Search

Architectures such as Linformer and Mamba have recently emerged as competitive linear time replacements for transformers. However, corresponding large pretrained models are often unavailable, especially in non-text domains. To remedy this, we present a Cross-Architecture Layerwise Distillation (CALD) approach that jointly converts a transformer model to a linear time substitute and fine-tunes it to a target task. We also compare several means to guide the fine-tuning to optimally retain the desired inference capability from the original model. The methods differ in their use of the target model and the trajectory of the parameters. In a series of empirical studies on language processing, language modeling, and speech processing, we show that CALD can effectively recover the result of the original model, and that the guiding strategy contributes to the result. Some reasons for the variation are suggested., Comment: 15 pages, 4 figures

We present Gemini near-infrared integral field spectrograph (NIFS) K-band observations of the central 400 pc of NGC 1266, a nearby (D$\approx$30 Mpc) post-starburst galaxy with a powerful multi-phase outflow and a shocked ISM. We detect 7 H$_2$ ro-vibrational emission lines excited thermally to $T$$\sim$2000 K, and weak Br$\gamma$ emission, consistent with a fast C-shock. With these bright H$_2$ lines, we observe the spatial structure of the shock with an unambiguous tracer for the first time. The Br$\gamma$ emission is concentrated in the central $\lesssim$100 pc, indicating that any remaining star-formation in NGC 1266 is in the nucleus while the surrounding cold molecular gas has little on-going star-formation. Though it is unclear what fraction of this Br$\gamma$ emission is from star-formation or the AGN, assuming it is entirely due to star-formation we measure an instantaneous star-formation rate of 0.7 M$_\odot$ yr$^{-1}$, though the star-formation rate may be significantly higher in the presence of additional extinction. NGC 1266 provides a unique laboratory to study the complex interactions between AGN, outflows, shocks, and star-formation, all of which are necessary to unravel the evolution of the post-starburst phase., Comment: ApJ accepted

Motivated by the sensitivity-based importance score of the adaptive low-rank adaptation (AdaLoRA), we utilize more theoretically supported metrics, including the signal-to-noise ratio (SNR), along with the Improved Variational Online Newton (IVON) optimizer, for adaptive parameter budget allocation. The resulting Bayesian counterpart not only has matched or surpassed the performance of using the sensitivity-based importance metric but is also a faster alternative to AdaLoRA with Adam. Our theoretical analysis reveals a significant connection between the two metrics, providing a Bayesian perspective on the efficacy of sensitivity as an importance score. Furthermore, our findings suggest that the magnitude, rather than the variance, is the primary indicator of the importance of parameters.

Whilst state of the art automatic speech recognition (ASR) can perform well, it still degrades when exposed to acoustic environments that differ from those used when training the model. Unfamiliar environments for a given model may well be known a-priori, but yield comparatively small amounts of adaptation data. In this experimental study, we investigate to what extent recent formalisations of modularity can aid adaptation of ASR to new acoustic environments. Using a conformer based model and fixed routing, we confirm that environment awareness can indeed lead to improved performance in known environments. However, at least on the (CHIME) datasets in the study, it is difficult for a classifier module to distinguish different noisy environments, a simpler distinction between noisy and clean speech being the optimal configuration. The results have clear implications for deploying large models in particular environments with or without a-priori knowledge of the environmental noise., Comment: 5 pages, 3 figures

The role of radio mode AGN feedback on galaxy evolution is still under debate. In this study, we utilize a combination of radio continuum observations and optical integral field spectroscopic (IFS) data to explore the impact of radio AGN on the evolution of their host galaxies at both global and sub-galactic scales. We construct a comprehensive radio-IFS sample comprising 5578 galaxies with redshift z < 0.15 by cross-matching the LOFAR Two-Metre Sky Survey (LoTSS) with the MaNGA survey. We revisit the tight linear radio continuum - star formation relation and quantify its intrinsic scatter, then use the relation to classify 616 radio-excess AGN with excessive radio luminosities over that expected from their star formation rate. Massive radio AGN host galaxies are predominantly quiescent systems, but the quenching level shows no correlation with the jet luminosity. The mass assembly histories derived from the stellar population synthesis model fitting agree with the cosmological simulations incorporating radio-mode AGN feedback models. We observe that radio AGN hosts grow faster than a control sample of galaxies matched in stellar mass, and the quenching age (~ 5 Gyr) is at larger lookback times than the typical radio jet age (< 1 Gyr). By stacking the spectra in different radial bins and comparing results for radio AGN hosts and their controls, we find emission line excess features in the nuclear region of radio AGN hosts. This indicates that radio AGN are ionizing the surrounding interstellar medium in the vicinity of the nucleus. We also find that ongoing star formation in the outer regions of the galaxy is weaker if a radio jet is detected. Our findings support the scenario that the observed present-day radio AGN activity is not responsible for the past quenching of their hosts, but may help the host galaxies maintain quiescence through ionizing and heating the surrounding gas., Comment: 16 pages, 15 figures, submitted to A&A

