Publications

The redshift range z=4-6 marks a transition phase between primordial and mature galaxy formation in which galaxies considerably increase their stellar mass, metallicity, and dust content. The study of galaxies in this redshift range is therefore important to understand early galaxy formation and the fate of galaxies at later times. Here, we investigate the burstiness of the recent star-formation history (SFH) of 221 z~4.5 main-sequence galaxies at log(M) > 9.7 by comparing their ultra-violet (UV) continuum, H? luminosity, and Hα equivalent-width (EW). The Hα properties are derived from the Spitzer [3.6μm]-[4.5μm] broad-band color, thereby properly taking into account model and photometric uncertainties. We find a significant scatter between H? and UV-derived luminosities and star-formation rates (SFRs). About half of the galaxies show a significant excess in H? compared to expectations from a constant smooth SFH. We also find a tentative anti-correlation between Hα EW and stellar mass, ranging from 1000A at log(M) < 10 to below 100A at log(M) > 11. Consulting models suggests that most z~4.5 galaxies had a burst of star-formation within the last 50 Myrs, increasing their SFRs by a factor of > 5. The most massive galaxies on the other hand might decrease their SFRs, and may be transitioning to a quiescent stage by z=4. We identify differential dust attenuation (f) between stars and nebular regions as the main contributor to the uncertainty. With local galaxies selected by increasing H? EW (reaching values similar to high-z galaxies), we predict that f approaches unity at z>4 consistent with the extrapolation of measurements out to z=2.

Machine-learning (ML) algorithms will play a crucial role in studying the large data sets delivered by new facilities over the next decade and beyond. Here, we investigate the capabilities and limits of such methods in finding galaxies with brightness-variable active galactic nuclei (AGNs). Specifically, we focus on an unsupervised method based on self-organizing maps (SOM) that we apply to a set of nonparametric variability estimators. This technique allows us to maintain domain knowledge and systematics control while using all the advantages of ML. Using simulated light curves that match the noise properties of observations, we verify the potential of this algorithm in identifying variable light curves. We then apply our method to a sample of ~8300 WISE color-selected AGN candidates in Stripe 82, in which we have identified variable light curves by visual inspection. We find that with ML we can identify these variable classified AGN with a purity of 86% and a completeness of 66%, a performance that is comparable to that of more commonly used supervised deep-learning neural networks. The advantage of the SOM framework is that it enables not only a robust identification of variable light curves in a given data set, but it is also a tool to investigate correlations between physical parameters in multidimensional space�such as the link between AGN variability and the properties of their host galaxies. Finally, we note that our method can be applied to any time-sampled light curve (e.g., supernovae, exoplanets, pulsars, and other transient events).

Thanks to deep optical to near-IR imaging and spectroscopy, significant progress is made in characterizing the rest-frame UV to optical properties of galaxies in the early universe (z > 4). Surveys with Hubble, Spitzer, and ground-based facilities (Keck, Subaru, and VLT) provide spectroscopic and photometric redshifts, measurements of the spatial structure, stellar masses, and optical emission lines for large samples of galaxies. Recently, the Atacama Large (Sub) Millimeter Array (ALMA) has become a major player in pushing studies of high redshift galaxies to far-infrared wavelengths, hence making panchromatic surveys over many orders of frequencies possible. While past studies focused mostly on bright sub-millimeter galaxies, the sensitivity of ALMA now enables surveys like ALPINE, which focuses on measuring the gas and dust properties of a large sample of normal main-sequence galaxies at z > 4. Combining observations across different wavelengths into a single, panchromatic picture of galaxy formation and evolution is currently and in the future an important focus of the astronomical community.

