Search

Star formation quenching is one of the key processes that shape the evolution of galaxies. In this study, we investigate the changes in molecular gas and star formation properties as galaxies transit from the star-forming main sequence to the passive regime. Our analysis reveals that as galaxies move away from the main sequence towards the green valley the radial profile of specific star formation rate surface density ($\Sigma_\mathrm{sSFR}$) is suppressed compared with main sequence galaxies out to a galactocentric radius of 1.5 $R_{e}$ ($\sim$ 7 kpc for our sample). By combining radial profiles of gas fraction ($f_\mathrm{gas}$) and star formation efficiency (SFE), we can discern the underlying mechanism that determines $\Sigma_\mathrm{sSFR}$ at different galactocentric radii. Analysis of relative contributions of $f_\mathrm{gas}$ and SFE to $\Sigma_\mathrm{sSFR}$ uncovers a diverse range of quenching modes. Star formation in approximately half of our quenching galaxies is primarily driven by a single mode (i.e. either $f_\mathrm{gas}$ or SFE), or a combination of both. A collective analysis of all galaxies reveals that the reduction in star formation within the central regions ($R$ $<$ 0.5 $R_{e}$) is primarily attributable to a decrease in SFE. Conversely, in the disk regions ($R$ $>$ 0.5 $R_{e}$), both $f_\mathrm{gas}$ and SFE contribute to the suppression of star formation. Our findings suggest that multiple quenching mechanisms may be at play in our sample galaxies, and even within a single galaxy. We also compare our observational outcomes with those from galaxy simulations and discuss the implications of our data., Comment: 27 pages, 8 figures, 1 table. Accepted for publication in The Astrophysical Journal

We present an analysis of the quenching of local observed and simulated galaxies, including an investigation of the dependence of quiescence on both intrinsic and environmental parameters. We apply an advanced machine learning technique utilizing random forest classification to predict when galaxies are star forming or quenched. We perform separate classification analyses for three groups of galaxies: (a) central galaxies; (b) high-mass satellites ($M_{*} > 10^{10.5}{\rm M_{\odot}}$); and (c) low-mass satellites ($M_{*} < 10^{10}{\rm M_{\odot}}$) for three cosmological hydrodynamical simulations (EAGLE, Illustris, and IllustrisTNG), and observational data from the SDSS. The simulation results are unanimous and unambiguous: quiescence in centrals and high-mass satellites is best predicted by intrinsic parameters (specifically central black hole mass), whilst it is best predicted by environmental parameters (specifically halo mass) for low-mass satellites. In observations, we find black hole mass to best predict quiescence for centrals and high mass satellites, exactly as predicted by the simulations. However, local galaxy over-density is found to be most predictive parameter for low-mass satellites. Nonetheless, both simulations and observations do agree that it is environment which quenches low mass satellites. We provide evidence which suggests that the dominance of local over-density in classifying low mass systems may be due to the high uncertainty in halo mass estimation from abundance matching, rather than it being fundamentally a more predictive parameter. Finally, we establish that the qualitative trends with environment predicted in simulations are recoverable in the observation space. This has important implications for future wide-field galaxy surveys., Comment: Accepted to MNRAS; 32 pages; 22 figures

We present observational evidence for a stellar Fundamental Metallicity Relation (FMR), a smooth relation between stellar mass, star-formation rate (SFR) and the light-weighted stellar metallicity of galaxies (analogous to the well-established gas-phase FMR). We use the flexible, non-parametric software pPXF to reconstruct simultaneously the star-formation and chemical-enrichment history of a representative sample of galaxies from the local MaNGA survey. We find that (i) the metallicity of individual galaxies increases with cosmic time and (ii) at all stellar masses, the metallicity of galaxies is progressively higher, moving from the star-burst region above the main sequence (MS) towards the passive galaxies below the MS, manifesting the stellar FMR. These findings are in qualitative agreement with theoretical expectations from IllustrisTNG, where we find a mass-weighted stellar FMR. The scatter is reduced when replacing the stellar mass $M_{*}$ with $M_{*}/R_{\rm e}$ (with $R_{\rm e}$ being the effective radius), in agreement with previous results using the velocity dispersion $\sigma_{\rm e}$, which correlates with $M_{*}/R_{\rm e}$. Our results point to starvation as the main physical process through which galaxies quench, showing that metal-poor gas accretion from the intergalactic/circumgalactic medium -- or the lack thereof -- plays an important role in galaxy evolution by simultaneously shaping both their star-formation and their metallicity evolutions, while outflows play a subordinate role. This interpretation is further supported by the additional finding of a young stellar FMR, tracing only the stellar populations formed in the last 300 Myr. This suggests a tight co-evolution of the chemical composition of both the gaseous interstellar medium and the stellar populations, where the gas-phase FMR is continuously imprinted onto the stars over cosmic times.