Self-interested behavior from individuals can collectively lead to poor societal outcomes. These outcomes can seemingly be improved through the actions of altruistic agents, which benefit other agents in the system. However, it is known in specific contexts that altruistic agents can actually induce worse outcomes compared to a fully selfish population -- a phenomenon we term altruistic perversity. This paper provides a holistic investigation into the necessary conditions that give rise to altruistic perversity. In particular, we study the class of two-strategy population games where one sub-population is altruistic and the other is selfish. We find that a population game can admit altruistic perversity only if the associated social welfare function is convex and the altruistic population is sufficiently large. Our results are a first step in establishing a connection between properties of nominal agent interactions and the potential impacts from altruistic behaviors.

Spectral energy distribution (SED) fitting has been extensively used to determine the nature of the faint radio source population. Recent efforts have combined fits from multiple SED-fitting codes to account for the host galaxy and any active nucleus that may be present. We show that it is possible to produce similar-quality classifications using a single energy-balance SED fitting code, Prospector, to model up to 26 bands of UV$-$far-infrared aperture-matched photometry for $\sim$31,000 sources in the ELAIS-N1 field from the LOFAR Two-Metre Sky Survey (LoTSS) Deep fields first data release. One of a new generation of SED-fitting codes, Prospector accounts for potential contributions from radiative active galactic nuclei (AGN) when estimating galaxy properties, including star formation rates (SFRs) derived using non-parametric star formation histories. Combining this information with radio luminosities, we classify 92 per cent of the radio sources as a star-forming galaxy, high/low-excitation radio galaxy, or radio-quiet AGN and study the population demographics as a function of 150 MHz flux density, luminosity, SFR, stellar mass, redshift and apparent $r$-band magnitude. Finally, we use Prospector SED fits to investigate the SFR$-$150 MHz luminosity relation for a sample of $\sim$$133,000~3.6~\mu$m-selected $z<1$ sources, finding that the stellar mass dependence is significantly weaker than previously reported, and may disappear altogether at $\log_{10} (\mathrm{SFR}/M_\odot~\mathrm{yr}^{-1}) > 0.5$. This approach makes it significantly easier to classify radio sources from LoTSS and elsewhere, and may have important implications for future studies of star-forming galaxies at radio wavelengths., Comment: 20 pages, 13 figures, accepted for publication in MNRAS

Understanding cognitive processes in the brain demands sophisticated models capable of replicating neural dynamics at large scales. We present a physiologically inspired speech recognition architecture, compatible and scalable with deep learning frameworks, and demonstrate that end-to-end gradient descent training leads to the emergence of neural oscillations in the central spiking neural network. Significant cross-frequency couplings, indicative of these oscillations, are measured within and across network layers during speech processing, whereas no such interactions are observed when handling background noise inputs. Furthermore, our findings highlight the crucial inhibitory role of feedback mechanisms, such as spike frequency adaptation and recurrent connections, in regulating and synchronising neural activity to improve recognition performance. Overall, on top of developing our understanding of synchronisation phenomena notably observed in the human auditory pathway, our architecture exhibits dynamic and efficient information processing, with relevance to neuromorphic technology.