We report the detection of CO(2-1) line emission from a Lyman Break Galaxy at z = 5.7 with the VLA. The CO line luminosity implies a massive molecular gas reservoir of (1.3±0.3)(α_CO/4.5M_☉ (K km s^-1 pc²)^-1) × 10¹¹ M_☉, suggesting low star formation efficiency, with a gas depletion timescale of order 1 Gyr. This efficiency is much lower than traditionally observed in z > 5 starbursts, indicating that star forming conditions in Main Sequence galaxies at z = 6 may be comparable to those of normal galaxies probed up to z = 3 to-date, but with rising gas fractions across the entire redshift range. We also obtain a deep CO upper limit for a Main Sequence galaxy at z = 5.3 with ~3 times lower SFR, perhaps implying a high α_CO conversion factor, as typically found in low metallicity galaxies. Using ALMA, we find faint [NII] 205μm emission relative to [CII] in all but the most IR-luminous "normal" galaxies at z = 5-6 for a sample including both CO targets, suggesting more intense or harder radiation fields in the ionized gas relative to lower redshift. These radiation properties indicate low metallicity may be common in typical ~10¹⁰ M_☉ galaxies at z = 5-6, consistent with our CO measurements. Our sample shows evidence for high dust temperatures, and a young starburst producing high radiation intensity and hardness even with substantial dust obscuration. While a fraction of Main Sequence star formation in the first billion years may take place in conditions not dissimilar to lower redshift, lower metallicity may affect the remainder of the population.

We present an empirical parameterization of the [NII]/Hα flux ratio as a function of stellar mass and redshift valid at 0 < z < 2.7 and 8.5 < log(M) < 11.0. This description can easily be applied to (i) simulations for modeling [NII] line emission, (ii) deblend [NII] and Hα in current low-resolution grism and narrow-band observations to derive intrinsic HHα fluxes, and (iii) to reliably forecast the number counts of Hα emission line galaxies for future surveys such as those planned for Euclid and WFIRST. Our model combines the evolution of the locus on the BPT diagram measured in spectroscopic data out to z ~ 2.5 with the strong dependence of [NII]/Hα on stellar mass and [OIII]/Hβ observed in local galaxy samples. We find large variations in the [NII]/Hα flux ratio at a fixed redshift due to its dependency on stellar mass, hence the assumption of a constant [NII] contamination fraction can lead to a significant under- or over-estimate of Hα luminosities. Specifically, measurements of the intrinsic Hα luminosity function derived from current low-resolution grism spectroscopy assuming a constant 29% contamination of [NII] are likely over-estimated by factors of 2-4 at log(L) > 43.0 and systematically under-estimated by ~50 at log(L) < 42.5 at redshifts z ~ 1.5. This has implications on the prediction of H-alpha emitters for Euclid and WFIRST. We also study the impact of blended Hα and [NII] on the accuracy of measured spectroscopic redshifts.

We trace the specific star formation rate (sSFR) of massive star-forming galaxies (> 10¹⁰ M_☉) from z = 2 to 7. Our method is substantially different from previous analyses, as it does not rely on direct estimates of star formation rate, but on the differential evolution of the galaxy stellar mass function (SMF). We show the reliability of this approach by means of semianalytical and hydrodynamical cosmological simulations. We then apply it to real data, using the SMFs derived in the COSMOS and CANDELS fields. We find that the sSFR is proportional to (1+z)^1.1±0.2 at z > 2, in agreement with other observations but in tension with the steeper evolution predicted by simulations from z = 4 to 2. We investigate the impact of several sources of observational bias, which, however, cannot account for this discrepancy. Although the SMF of high-redshift galaxies is still affected by significant errors, we show that future large-area surveys will substantially reduce them, making our method an effective tool to probe the massive end of the main sequence of star-forming galaxies.

Recent studies have found a significant evolution and scatter in the IRX-β relation at z > 4, suggesting different dust properties of these galaxies. The total far-infrared (FIR) luminosity is key for this analysis but poorly constrained in normal (main-sequence) star-forming z > 5 galaxies where often only one single FIR point is available. To better inform estimates of the FIR luminosity, we construct a sample of local galaxies and three low-redshift analogs of z > 5 systems. The trends in this sample suggest that normal high-redshift galaxies have a warmer infrared (IR) SED compared to average z < 4 galaxies that are used as prior in these studies. The blue-shifted peak and mid-IR excess emission could be explained by a combination of a larger fraction of the metal-poor inter-stellar medium (ISM) being optically thin to ultra-violet (UV) light and a stronger UV radiation field due to high star formation densities. Assuming a maximally warm IR SED suggests 0.6 dex increased total FIR luminosities, which removes some tension between dust attenuation models and observations of the IRX-β relation at z > 5. Despite this, some galaxies still fall below the minimum IRX-? relation derived with standard dust cloud models. We propose that radiation pressure in these highly star-forming galaxies causes a spatial offset between dust clouds and young star-forming regions within the lifetime of O/B stars. These offsets change the radiation balance and create viewing-angle effects that can change UV colors at fixed IRX. We provide a modified model that can explain the location of these galaxies on the IRX-β diagram.