One of the most important questions in astrophysics is what causes galaxies to stop forming stars. Previous studies have shown a tight link between quiescence and black hole mass. Other studies have revealed that quiescence is also associated with 'starvation', the halting of gas inflows, which results in the remaining gas being used up rapidly by star formation and in rapid chemical enrichment. In this work we find the final missing link between these two findings. Using a large sample of galaxies, we uncover the intrinsic dependencies of the stellar metallicity on galaxy properties. In the case of the star-forming galaxies, the stellar metallicity is driven by stellar mass. However, for passive galaxies the stellar metallicity is primarily driven by the black hole mass, as traced by velocity dispersion. This result finally reveals the connection between previous studies, where the integrated effect of black hole feedback prevents gas inflows, starving the galaxy, which is seen by the rapid increase in the stellar metallicity, leading to the galaxy becoming passive., Comment: 20 pages, 6 figures, submitted to Nature Astronomy

We present the first statistical study of spatially integrated non-Gaussian stellar kinematics spanning 7 Gyr in cosmic time. We use deep, rest-frame optical spectroscopy of massive galaxies (stellar mass $M_\star > 10^{10.5} {\rm M}_\odot$) at redshifts z = 0.05, 0.3 and 0.8 from the SAMI, MAGPI and LEGA-C surveys, to measure the excess kurtosis $h_4$ of the stellar velocity distribution, the latter parametrised as a Gauss-Hermite series. We find that at all redshifts where we have large enough samples, $h_4$ anti-correlates with the ratio between rotation and dispersion, highlighting the physical connection between these two kinematic observables. In addition, and independently from the anti-correlation with rotation-to-dispersion ratio, we also find a correlation between $h_4$ and $M_\star$, potentially connected to the assembly history of galaxies. In contrast, after controlling for mass, we find no evidence of independent correlation between $h_4$ and aperture velocity dispersion or galaxy size. These results hold for both star-forming and quiescent galaxies. For quiescent galaxies, $h_4$ also correlates with projected shape, even after controlling for the rotation-to-dispersion ratio. At any given redshift, star-forming galaxies have lower $h_4$ compared to quiescent galaxies, highlighting the link between kinematic structure and star-forming activity., Comment: 26 pages, 15 figures Accepted for publication in MNRAS

We present the first study of spatially integrated higher-order stellar kinematics over cosmic time. We use deep rest-frame optical spectroscopy of quiescent galaxies at redshifts z=0.05, 0.3 and 0.8 from the SAMI, MAGPI and LEGA-C surveys to measure the excess kurtosis $h_4$ of the stellar velocity distribution, the latter parametrised as a Gauss-Hermite series. Conservatively using a redshift-independent cut in stellar mass ($M_\star = 10^{11}\,{\rm M}_\odot$), and matching the stellar-mass distributions of our samples, we find 7 $\sigma$ evidence of $h_4$ increasing with cosmic time, from a median value of 0.019$\pm$0.002 at z=0.8 to 0.059$\pm$0.004 at z=0.06. Alternatively, we use a physically motivated sample selection, based on the mass distribution of the progenitors of local quiescent galaxies as inferred from numerical simulations; in this case, we find 10 $\sigma$ evidence. This evolution suggests that, over the last 7 Gyr, there has been a gradual decrease in the rotation-to-dispersion ratio and an increase in the radial anisotropy of the stellar velocity distribution, qualitatively consistent with accretion of gas-poor satellites. These findings demonstrate that massive galaxies continue to accrete mass and increase their dispersion support after becoming quiescent., Comment: 19 pages, 9 figures Accepted for publication in MNRAS