Self-interested routing polices from individual users in a system can collectively lead to poor aggregate congestion in routing networks. The introduction of altruistic agents, whose goal is to benefit other agents in the system, can seemingly improve aggregate congestion. However, it is known in that in some network routing problems, altruistic agents can actually worsen congestion compared to that which would arise in the presence of a homogeneously selfish population. This paper provides a thorough investigation into the necessary conditions for altruists to be guaranteed to improve total congestion. In particular, we study the class of series-parallel non-atomic congestion games, where one sub-population is altruistic and the other is selfish. We find that a game is guaranteed to have improved congestion in the presence of altruistic agents (even if only a small part of the total population) compared to the homogeneously selfish version of the game, provided the network is symmetric, where all agents are given access to all paths in the network, and the series-parallel network for the game does not have sub-networks which emulate Braess's paradox -- a phenomenon we refer to as a Braess-resistant network. Our results appear to be the most complete characterization of when behavior that is designed to improve total congestion (which we refer to as altruism) is actually guaranteed to do so., Comment: 7 pages, 2 figures

Morphological features in galaxies, like spiral arms, bars, rings, tidal tails etc. carry information about their structure, origin and evolution. It is therefore important to catalog and study such features and to correlate them with other basic galaxy properties, the environment in which the galaxies are located and their interactions with other galaxies. The volume of present and future data on galaxies is so large that traditional methods, which involve expert astronomers identifying morphological features through visual inspection, are no longer sufficient. It is therefore necessary to use AI based techniques like machine learning and deep learning for finding morphological structures quickly and efficiently. We report in this study the application of deep learning for finding ring like structures in galaxy images from the Sloan Digital Sky Survey (SDSS) data release DR18. We use a catalog by Buta (2017) of ringed galaxies from the SDSS to train the network, reaching good accuracy and recall, and generate a catalog of 29420 galaxies of which 4855 have ring like structures with prediction confidence exceeding 90 percent. Using a catalog of barred galaxy images identified by Abraham et. al. (2018) using deep learning techniques, we identify a set of 2087 galaxies with bars as well as rings. The catalog should be very useful in understanding the origin of these important morphological structures. As an example of the usefulness of the catalog, we explore the environments and star formation characteristics of ring galaxies in our sample., Comment: 22 pages, 15 figures, 1 Table. Accepted for publication in The Astrophysical Journal (ApJ)

In this study, we introduce a toll lane framework that optimizes the mixed flow of autonomous and high-occupancy vehicles on freeways, where human-driven and autonomous vehicles of varying commuter occupancy share a segment. Autonomous vehicles, with their ability to maintain shorter headways, boost traffic throughput. Our framework designates a toll lane for autonomous vehicles with high occupancy to use free of charge, while others pay a toll. We explore the lane choice equilibria when all vehicles minimize travel costs, and characterize the equilibria by ranking vehicles by their mobility enhancement potential, a concept we term the mobility degree. Through numerical examples, we demonstrate the framework's utility in addressing design challenges such as setting optimal tolls, determining occupancy thresholds, and designing lane policies, showing how it facilitates the integration of high-occupancy and autonomous vehicles. We also propose an algorithm for assigning rational tolls to decrease total commuter delay and examine the effects of toll non-compliance. Our findings suggest that self-interest-driven behavior mitigates moderate non-compliance impacts, highlighting the framework's resilience. This work presents a pioneering comprehensive analysis of a toll lane framework that emphasizes the coexistence of autonomous and high-occupancy vehicles, offering insights for traffic management improvements and the integration of autonomous vehicles into existing transportation infrastructures.

Our knowledge of galaxy formation and evolution has incredibly progressed through multi-wavelength observational constraints of the interstellar medium (ISM) of galaxies at all cosmic epochs. However, little is known about the physical properties of the more diffuse and lower surface brightness reservoir of gas and dust that extends beyond ISM scales and fills dark matter haloes of galaxies up to their virial radii, the circumgalactic medium (CGM). New theoretical studies increasingly stress the relevance of the latter for understanding the feedback and feeding mechanisms that shape galaxies across cosmic times, whose cumulative effects leave clear imprints into the CGM. Recent studies are showing that a -- so far unconstrained -- fraction of the CGM mass may reside in the cold (T < 1e4 K) molecular and atomic phase, especially in high-redshift dense environments. These gas phases, together with the warmer ionised phase, can be studied in galaxies from z ~ 0 to z ~ 10 through bright far-infrared and sub-millimeter emission lines such as [C II] 158${\mu}$m, [O III] 88 ${\mu}$m, [C I] 609${\mu}$m, [C I] 370${\mu}$m, and the rotational transitions of CO. Imaging such hidden cold CGM can lead to a breakthrough in galaxy evolution studies but requires a new facility with the specifications of the proposed Atacama Large Aperture Submillimeter Telescope (AtLAST). In this paper, we use theoretical and empirical arguments to motivate future ambitious CGM observations with AtLAST and describe the technical requirements needed for the telescope and its instrumentation to perform such science., Comment: Submitted to Open Research Europe as part of the AtLAST collection: https://open-research-europe.ec.europa.eu/collections/atlast/about

