India; evolution of cool cluster cores heated by feedback-drivendeboivproalta@r pahctyisviecsg.iailsacc.etircnentu.icnl,epi r(aAteGekN@s)pjheytss.icCs.oiinsdce.enrsnaetti.oinn 2 of cold gas, and the consequenDt eepnahrtamnceendt oafccPrheytisoicns, aisndreAqustirroendofmoyr,AUGnNivefreseitdyboafckVticotobraiala,nVciectroardiaia, tBivCe Vco8oPli1nAg 1, Canada; [email protected] fi fi with reasonable ef ciencies, and to match theRoecbesievrevded20c1o5olAcporirle1p2r;oapcecrteipetse.dA20fe1e5dJbualcyk2e8f; pcuiebnlicsyhe(dm2e0ch1a5nSiceapltember 28 luminosity » (cid:139)M˙ c2; where M˙ is the mass accretion rate at 1 kpc) as small as 6 × 10−5 is sufficient to reduce acc acc the cooling/accretion rate by ∼10 compared to a pure cooling flow in clusters (with M (cid:49) 7 ´ 1014 M ). This ABSTRA2C00T (cid:58) fi value is much smaller compared to the ones considered earlier, and is consistent with the jet ef ciency and the fact that only a smalUl fsriancgtiohnigohf -graessoatlu1tkiopnc i3sDaccarnedted2Dont(oatxhiesysumpmermetarsisci)vehbyldacrokdhyonleam(SiMcBsHim).uTlhaetiofenesdbianckspherical geometry, we study the fi ef ciency in eareliveor lwutoirokns wofascosoolhciglhusttheart cthoerecsluhsteeartecodrebyreafecheeddbaeqcuki-lidbrriivumeninbiapohloatrsatactteivweitghoaulatcmtiucchnuclei (AGNs) jets. Condensation fi precipitation, unlike what is observed in cool-core clusters. We nd hysteresis cycles in all our simulations with of cold gas, and the consequent enhanced accretion, is required for AGN feedback to balance radiative cooling cold mode feedback: condensation of cold gas when the ratio of the cooling-time to the free-fall time (t t ) is with reasonable efficiencies, and to match the observed cool core propercotoiles.ff A feedback efficiency (mechanical 10 leads to a sudden enhancement in the accretion rate; a large accretion rate causes strong jets and overheating  luminosity » (cid:139)M˙ c2; where M˙ is the mass accretion rate at 1 kpc) as small as 6 × 10−5 is sufficient to reduce of the hot intracluster medium suchactchat t t > 1a0c;c further condensation of cold gas is suppressed and the cool ff fl accretion rate fatlhlse, cleoaodliinnggt/oacslcorweticoonolirnagteobf yth∼e 1co0recoamndpcaorenddentosataiopnuoref ccoolodlignags, reoswtartiinngcltuhestecyrscle(w. ith M (cid:49) 7 ´ 1014 M ). This 200 (cid:58) fi Therefore, therevisalauespirseamd uinchcosrme parloleprerctioesm, psuacrhedastothethjeetopnoewserc,oancscirdeteiorendraetea,rlfioerr,thaensdamise cvoanluseisotfencotrwe ith the jet ef ciency and the fact fi entropy or t t . A smaller number of cycles is observed for higher ef ciencies and for lower mass halos because that only a small fraction of gas at 1 kpc is accreted onto the supermassive black hole (SMBH). The feedback cool ff the core is overheated to a longer cooling time. The 3D simulations show the formation of a few-kpc scale, fi ef ciency in earlier works was so high that the cluster core reached equilibrium in a hot state without much rotationally supported, massive (~1011 M ) cold gas torus. Since the torus gas is not accreted onto the SMBH, it is (cid:58) fi precipitation, unlike what is observed in cool-core clusters. We nd hysteresis cycles in all our simulations with 4 largely decoupled from the feedback cycle. The radially dominant cold gas (T < 5 × 10 K; v > v ) consists of ∣ ∣ ∣ ∣ r f cold mode feedback: condensation of cold gas when the ratio of the cooling-time to the free-fall time (t t ) is fast cold gas uplifted by AGN jets and freely infalling cold gas condensing out of the core. The radially dominant cool ff cold gas extends ou1t0tole25adkspctoforathseudfidduecniael nruhna(nhcaelommeansts 7in´th1e01a4cMcreatinodnfereadtbea;cak elaffircgieenaccyc6re×ti1o0n−5r)a,tweicthauses strong jets and overheating  (cid:58) fl fl the average masos finthoewhroatte idnotmraincalutinstgerthemoeudtiuomw rsatuecbhy