Nebular HeII$\lambda$4686\AA~line emission is useful to unveil active galactic nuclei (AGN) residing in actively star-forming (SF) galaxies, typically missed by the standard BPT classification. Here we adopt the HeII diagnostic to identify hidden AGN in the Local Universe using for the first time spatially-resolved data from the Data Release 15 of the Mapping Nearby Galaxies at APO survey (MaNGA DR15). By combining results from HeII and BPT diagnostics, we overall select 459 AGN host candidates ($\sim$10% in MaNGA DR15), out of which 27 are identified as AGN by the HeII diagram only. The HeII-only AGN population is hosted by massive (M$_*\gtrsim10^{10}$ M$_{\odot}$) SF Main Sequence galaxies, and on average less luminous than the BPT-selected AGN. Given the HeII line faintness, we revisit our census accounting for incompleteness effects due to the HeII sensitivity limit of MaNGA. We thus obtain an overall increased fraction (11%) of AGN in MaNGA compared to the BPT-only census (9%), which further increases to 14% for galaxies more massive than $10^{10}$ M$_{\odot}$; interestingly, on the SF Main Sequence the increase is by about a factor of 2. A substantial number of AGN in SF galaxies points to significant, coeval star formation and black hole accretion, consistently with results from hydrodynamical simulations and with important implications on quenching scenarios. In view of exploring unprecedented high redshifts with JWST and new ground-based facilities, revisiting the standard BPT classification through novel emission-line diagnostics is fundamental to discover AGN in highly SF environments., Comment: 13 pages, 12 figures. Accepted for publication in MNRAS

We analyse the chemical properties of three z similar to 8 galaxies behind the galaxy cluster SMACSJ0723.3-7327, observed as part of the Early Release Observations programme of the James Webb Space Telescope. Exploiting [O III]lambda 4363 auroral line detections in NIRSpec spectra, we robustly apply the direct T-e method for the very first time at such high redshift, measuring metallicities ranging from extremely metal poor (12 + log(O/H)approximate to 7) to about one-third solar. We also discuss the excitation properties of these sources, and compare them with local strong-line metallicity calibrations. We find that none of the considered diagnostics match simultaneously the observed relations between metallicity and strong-line ratios for the three sources, implying that a proper re-assessment of the calibrations may be needed at these redshifts. On the mass-metallicity plane, the two galaxies at z similar to 7.6 (log(M-*/M-circle dot) = 8.1, 8.7) have metallicities that are consistent with the extrapolation of the mass-metallicity relation at z similar to 2-3, while the least massive galaxy at z similar to 8.5 (Log(M-*/M-circle dot) = 7.8) shows instead a significantly lower metallicity. The three galaxies show different level of offset relative to the Fundamental Metallicity Relation, with two of them (at z similar to 7.6) being marginally consistent, while the z similar to 8.5 source deviating significantly, being probably far from the smooth equilibrium between gas flows, star formation, and metal enrichment in place at later epochs.

We analyse the chemical properties of three z similar to 8 galaxies behind the galaxy cluster SMACSJ0723.3-7327, observed as part of the Early Release Observations programme of the James Webb Space Telescope. Exploiting [O III]lambda 4363 auroral line detections in NIRSpec spectra, we robustly apply the direct T-e method for the very first time at such high redshift, measuring metallicities ranging from extremely metal poor (12 + log(O/H)approximate to 7) to about one-third solar. We also discuss the excitation properties of these sources, and compare them with local strong-line metallicity calibrations. We find that none of the considered diagnostics match simultaneously the observed relations between metallicity and strong-line ratios for the three sources, implying that a proper re-assessment of the calibrations may be needed at these redshifts. On the mass-metallicity plane, the two galaxies at z similar to 7.6 (log(M-*/M-circle dot) = 8.1, 8.7) have metallicities that are consistent with the extrapolation of the mass-metallicity relation at z similar to 2-3, while the least massive galaxy at z similar to 8.5 (Log(M-*/M-circle dot) = 7.8) shows instead a significantly lower metallicity. The three galaxies show different level of offset relative to the Fundamental Metallicity Relation, with two of them (at z similar to 7.6) being marginally consistent, while the z similar to 8.5 source deviating significantly, being probably far from the smooth equilibrium between gas flows, star formation, and metal enrichment in place at later epochs.