We are motivated primarily by the adaptation of text-to-speech synthesis models; however we argue that more generic parameter-efficient fine-tuning (PEFT) is an appropriate framework to do such adaptation. Nevertheless, catastrophic forgetting remains an issue with PEFT, damaging the pre-trained model's inherent capabilities. We demonstrate that existing Bayesian learning techniques can be applied to PEFT to prevent catastrophic forgetting as long as the parameter shift of the fine-tuned layers can be calculated differentiably. In a principled series of experiments on language modeling and speech synthesis tasks, we utilize established Laplace approximations, including diagonal and Kronecker-factored approaches, to regularize PEFT with the low-rank adaptation (LoRA) and compare their performance in pre-training knowledge preservation. Our results demonstrate that catastrophic forgetting can be overcome by our methods without degrading the fine-tuning performance, and using the Kronecker-factored approximation produces a better preservation of the pre-training knowledge than the diagonal ones.

The discipline of game theory was introduced in the context of economics, and has been applied to study cyber attacker and defender behaviors. While adaptions have been made to accommodate features in the cyber domain, these studies are inherently limited by the root of game theory in economic systems where players (i.e., agents) may be selfish but not malicious. In this SoK, we systematize the major cybersecurity problems that have been studied with the game-theoretic approach, the assumptions that have been made, the models and solution concepts that have been proposed. The systematization leads to a characterization of the technical gaps that must be addressed in order to make game-theoretic cybersecurity models truly useful. We explore bridges to address them., Comment: 21 pages, Finished October 17th, 2023

The enormous increase in mid-IR sensitivity and spatial and spectral resolution provided by the JWST spectrographs enables, for the first time, detailed extragalactic studies of molecular vibrational bands. This opens an entirely new window for the study of the molecular interstellar medium in luminous infrared galaxies (LIRGs). We present a detailed analysis of rovibrational bands of gas-phase CO, H$_2$O, C$_2$H$_2$ and HCN towards the heavily-obscured eastern nucleus of the LIRG VV 114, as observed by NIRSpec and MIRI MRS. Spectra extracted from apertures of 130 pc in radius show a clear dichotomy between the obscured AGN and two intense starburst regions. We detect the 2.3 $\mu$m CO bandheads, characteristic of cool stellar atmospheres, in the star-forming regions, but not towards the AGN. Surprisingly, at 4.7 $\mathrm{\mu}$m we find highly-excited CO ($T_\mathrm{ex} \approx 700-800$ K out to at least rotational level $J = 27$) towards the star-forming regions, but only cooler gas ($T_\mathrm{ex} \approx 200$ K) towards the AGN. We conclude that only mid-infrared pumping through the rovibrational lines can account for the equilibrium conditions found for CO and H$_2$O in the deeply-embedded starbursts. Here the CO bands probe regions with an intense local radiation field inside dusty young massive star clusters or near the most massive young stars. The lack of high-excitation molecular gas towards the AGN is attributed to geometric dilution of the intense radiation from the bright point source. An overview of the relevant excitation and radiative transfer physics is provided in an appendix., Comment: 28 pages, 17 figures. Accepted for publication in ApJ. This version includes small revisions following the referee report

Spitzer spectral maps reveal a disk of highly luminous, warm (>150 K) H2 in the center of the massive spiral galaxy Messier 58, which hosts a radio-loud AGN. The inner 2.6 kpc of the galaxy appears to be overrun by shocks from the radio jet cocoon. Gemini NIRI imaging of the H2 1-0 S(1) emission line, ALMA CO 2-1, and HST multiband imagery indicate that much of the molecular gas is shocked in-situ, corresponding to lanes of dusty molecular gas that spiral towards the galaxy nucleus. The CO 2-1 and ionized gas kinematics are highly disturbed, with velocity dispersion up to 300 km/s. Dissipation of the associated kinetic energy and turbulence, likely injected into the ISM by radio-jet driven outflows, may power the observed molecular and ionized gas emission from the inner disk. The PAH fraction and composition in the inner disk appear to be normal, in spite of the jet and AGN activity. The PAH ratios are consistent with excitation by the interstellar radiation field from old stars in the bulge, with no contribution from star formation. The phenomenon of jet-shocked H2 may substantially reduce star formation and help to regulate the stellar mass of the inner disk and supermassive black hole in this otherwise normal spiral galaxy. Similarly strong H2 emission is found at the centers of several nearby spiral and lenticular galaxies with massive bulges and radio-loud AGN., Comment: 18 pages, 12 figures, 1 Table, accepted for publication in the Astrophysical Journal

