AGN feedback
A black hole at a galaxy's center is minuscule compared to the galaxy, so you might expect it to be a passive resident, irrelevant to its giant host. Yet when it switches on as an AGN, it dumps out so much energy that it can reshape the whole galaxy around it — heating gas, blowing winds, and starving the galaxy of the cold gas it needs to make new stars. This push-back from the central engine onto its host is AGN feedback.
Feedback works through two broad channels. In 'radiative' (or quasar) mode, the blinding light and fast winds from a vigorously feeding black hole sweep up and heat surrounding gas, and can drive it right out of the galaxy. In 'kinetic' (or radio/jet) mode, relativistic jets inflate huge bubbles in the hot gas of a galaxy or cluster, continually reheating it so it cannot cool and condense into stars. Either way the AGN injects energy that opposes the gravitational collapse of gas, throttling star formation. Crucially, it can also choke off the black hole's own fuel supply, regulating its growth.
AGN feedback is now considered essential to understanding galaxies. Without it, computer models predict that the biggest galaxies should keep gobbling cold gas and forming far more stars than we actually observe — feedback is the 'thermostat' that quenches them, leaving giant ellipticals 'red and dead'. It is the favored explanation for the M-sigma relation, tying black hole and galaxy growth together. The honest caveat: that feedback matters is widely accepted, but exactly how the energy couples to the gas, and how much each mode contributes, remain active areas of research.
At the heart of the Perseus galaxy cluster, jets from the central galaxy's black hole have punched giant bubbles into the hot X-ray gas, even sending out pressure waves recorded as the lowest 'note' ever detected — a B-flat some 57 octaves below middle C. That energy keeps the cluster gas too hot to form many new stars.
The central engine pushing back — heating gas and shutting down star formation.
AGN feedback can both quench and, in places, trigger star formation, and its exact workings are not fully pinned down. What is robust is that without some form of AGN feedback, simulations make far too many overgrown, star-forming giant galaxies compared to the real universe.