tahafatcttor of t≈2>. W1e0c;omfupratrheeorurcosinmduelnatsioantiorensuoltsf cold gas is suppressed and the cool ff with recent observations. accretion rate falls, leading to slow cooling of the core and condensation of cold gas, restarting the cycle. Key words: galaxies: clusters: intracluster medium – galaxies: halos – galaxies: jets Therefore, there is a spread in core properties, such as the jet power, accretion rate, for the same value of core fi entropy or t t . A smaller number of cycles is observed for higher ef ciencies and for lower mass halos because cool ff the core is overheated to a longer cooling time. The 3D simulations show the formation of a few-kpc scale, 1. INTRODUCTION While there are potential heat sources, such as the kinetic rotationally supported, massive (~1011 M ) cold gas torus. Since the torus gas is not accreted onto the SMBH, it is energy of infalling galaxies and sub-halos (e.g., Dekel & (cid:58) The majority of baryons in galaxy clusters are in the form of 4 largely decoupled from the feedbackBciyrncbloei.mTh20e0r8a)d, ithaellrymadlocmonidnuacntitocnoflrdomgatshe(Tho<tter5o×uts1k0irtsK; v > v ) consists of ∣ ∣ ∣ ∣ r f a hot plasma known as the intracluster medium (ICM). In the fast cold gas uplifted by AGN jets an(ed.gf.,reVeloyigitn&falFlianbgianco2ld00g4a; sVcooitnd20e1n1s)i,nga ogluotbaolflythsetacbloere. The radially dominant absence of cooling and heating, the ICM is expected to follow cold gas extends out to 25 kpc for themfiedchuacniiasml r,uwnhi(chhailnocrmeaassess 7rap´idly10w1i4thMan ianncrdeafseinegdbhoatcgkaes fficiency 6 × 10−5), with fi self-similar pro les for density, temperature, etc., irrespective (cid:58) fl density in the cflore, is required to prevent catastrophic cooling. the average mass in ow rate dominating the out ow rate by a factor of ≈2. We compare our simulation results of the halo mass (Kaiser 1986, 1991; see also the review by Observations of several cool-core clusters by Chandra and Voit 2005). However, sewlf-istihmirlearcietyntisonbosteorbvsaetrivoendsi.n either XMM-Newton have uncovered active galactic nucleus (AGN)- groups or clusters (e.g., Balogh et al. 1999; Ponman et al. 1999; Key words: galaxies: clusters: intracjleut-sdtreirvemn Xed-riauymca–vitgieasl,awxhioesse: mheaclohasni–cagl aploawxeireiss:ejneotusgh to Babul et al. 2002). Moreover, the core cooling times in about a balance radiative cooling in the core (e.g., Böhringer third of clusters are shorter than 1 Gyr, much shorter than their et al. 2002; Bîrzan et al. 2004; McNamara & Nulsen 2007). age (∼Hubble time; e.g., Cavagnolo et al. 2009; Pratt The AGN jets are powered by the accretion of the cooling ICM 1. INTRODUCTION While there are potential heat sources, such as the kinetic et al. 2009). Thus, we expect cooling to shape the distribution onto the supermassive black hole (SMBH) at the center of the energy of infalling galaxies and sub-halos (e.g., Dekel & of baryons in these cool-core clusters. The majority of baryons in galaxy clusters daoremiinnantht eclufsotremr gaolfaxy. Thus, more cooling/accretion leads to The existence of cool cores with short cooling times in a Birnboim 2008), thermal conduction from the hotter outskirts an enhanced jet power and ICM heating, closing a feedback a hot plasma known as the intracluster medium (ICM). In the good fraction of galaxy clusters is a long-standing puzzle. (e.g., Voigt & Fabian 2004; Voit 2011), a globally stable loop that prevents runaway cooling in the core. absence of coolingfland heating, the ICM is expected to follow According to the classical cooling ow model, cluster cores AGN feedback has been lomnge-cshusapneicstmed,towphlaicyha irnolcerienasseelsf- rapidly with an increasing hot gas fi with such shsoerltf-sciomoliilnagr ptirmoesleswefroer dexepnescitteyd, tteompcoeorlature, etc., irrespective regulating the ICM (e.g., Binney & Tabor 1995; Ciotti & density in the core, is required to prevent catastrophic cooling. catastrophically and to fuel star formation at a rate of of the halo mass (Kaiser 1986, 1991; see also the review by Ostriker 2001; Soker et al. 2001; Babul et al. 2002; McCarthy 100–1000 M yr−1 (e.g., Fabian 1994; Lewis et al. 2000). Observations of several cool-core clusters by Chandra and (cid:58) Voit 2005). However, self-similarity is not oebt saelr.v2e0d08i)n, beuitthaerclear picture has emerged only recently. However, cooling, dropout, and star formation at these high XMM-Newton have uncovered active galactic nucleus (AGN)- While AGN feedback should provide feedback heating in groups or clusters (e.g., Balogh et al. 1999; Ponman et al. 1999; rates are never seen in cluster cores (e.g., Edge 2001; Peterson jet-driven X-ray cavities, whose mechanical power is enough to cluster cores (as it is enhanced with ICM cooling), it is not Babul et al. 2002). Moreover, the core cooling times in about a et al. 2003; O’Dea et al. 2008). This means that some source(s) obvious if, for reasonable pabraalmaentceres, AraGdNiahtievaetingccoaonliknegep in the core (e.g., Böhringer of heating is(athrei)rdabolef ctolursetpelresniasrhetshheocroterer tchoaolnin1g Gloyssre,sm, uch shorter than their pace with cooling that increases rapidly with an increasing core et al. 2002; Bîrzan et al. 2004; McNamara & Nulsen 2007). thereby prevenatigneg ru(n∼awHauybcboloelintgimaned; staer.gfo.,rmCatiaovna. gnolo et al. 2009; Pratt density. Moreover, the dense core gas is expected to be highly The AGN jets are powered by the accretion of the cooling ICM et al. 2009). Thus, we expect cooling to shape the distribution onto the supermassive black hole (SMBH) at the center of the 1 of baryons in these cool-core clusters. dominant cluster galaxy. Thus, more cooling/accretion leads to The existence of cool cores with short cooling times in a an enhanced jet power and ICM heating, closing a feedback good fraction of galaxy clusters is a long-standing puzzle. loop that prevents runaway cooling in the core. fl According to the classical cooling ow model, cluster cores AGN feedback has been long-suspected to play a role in self- with such short cooling times were expected to cool regulating the ICM (e.g., Binney & Tabor 1995; Ciotti & catastrophically and to fuel star formation at a rate of Ostriker 2001; Soker et al. 2001; Babul et al. 2002; McCarthy −1 100–1000 M yr (e.g., Fabian 1994; Lewis et al. 2000). (cid:58) et al. 2008), but a clear picture has emerged only recently. However, cooling, dropout, and star formation at these high While AGN feedback should provide feedback heating in rates are never seen in cluster cores (e.g., Edge 2001; Peterson cluster cores (as it is enhanced with ICM cooling), it is not et al. 2003; O’Dea et al. 2008). This means that some source(s) obvious if, for reasonable parameters, AGN heating can keep of heating is(are) able to replenish the core cooling losses, pace with cooling that increases rapidly with an increasing core thereby preventing runaway cooling and star formation. density. Moreover, the dense core gas is expected to be highly 1 Cold gas condensation allows feedback to act sufficiently fast, unlike Bondi • t /t ~10 seems robust • cool ff cooling & heating cycles • push ε to smallest allowed by observations • cold gas inflows & outflows • angular momentum: stochastic cold accretion • AGN jet-ICM sims. @⇢ + ⇢v = S mass ⇢ @t r · @v ⇢ + v v = p ⇢ � + S v ˆr momentum ⇢ jet @t · r �r � r ✓ ◆ p d � 2 ln(p/⇢ ) = n ⇤ � 1 dt � � source terms to mimic injection by feedback AGN jets The Astrophysical Journal, 811:108 (21pp), 2015 October 1 Prasad, Sharma, & Babul AGN jet-ICM sims. 15 1.8 ´ 10 M , respectively, and adopt c = 4.7 for all in Figure 1). The jet radius r is scaled with the halo mass; i.e., (cid:58) 200 jet @⇢ models. + ⇢v = S mass1 3 ⎛ ⇢ ⎞ M @t r · We include the source