We present upgraded infrastructure for Searches after Gravitational Waves Using ARizona Observatories (SAGUARO) during LIGO, Virgo, and KAGRA's fourth gravitational-wave (GW) observing run (O4). These upgrades implement many of the lessons we learned after a comprehensive analysis of potential electromagnetic counterparts to the GWs discovered during the previous observing run. We have developed a new web-based target and observation manager (TOM) that allows us to coordinate sky surveys, vet potential counterparts, and trigger follow-up observations from one centralized portal. The TOM includes software that aggregates all publicly available information on the light curves and possible host galaxies of targets, allowing us to rule out potential contaminants like active galactic nuclei, variable stars, solar-system objects, and preexisting supernovae, as well as to assess the viability of any plausible counterparts. We have also upgraded our image-subtraction pipeline by assembling deeper reference images and training a new neural network-based real-bogus classifier. These infrastructure upgrades will aid coordination by enabling the prompt reporting of observations, discoveries, and analysis to the GW follow-up community, and put SAGUARO in an advantageous position to discover kilonovae in the remainder of O4 and beyond. Many elements of our open-source software stack have broad utility beyond multimessenger astronomy, and will be particularly relevant in the "big data" era of transient discoveries by the Vera C. Rubin Observatory., Comment: updated to match accepted version

Recently, large pretrained language models have demonstrated strong language understanding capabilities. This is particularly reflected in their zero-shot and in-context learning abilities on downstream tasks through prompting. To assess their impact on spoken language understanding (SLU), we evaluate several such models like ChatGPT and OPT of different sizes on multiple benchmarks. We verify the emergent ability unique to the largest models as they can reach intent classification accuracy close to that of supervised models with zero or few shots on various languages given oracle transcripts. By contrast, the results for smaller models fitting a single GPU fall far behind. We note that the error cases often arise from the annotation scheme of the dataset; responses from ChatGPT are still reasonable. We show, however, that the model is worse at slot filling, and its performance is sensitive to ASR errors, suggesting serious challenges for the application of those textual models on SLU., Comment: 6 pages, 2 figures; Accepted by Interspeech 2023

End-to-end spoken language understanding (SLU) remains elusive even with current large pretrained language models on text and speech, especially in multilingual cases. Machine translation has been established as a powerful pretraining objective on text as it enables the model to capture high-level semantics of the input utterance and associations between different languages, which is desired for speech models that work on lower-level acoustic frames. Motivated particularly by the task of cross-lingual SLU, we demonstrate that the task of speech translation (ST) is a good means of pretraining speech models for end-to-end SLU on both intra- and cross-lingual scenarios. By introducing ST, our models reach higher performance over baselines on monolingual and multilingual intent classification as well as spoken question answering using SLURP, MINDS-14, and NMSQA benchmarks. To verify the effectiveness of our methods, we also create new benchmark datasets from both synthetic and real sources, for speech summarization and low-resource/zero-shot transfer from English to French or Spanish. We further show the value of preserving knowledge for the ST pretraining task for better downstream performance, possibly using Bayesian transfer regularizers., Comment: 16 pages, 3 figures; accepted by Findings of EMNLP 2023

We present the first deep polarimetric study of Galactic synchrotron emission at low radio frequencies. Our study is based on 21 observations of the European Large Area Infrared Space Observatory Survey-North 1 (ELAIS-N1) field using the Low-Frequency Array (LOFAR) at frequencies from 114.9 to 177.4 MHz. These data are a part of the LOFAR Two-metre Sky Survey Deep Fields Data Release 1. We used very low-resolution ($4.3'$) Stokes QU data cubes of this release. We applied rotation measure (RM) synthesis to decompose the distribution of polarised structures in Faraday depth, and cross-correlation RM synthesis to align different observations in Faraday depth. We stacked images of about 150 hours of the ELAIS-N1 observations to produce the deepest Faraday cube at low radio frequencies to date, tailored to studies of Galactic synchrotron emission and the intervening magneto-ionic interstellar medium. This Faraday cube covers $\sim36~{\rm deg^{2}}$ of the sky and has a noise of $27~{\rm \mu Jy~PSF^{-1}~RMSF^{-1}}$ in polarised intensity. This is an improvement in noise by a factor of approximately the square root of the number of stacked data cubes ($\sim\sqrt{20}$), as expected, compared to the one in a single data cube based on five-to-eight-hour observations. We detect a faint component of diffuse polarised emission in the stacked cube, which was not detected previously. Additionally, we verify the reliability of the ionospheric Faraday rotation corrections estimated from the satellite-based total electron content measurements to be of $~\sim0.05~{\rm rad~m^{-2}}$. We also demonstrate that diffuse polarised emission itself can be used to account for the relative ionospheric Faraday rotation corrections with respect to a reference observation., Comment: 15 pages, 15 figures, accepted for publication in A&A