We identify the intrinsic dependence of star formation quenching on a variety of galactic and environmental parameters, utilizing a machine learning approach with Random Forest classification. We have previously demonstrated the power of this technique to isolate causality, not mere correlation, in complex astronomical data. First, we analyze three cosmological hydrodynamical simulations (Eagle, Illustris, and IllustrisTNG), selecting snapshots spanning the bulk of cosmic history from comic noon ($z \sim 2$) to the present epoch, with stellar masses in the range $9 < \log(M_*/M_{\odot}) < 12$. In the simulations, black hole mass is unanimously found to be the most predictive parameter of central galaxy quenching at all epochs. Perhaps surprisingly, black hole accretion rate (and hence the bolometric luminosity of active galactic nuclei, AGN) is found to be of little predictive power over quenching. This theoretical result is important for observational studies of galaxy quenching as it cautions against using the current AGN state of a galaxy as a useful proxy for the cumulative impact of AGN feedback on a galactic system. The latter is traced by black hole mass not AGN luminosity. Additionally, we explore a sub-set of 'observable' parameters, which can be readily measured in extant wide-field galaxy surveys targeting $z = 0 - 2$, at $9 < \log(M_*/M_{\odot}) < 12$. All three simulations predict that in lieu of black hole mass, the stellar gravitational potential will outperform the other parameters in predicting quenching. We confirm this theoretical prediction observationally in the SDSS (at low redshifts) and in CANDELS (at intermediate and high redshifts)., Comment: Accepted to ApJ. 26 pages; 12 figures

Parsec-scale separation supermassive black hole binaries in the centre of gas-rich galaxy merger remnants could be surrounded by massive circumbinary discs (CBDs). Black hole mass and spin evolution during the gas-rich binary inspiral are crucial in determining the direction and power of relativistic jets that radio observations with LOFAR and SKAO will probe, and for predicting gravitational wave (GW) emission that IPTA and LISA will measure. We present 3D hydrodynamic simulations capturing gas-rich, self-gravitating CBDs around a $2\times 10^6$M$_{\odot}$ supermassive black hole binary, that probe different mass ratios, eccentricities and inclinations. We employ a sub-grid Shakura-Sunyaev accretion disc to self-consistently model black hole mass and spin evolution together with super-Lagrangian refinement techniques to resolve gas flows, streams and mini-discs within the cavity, which play a fundamental role in torquing and feeding the binary. We find that higher mass ratio and eccentric binaries result in larger cavities, while retrograde binaries result in smaller cavities. All of the simulated binaries are expected to shrink with net gravitational torques being negative. Unlike previous simulations, we do not find preferential accretion onto the secondary black hole. This implies smaller chirp masses at coalescence and hence a weaker GW background. Critically this means that spin-alignment is faster than the binary inspiral timescale even for low mass ratios. However, we find that mini-disc and hence spin alignment is not guaranteed in initially misaligned systems, potentially leading to a significant fraction of recoiled remnants displaced from their host galaxies if chaotic accretion is the dominant feeding channel., Comment: 27 pages, 14 figures, submitted to MNRAS. Comments and feedback welcome