terms S for mass and S v r for the ˆ 200 ρ r jet r = 2 kpc . ⎜ ⎟ jet @v radial momentum to drive AGN jets (v is the velocity which 14 7 ´ 10 M ⎝ ⎠ jet ⇢ + v v = p ⇢ � + S v ˆr (cid:58) momentum ⇢ jet 3 the jet matter is put in). These source terms and the cooling @t · r �r � r ✓ ◆ term (in Equation (3)) are applied in an operator-split fashion. p d � 2 ln(p/⇢ ) = n ⇤ The jet mass-loading rate is calculated from the current mass The mass and momentum source terms are approximated � 1 dt � accretion rate�(M˙ ) evaluated at the inner radial boundary such forward in time and centered in space. The cooling term is acc source term applied in a small fi that the increase in the jet kinetic energy is a xed fraction of applied using a semi-implicit method described in Equations bipolar cone at the center: the energy released via accretion; i.e., (7) of McCourt et al. (2012). opening angle of 300, size 2 kpc 2 kpc Our simulations do not include physical processes like star 2 2 M˙ v = (cid:139)M˙ c . 6 ( ) jet acc formation and supernova feedback. Star formation may deplete jet some of the cold gas available in the cores (see Li et al. 2015), v =0.1c, ε=6x10-5, r =1, 2004 kpc −1 We choose the jejett velocity v =in,o3ut ´ 10 km s (0.1 c; c is jet but this is unlikely to change our results for a realistic model of robust to variations the speed of light); such fast velocities are seen in X-ray star formation. Supernova feedback is energetically subdomi- fl observations of small-scale out ows in radio galaxies (Tombesi nant compared to AGN feedback, and cannot realistically fi fi et al. 2010). The jet ef ciency (ò; our ducial value is fl suppress cluster cooling ows (e.g., Saro et al. 2006). We only −5 6 × 10 ) accounts for both the fraction of the infalling mass at include the most relevant physical processes, namely cooling the inner boundary (at 1 kpc for the cluster runs) that is accreted and AGN jet feedback, in our present simulations. by the SMBH and for the fraction of accretion energy that is channeled into the jet kinetic energy. Our results are insensitive 2.1. Jet Implementation to a reasonable variation in jet parameters (v , r , q , s , s ), jet jet jet r q fi Jets are implemented in the active domain by adding mass but depend on the jet ef ciency (ò). Like Gaspari et al. (2012), the jet energy is injected only in and momentum source terms as shown in Equations (1) and the form of kinetic energy; we do not add a thermal energy (2). The source terms are negligible outside a small biconical region centered at the origin around q = 0, p, mimicking mass source term corresponding to the jet. We note that Li & Bryan (2014b) have shown that the core evolution does not depend and momentum injection by fast bipolar AGN jets. sensitively on the manner in which the feedback energy is The density source term is implemented as partitioned into kinetic or thermal form. Another difference from previous approaches, which use few grid points to inject S r, q = (cid:38)M˙ y r, q , ( ) ( ) r jet jet mass/energy, is that our jet injection region is well- resolved. where M˙ is the single-jet mass loading rate, jet 2.2. Grid, Initial, and Boundary Conditions ⎡ ⎛ q - q ⎞ ⎛ q + q - p ⎞ ⎤ jet jet y r, q = 2 + tanh + tanh ( ) ⎢ ⎜ ⎟ ⎜ ⎟ ⎥ Most AGN feedback simulations evolved for cosmological s s ⎝ ⎠ ⎝ ⎠ ⎣ ⎦ q q timescales (e.g., Gaspari et al. 2012; Li & Bryan 2014a) use fi ⎡ ⎛ r - r ⎞ ⎤ 1 Cartesian grids with mesh re nement. However, we use jet ´ 1 + tanh ´ ⎢ ⎜ ⎟ ⎥ spherical coordinates with a logarithmically spaced grid in s 4 ⎝ ⎠ ⎣ ⎦ r 5 radius, and equal spacing in θ and f. The advantage of a ( ) fi spherical coordinate system is that it gives ne resolution at smaller scales without a complex algorithm. Perhaps more that describes the spatial distribution of the source term which importantly, a spherical setup allows for 2D axisymmetric falls smoothly to