At the highest stellar masses (log(\mstar) $\gtrsim$ 11.5 \msun), only a small fraction of galaxies are disk-like and actively star-forming objects. These so-called `super spirals' are ideal objects to better understand how galaxy evolution proceeds and to extend our knowledge about the relation between stars and gas to a higher stellar mass regime. We present new CO(1-0) data for a sample of 46 super spirals and for 18 slightly lower-mass (log(\mstar) $>$ 11.0 \msun ) galaxies with broad HI lines -- HI fast-rotators (HI-FRs). We analyze their molecular gas mass, derived from CO, in relation to their star formation rate (SFR) and stellar mass, and compare the results to values and scaling relations derived from lower-mass galaxies. We confirm that super spirals follow the same star-forming main sequence (SFMS) as lower-mass galaxies. We find that they possess abundant molecular gas, which lies above the extrapolation of the scaling relation with stellar mass derived from lower-mass galaxies, but within the relation between \mmol/\mstar and the distance to the SFMS. The molecular gas depletion time, \taudep = \mmol/SFR, is higher than for lower-mass galaxies on the SFMS (\taudep = 9.30 $\pm$ 0.03, compared to \taudep = 9.00 $\pm$ 0.02 for the comparison sample) and seems to continue an increasing trend with stellar mass. HI-FR galaxies have an atomic-to-molecular gas mass ratio that is in agreement with that of lower-mass galaxies, indicating that the conversion from the atomic to molecular gas proceeds in a similar way. We conclude that the availability of molecular gas is a crucial factor to enable star formation to continue and that, if gas is present, quenching is not a necessary destiny for high-mass galaxies. The difference in gas depletion time suggests that the properties of the molecular gas at high stellar masses are less favorable for star formation., Comment: 28 pages, accepted for publication in Astronomy & Astrophysics

Given the recent success of diffusion in producing natural-sounding synthetic speech, we investigate how diffusion can be used in speaker adaptive TTS. Taking cues from more traditional adaptation approaches, we show that adaptation can be included in a diffusion pipeline using conditional layer normalization with a step embedding. However, we show experimentally that, whilst the approach has merit, such adaptation alone cannot approach the performance of Transformer-based techniques. In a second experiment, we show that diffusion can be optimally combined with Transformer, with the latter taking the bulk of the adaptation load and the former contributing to improved naturalness.

Feedback from both supermassive black holes and massive stars plays a fundamental role in the evolution of galaxies and the inter-galactic medium. In this paper we use available data to estimate the total amount of kinetic energy and momentum created per co-moving volume element over the history of the universe from three sources: massive stars and supernovae, radiation pressure and winds driven by supermassive black holes, and radio jets driven by supermassive black holes. Kinetic energy and momentum injection from jets peaks at z ~ 1, while the other two sources peak at z ~ 2. Massive stars are the dominant global source of momentum injection. For supermassive black holes, we find that the amount of kinetic energy from jets is about an order-of-magnitude larger than that from winds. We also find that amount of kinetic energy created by massive stars is about 2.5 epsilon times that carried by jets (where epsilon is the fraction of injected energy not lost to radiative cooling). We discuss the implications of these results for the evolution of galaxies and the IGM. Because the ratio of black hole mass to galaxy mass is a steeply increasing function of mass, we show that the relative importance of black hole feedback to stellar feedback likewise increases with mass. We show that there is a trend in the present-day universe which, in the simplest picture, is consistent with galaxies that have been dominated by black hole feedback being generally quenched, while galaxies that have been dominated by stellar feedback are star-forming. We also note that the amount of kinetic energy carried by jets and winds appears sufficient to explain the properties of hot gas in massive halos (> 10^13 solar masses).