We present an analysis of the quenching of star formation in massive galaxies ($M_* > 10^{9.5} M_\odot$) within the first 0.5 - 3 Gyr of the Universe's history utilizing JWST-CEERS data. We utilize a combination of advanced statistical methods to accurately constrain the intrinsic dependence of quenching in a multi-dimensional and inter-correlated parameter space. Specifically, we apply Random Forest (RF) classification, area statistics, and a partial correlation analysis to the JWST-CEERS data. First, we identify the key testable predictions from two state-of-the-art cosmological simulations (IllustrisTNG & EAGLE). Both simulations predict that quenching should be regulated by supermassive black hole mass in the early Universe. Furthermore, both simulations identify the stellar potential ($\phi_*$) as the optimal proxy for black hole mass in photometric data. In photometric observations, where we have no direct constraints on black hole masses, we find that the stellar potential is the most predictive parameter of massive galaxy quenching at all epochs from $z = 0 - 8$, exactly as predicted by simulations for this sample. The stellar potential outperforms stellar mass, galaxy size, galaxy density, and S\'ersic index as a predictor of quiescence at all epochs probed in JWST-CEERS. Collectively, these results strongly imply a stable quenching mechanism operating throughout cosmic history, which is closely connected to the central gravitational potential in galaxies. This connection is explained in cosmological models via massive black holes forming and growing in deep potential wells, and subsequently quenching galaxies through a mix of ejective and preventative active galactic nucleus (AGN) feedback., Comment: Accepted to ApJ. 32 pages; 10 figures

We present an analysis of the quenching of star formation in galaxies, bulges, and disks throughout the bulk of cosmic history, from $z=2-0$. We utilise observations from the SDSS and MaNGA at low redshifts. We complement these data with observations from CANDELS at high redshifts. Additionally, we compare the observations to detailed predictions from the LGalaxies semi-analytic model. To analyse the data, we developed a machine learning approach utilising a Random Forest classifier. We first demonstrate that this technique is extremely effective at extracting causal insight from highly complex and inter-correlated model data, before applying it to various observational surveys. Our primary observational results are as follows: At all redshifts studied in this work, we find bulge mass to be the most predictive parameter of quenching, out of the photometric parameter set (incorporating bulge mass, disk mass, total stellar mass, and $B/T$ structure). Moreover, we also find bulge mass to be the most predictive parameter of quenching in both bulge and disk structures, treated separately. Hence, intrinsic galaxy quenching must be due to a stable mechanism operating over cosmic time, and the same quenching mechanism must be effective in both bulge and disk regions. Despite the success of bulge mass in predicting quenching, we find that central velocity dispersion is even more predictive (when available in spectroscopic data sets). In comparison to the LGalaxies model, we find that all of these observational results may be consistently explained through quenching via preventative `radio-mode' active galactic nucleus (AGN) feedback. Furthermore, many alternative quenching mechanisms (including virial shocks, supernova feedback, and morphological stabilisation) are found to be inconsistent with our observational results and those from the literature., Comment: Accepted to A&A; 40 pages; 20 figures

We present an analysis of the quenching of star formation in galaxies, bulges, and disks throughout the bulk of cosmic history, from $z=2-0$. We utilise observations from the SDSS and MaNGA at low redshifts. We complement these data with observations from CANDELS at high redshifts. Additionally, we compare the observations to detailed predictions from the LGalaxies semi-analytic model. To analyse the data, we developed a machine learning approach utilising a Random Forest classifier. We first demonstrate that this technique is extremely effective at extracting causal insight from highly complex and inter-correlated model data, before applying it to various observational surveys. Our primary observational results are as follows: At all redshifts studied in this work, we find bulge mass to be the most predictive parameter of quenching, out of the photometric parameter set (incorporating bulge mass, disk mass, total stellar mass, and $B/T$ structure). Moreover, we also find bulge mass to be the most predictive parameter of quenching in both bulge and disk structures, treated separately. Hence, intrinsic galaxy quenching must be due to a stable mechanism operating over cosmic time, and the same quenching mechanism must be effective in both bulge and disk regions. Despite the success of bulge mass in predicting quenching, we find that central velocity dispersion is even more predictive (when available in spectroscopic data sets). In comparison to the LGalaxies model, we find that all of these observational results may be consistently explained through quenching via preventative `radio-mode' active galactic nucleus (AGN) feedback. Furthermore, many alternative quenching mechanisms (including virial shocks, supernova feedback, and morphological stabilisation) are found to be inconsistent with our observational results and those from the literature., Comment: Accepted to A&A; 40 pages; 20 figures