zero outside the small biconical jet region of simulations which are much faster and capture a lot (but not all) radius r and half-opening angle q . We smooth the jet source jet jet of essential physics. terms in space because the Kelvin–Helmholtz instability is We perform our simulations in spherical coordinates with known to be suppressed due to numerical diffusion in a fast 0 (cid:45) q (cid:45) p, 0 (cid:45) f (cid:45) 2p, and r (cid:45) r (cid:45) r , with fl min max ow if the shear layer is unresolved (e.g., Robertson 1 3 et al. 2010). The normalization factor ⎛ ⎞ M 200 r = 1, 200 kpc . [ ] ⎜ ⎟ min,max [ ] 14 7 ´ 10 M ⎝ ⎠ (cid:58) 3 (cid:38) = 3 2pr ( 1 - cos q ) According to self similar scaling, we have scaled all length jet jet scales in our simulations (inner/outer radii r r , r , jet min max 200 1 3 ensures that the total mass added due to jets per unit time is radius r ) as M . jet 200 fl 2M˙ . All our simulations use the following jet parameters: We apply out ow boundary conditions (gas is allowed to jet leave the computational domain but prevented from entering it) s = 0.05 kpc, q = p 6, and s = 0.05. The jet source region r jet q fi at the inner radial boundary. We x the density and pressure at with an opening angle of 30° may sound large but we get the outer radial boundary to the initial value and prevent gas similar results with narrower jets. Also, the fast jet extends well fl from leaving or entering through the outer boundary. Re ective beyond the source region and is much narrower (c.f. third panel boundary conditions are applied in θ (with the sign of v f fl ipped) and periodic boundary conditions are used in f. We 3 We have also carried out narrow-jet simulations with momentum injection fi noticed that cold gas has a tendency to arti cially “stick” at the fi in the vertical z direction, but do not nd much difference from our runs with [ˆ] θ boundaries (mainly in 2D axisymmetric simulations) for our momentum injection in the radial [r ; see Equation (2)] direction. ˆ 4 Dependence on halo mass & efficiency larger ε 1.8x1015 M sun suppresses 7x1014 M sun accretion more massive halos require larger ε depends on where Mdot calculated Density movie 150 kpc BCG+NFW in PLUTO 256x128x32 in (logr,θ,φ) r =0.5 kpc, r =0.5 Mpc min max evolution for ~2.8 Gyr made by Deovrat Prasad r-θ slices The Astrophysical Journal, 811:108 (21pp), 2015 October 1 Prasad, Sharma, & Babul sound/weak shock waves pressure 50 kpc buoyant bubbles bubble mixing with ICM density Figure 1. Pressure (upper panel), electron number density (middle panel), and temperature (lower panel) contour plots (R–z plane at f = 0) in the core at different times for the 3D fiducial run. The density is cutoff at the maximum and the minimum contour level shown. The low-density bubbles/cavities are not symmetric and there are signatures of mixing in the core. The left panel corresponds to a time just before a cooling time in the core. The second panel from the left shows cold gas dredged up by the outgoing jets. The rightmost panel shows infalling extended cold clouds. The pressure maps show the weak outer shock, but the bubbles/cavities so prominent in the density/temperature plot are indiscernible in the pressure map, implying that the bubbles are in pressure equilibrium and buoyant. Also notice the outward-propagating sound waves in the two middle pressure panels in which the jet is active. The infalling/rotationally supported cold gas has a much lower temperature and pressure than the hot phase. The arrows in the temperature plots denote the projected gas velocity unit vectors. 5 r-θ slices The Astrophysical Journal, 811:108 (21pp), 2015 October 1 Prasad, Sharma, & Babul sound/weak shock waves pressure 50 kpc buoyant bubbles bubble mixing with ICM density Figure 1. Pressure (upper panel), electron number density (middle panel), and temperature (lower panel) contour plots (R–z plane at f = 0) in the core at different times for the 3D fiducial run. The density is cutoff at the maximum and the minimum contour level shown. The low-density bubbles/cavities are not symmetric and there are signatures of mixing in the core. The left panel corresponds to a time just before a cooling time in the core. The second panel from the left shows cold gas dredged up by the outgoing jets. The rightmost panel shows infalling extended cold clouds. The pressure maps show the weak outer shock, but the bubbles/cavities so prominent in the density/temperature plot are indiscernible in the pressure map, implying that the bubbles are in pressure equilibrium and buoyant. Also notice the outward-propagating sound waves in the two middle pressure panels in which the jet is active. The infalling/rotationally supported cold gas has a much lower temperature and pressure than the hot phase. The arrows in the temperature plots denote the projected gas velocity unit vectors. 5 Cold rotating torus 6 Deovrat Prasad & Prateek Sharma few kpc scale molecular torus Fig. 3.— The 2-D (z = 0) contour plots of density in the very inner region at different times for the fiducial 3-D run, with the projection of the velocity unit-vector represented by arrows. The top-left panel shows the beginning of the infall of cold gas with random angular momentum. The top-middle panel shows an anti-clockwise transient torus. All times after this show a clockwise torus which waxes and wanes because of cooling and AGN heating cycles. Even at late times the cold torus is not stable and gets disrupted by jets. it reforms over a few cooling times. Figure 3 shows the that these explosive ‘events’ are rare and the jet mate- evolution of the torus at various stages of the simulation. rial is quickly mixed with the ICM, and the core settles The top-left panel of Figure 3 shows the cluster center at back to a quiescent state (see the top panel of Fig. 9 0.5 Gyr. Small cold gas clouds are accumulating in the which shows a large peak in jet energy at 2.4 Gyr). In core after the first active AGN phase. At 1.3 Gyr, cold reality, the cold gas in the torus is mainly consumed by gas accreting through the inner boundary has an anti- star-formation and only a part of it reaches the SMBH, clockwise rotational sense. At 1.98 Gyr, cold gas (and and that too at the slow viscous timescale. In addition, the hot gas out of which it condenses) is rotating clock- the rapidly reorienting AGN jets can disrupt the massive wise. Jet activity leading up to this phase has reversed cold torus (e.g., see Babul et al. 2012). the azimuthal velocity of the cold gas. At all times af- Li & Bryan (2014b) show that after 3 Gyr the cold ter this the dynamic cold gas torus rotates in a clockwise gas settles down in form of a stable torus, with no further sense, essentially because the mass (and angular momen- condensation of extended cold gas. This is inconsistent tum) in the rotating torus is much larger than the newly with observations. The bottom panels in Figure 3 from condensing cold gas. our fiducial run shows that the torus is unsteady even at The middle panels of Figure 3 show the dynamic na- late times. Moreover, unlike them, we see extended cold ture of the rotationally supported torus. The torus gets gas condensing out till the end. We compare our results disrupted due to jet activity as seen in the middle panel in detail with Li & Bryan (2014b) in section 4.1. of Figure3, but forms again quickly. The snapshots at 2.4 To test the role of cooling in maintaining the cold ro- and 2.4 Gyr show that the inner region is covered by the tating torus, we restarted the 3-D fiducial run after a very hot/dilute jet material. This unphysical behavior massive cold torus had formed (3 Gyr), and re-ran it is mainly because of our feedback prescription; we scale without radiative cooling or feedback heating. While