In this paper we investigate how massive central galaxies cease their star formation by comparing theoretical predictions from cosmological simulations: EAGLE, Illustris and IllustrisTNG with observations of the local Universe from the Sloan Digital Sky Survey (SDSS). Our machine learning (ML) classification reveals supermassive black hole mass ($M_{\rm BH}$) as the most predictive parameter in determining whether a galaxy is star forming or quenched at redshift $z=0$ in all three simulations. This predicted consequence of active galactic nucleus (AGN) quenching is reflected in the observations, where it is true for a range of indirect estimates of $M_{\rm BH}$ via proxies as well as its dynamical measurements. Our partial correlation analysis shows that other galactic parameters lose their strong association with quiescence, once their correlations with $M_{\rm BH}$ are accounted for. In simulations we demonstrate that it is the integrated power output of the AGN, rather than its instantaneous activity, which causes galaxies to quench. Finally, we analyse the change in molecular gas content of galaxies from star forming to passive populations. We find that both gas fractions ($f_{\rm gas}$) and star formation efficiencies (SFEs) decrease upon transition to quiescence in the observations but SFE is more predictive than $f_{\rm gas}$ in the ML passive/star-forming classification. These trends in the SDSS are most closely recovered in IllustrisTNG and are in direct contrast with the predictions made by Illustris. We conclude that a viable AGN feedback prescription can be achieved by a combination of preventative feedback and turbulence injection which together quench star formation in central galaxies., Comment: 29 pages, 16 figures, accepted for publication in MNRAS

In this paper we investigate how massive central galaxies cease their star formation by comparing theoretical predictions from cosmological simulations: EAGLE, Illustris and IllustrisTNG with observations of the local Universe from the Sloan Digital Sky Survey (SDSS). Our machine learning (ML) classification reveals supermassive black hole mass ($M_{\rm BH}$) as the most predictive parameter in determining whether a galaxy is star forming or quenched at redshift $z=0$ in all three simulations. This predicted consequence of active galactic nucleus (AGN) quenching is reflected in the observations, where it is true for a range of indirect estimates of $M_{\rm BH}$ via proxies as well as its dynamical measurements. Our partial correlation analysis shows that other galactic parameters lose their strong association with quiescence, once their correlations with $M_{\rm BH}$ are accounted for. In simulations we demonstrate that it is the integrated power output of the AGN, rather than its instantaneous activity, which causes galaxies to quench. Finally, we analyse the change in molecular gas content of galaxies from star forming to passive populations. We find that both gas fractions ($f_{\rm gas}$) and star formation efficiencies (SFEs) decrease upon transition to quiescence in the observations but SFE is more predictive than $f_{\rm gas}$ in the ML passive/star-forming classification. These trends in the SDSS are most closely recovered in IllustrisTNG and are in direct contrast with the predictions made by Illustris. We conclude that a viable AGN feedback prescription can be achieved by a combination of preventative feedback and turbulence injection which together quench star formation in central galaxies., Comment: 29 pages, 16 figures, accepted for publication in MNRAS