the jet power with the instantaneous mass inflow rate the cold torus is long-lived even without cooling (Kelvin- through the inner boundary (see Eq. 6). Even small os- Helmholtz instability does not grow for at least the next cillations of the cold torus can sometimes lead to a large Gyr), it is heated (by numerical dissipation) to 105 K, instantaneous mass inflow through the inner boundary and is no longer maintained at the temperature of the and hence an explosive jet feedback in which the jet ma- stable phase (104 K). Thus, radiative cooling function is terial encompasses the inner core. The reassuring fact is what dictates the temperature of the cold phase. The 6 Hamer et al. Cold torus in Hydra A 6 Hamer et al. [Hamer et al. 2014] ~5 kpc cold torus more examples from ALMA, Hershel Figure 2. This figure shows the IFU maps of the Hα emission as taken from fits to the Hα/[NII] triplet observed in the VIMOS cubes. Panel (A) shows a continuum image made by collapsing the cube, the contours show the Hα emission clearly centred on the BCG. Panel (B) is a Hα Flux map which shows a disc of bright emission running across the BCG. In panel (C) we show the relative velocity of the Hα line to the galaxy redshift, a strong velocity gradient of ∼ 600 km s−1 can clearly be seen. Contoured on this plot are lines of constant velocity created by fitting a disc model to the velocity map. The final panel (D) shows the measured Full Width Half Maximum (FWHM) of the line which can be seen to broaden at the centre of the velocity gradient. is likely to be due to the lower signal–to-noise as the lines are BCG. The luminosity of [FeII] emission has a high depen- present within the total spectrum of this region (extracted dence on the gas density (Bautista et al. 1994) so we would 1×1 arcsec2 centred on the offset Paα peak) though the line expect it to be brightest in the central regions where the is weak compared to Paα. gas density is higher. Despite being compact the line does appear to be extended to the east on scales slightly greater [FeII] emission was the only line detected in the H-band than the seeing. Within this small extent there appears to be observations. The maps presented in Figure 5 show that the a velocity change of ∼ 200 km s−1 across the emission. This [FeII] emission is compact and located at the centre of the Figure 2. This figure shows the IFU maps of the Hα emission as taken from fits to the Hα/[NII] triplet observed in the VIMOS cubes. Panel (A) shows a continuum image made by collapsing the cube, the contours show the Hα emission clearly centred on the BCG. Panel (B) is a Hα Flux map which shows a disc of bright emission running across the BCG. In panel (C) we show the relative velocity of the Hα line to the galaxy redshift, a strong velocity gradient of ∼ 600 km s−1 can clearly be seen. Contoured on this plot are lines of constant velocity created by fitting a disc model to the velocity map. The final panel (D) shows the measured Full Width Half Maximum (FWHM) of the line which can be seen to broaden at the centre of the velocity gradient. is likely to be due to the lower signal–to-noise as the lines are BCG. The luminosity of [FeII] emission has a high depen- present within the total spectrum of this region (extracted dence on the gas density (Bautista et al. 1994) so we would 1×1 arcsec2 centred on the offset Paα peak) though the line expect it to be brightest in the central regions where the is weak compared to Paα. gas density is higher. Despite being compact the line does appear to be extended to the east on scales slightly greater [FeII] emission was the only line detected in the H-band than the seeing. Within this small extent there appears to be observations. The maps presented in Figure 5 show that the a velocity change of ∼ 200 km s−1 across the emission. This [FeII] emission is compact and located at the centre of the