We investigate how star formation quenching proceeds within central and satellite galaxies using spatially resolved spectroscopy from the SDSS-IV MaNGA DR15. We adopt a complete sample of star formation rate surface densities ($\Sigma_{\rm SFR}$), derived in Bluck et al. (2020), to compute the distance at which each spaxel resides from the resolved star forming main sequence ($\Sigma_{\rm SFR} - \Sigma_*$ relation): $\Delta \Sigma_{\rm SFR}$. We study galaxy radial profiles in $\Delta \Sigma_{\rm SFR}$, and luminosity weighted stellar age (${\rm Age_L}$), split by a variety of intrinsic and environmental parameters. Via several statistical analyses, we establish that the quenching of central galaxies is governed by intrinsic parameters, with central velocity dispersion ($\sigma_c$) being the most important single parameter. High mass satellites quench in a very similar manner to centrals. Conversely, low mass satellite quenching is governed primarily by environmental parameters, with local galaxy over-density ($\delta_5$) being the most important single parameter. Utilising the empirical $M_{BH}$ - $\sigma_c$ relation, we estimate that quenching via AGN feedback must occur at $M_{BH} \geq 10^{6.5-7.5} M_{\odot}$, and is marked by steeply rising $\Delta \Sigma_{\rm SFR}$ radial profiles in the green valley, indicating `inside-out' quenching. On the other hand, environmental quenching occurs at over-densities of 10 - 30 times the average galaxy density at z$\sim$0.1, and is marked by steeply declining $\Delta \Sigma_{\rm SFR}$ profiles, indicating `outside-in' quenching. Finally, through an analysis of stellar metallicities, we conclude that both intrinsic and environmental quenching must incorporate significant starvation of gas supply., Comment: Accepted for publication in MNRAS. 40 pages; 24 figures

We present an analysis of star formation and quenching in the SDSS-IV MaNGA-DR15, utilising over 5 million spaxels from $\sim$3500 local galaxies. We estimate star formation rate surface densities ($\Sigma_{\rm SFR}$) via dust corrected $H\alpha$ flux where possible, and via an empirical relationship between specific star formation rate (sSFR) and the strength of the 4000 Angstrom break (D4000) in all other cases. We train a multi-layered artificial neural network (ANN) and a random forest (RF) to classify spaxels into `star forming' and `quenched' categories given various individual (and groups of) parameters. We find that global parameters (pertaining to the galaxy as a whole) perform collectively the best at predicting when spaxels will be quenched, and are substantially superior to local/ spatially resolved and environmental parameters. Central velocity dispersion is the best single parameter for predicting quenching in central galaxies. We interpret this observational fact as a probable consequence of the total integrated energy from AGN feedback being traced by the mass of the black hole, which is well known to correlate strongly with central velocity dispersion. Additionally, we train both an ANN and RF to estimate $\Sigma_{\rm SFR}$ values directly via regression in star forming regions. Local/ spatially resolved parameters are collectively the most predictive at estimating $\Sigma_{\rm SFR}$ in these analyses, with stellar mass surface density at the spaxel location ($\Sigma_*$) being by far the best single parameter. Thus, quenching is fundamentally a global process but star formation is governed locally by processes within each spaxel., Comment: Accepted for publication in MNRAS. 45 pages; 20 figures

We present an analysis of star formation and quenching in the SDSS-IV MaNGA-DR15, utilising over 5 million spaxels from $\sim$3500 local galaxies. We estimate star formation rate surface densities ($\Sigma_{\rm SFR}$) via dust corrected $H\alpha$ flux where possible, and via an empirical relationship between specific star formation rate (sSFR) and the strength of the 4000 Angstrom break (D4000) in all other cases. We train a multi-layered artificial neural network (ANN) and a random forest (RF) to classify spaxels into `star forming' and `quenched' categories given various individual (and groups of) parameters. We find that global parameters (pertaining to the galaxy as a whole) perform collectively the best at predicting when spaxels will be quenched, and are substantially superior to local/ spatially resolved and environmental parameters. Central velocity dispersion is the best single parameter for predicting quenching in central galaxies. We interpret this observational fact as a probable consequence of the total integrated energy from AGN feedback being traced by the mass of the black hole, which is well known to correlate strongly with central velocity dispersion. Additionally, we train both an ANN and RF to estimate $\Sigma_{\rm SFR}$ values directly via regression in star forming regions. Local/ spatially resolved parameters are collectively the most predictive at estimating $\Sigma_{\rm SFR}$ in these analyses, with stellar mass surface density at the spaxel location ($\Sigma_*$) being by far the best single parameter. Thus, quenching is fundamentally a global process but star formation is governed locally by processes within each spaxel., Comment: Accepted for publication in MNRAS. 45 pages; 20 figures