We examine the rest-frame ultra-violet (UV) properties of 10 [C II]λ158μm-detected galaxies at z ~ 5.5 in COSMOS using new HST/WFC3 near-infrared imaging. Together with pre-existing 158μm-continuum and [C II] line measurements by ALMA, we study their dust attenuation properties on the IRX-β diagram, which connects the total dust emission (~IRX = log(L_FIR/L₁₆₀₀)) to the line-of-sight dust column (~β). We find systematically bluer UV continuum spectral slopes (β) compared to previous low-resolution ground-based measurements, which relieves some of the tension between models of dust attenuation and observations at high redshifts. While most of the galaxies are consistent with local starburst or Small Magellanic cloud like dust properties, we find galaxies with low IRX values and a large range in β that cannot be explained by models of a uniform dust distribution well mixed with stars. A stacking analysis of Keck/DEIMOS optical spectra indicates that these galaxies are metal-poor with young stellar populations which could significantly alter their spatial dust distribution.

We use >9400 log(m/M_sun)>10 quiescent and star-forming galaxies at z < 2 in COSMOS/UltraVISTA to study the average size evolution of these systems, with focus on the rare, ultra-massive population at log(m/M_sun)>11.4. The large 2-square degree survey area delivers a sample of ~400 such ultra-massive systems. Accurate sizes are derived using a calibration based on high-resolution images from the Hubble Space Telescope. We find that, at these very high masses, the size evolution of star-forming and quiescent galaxies is almost indistinguishable in terms of normalization and power-law slope. We use this result to investigate possible pathways of quenching massive m > M* galaxies at z < 2. We consistently model the size evolution of quiescent galaxies from the star-forming population by assuming different simple models for the suppression of star-formation. These models include an instantaneous and delayed quenching without altering the structure of galaxies and a central starburst followed by compaction. We find that instantaneous quenching reproduces well the observed mass-size relation of massive galaxies at z > 1. Our starburst+compaction model followed by individual growth of the galaxies by minor mergers is preferred over other models without structural change for log(m/M_sun) > 11.0 galaxies at z > 0.5. None of our models is able to meet the observations at m > M* and z < 1 with out significant contribution of post-quenching growth of individual galaxies via mergers. We conclude that quenching is a fast process in galaxies with m > 10¹¹ M_sun, and that major mergers likely play a major role in the final steps of their evolution.

The intrinsic escape fraction of ionizing Lyman continuum photons (f_esc) is crucial to understand whether galaxies are capable of reionizing the neutral hydrogen in the early universe at z>6. Unfortunately, it is not possible to access fesc at z>4 with direct observations and the handful of measurements from low redshift galaxies consistently find f_esc < 10%, while at least f_esc ~ 10% is necessary for galaxies dominate reionization. Here, we present the first empirical prediction of f_esc at z>6 by combining the (sparsely populated) relation between [OIII]/[OII] and fesc with the redshift evolution of [OIII]/[OII] as predicted from local high-z analogs selected by their H? equivalent-width. We find f_esc = 5.7 (+8.3) (-3.3)% at z=6 and fesc = 10.4 (+15.5) (-6.3)% at z=9 for galaxies with log(M/M_sun) ~ 9.0 (errors given as 1σ). However, there is a negative correlation with stellar mass and we find up to 50% larger fesc per 0.5 dex decrease in stellar mass. The population averaged escape fraction increases according to f_esc = f_esc,0((1+z)/3)^a, with f_esc,0=2.3 +/-0.05% and a=1.17 +/-0.02 at z>2 for log(M/M_sun) ~ 9.0. With our empirical prediction of f_esc (thus fixing an important previously unknown variable) and further reasonable assumption on clumping factor and the production efficiency of Lyman continuum photons, we conclude that the average population of galaxies is just capable to reionize the universe by z~6.

The offset of high redshift star-forming galaxies in the [OII]/Hβ versus [NII]/Hα (O3N2) diagram in comparison with the local star-forming galaxy sequence is now well established. The physical origin of the shift is the subject of some debate, and has important implications for metallicity measurements based on strong lines at all redshifts. To investigate the origin of the O3N2 offset, we use a sample of ~100,000 star-forming galaxies from SDSS DR12 to probe the empirical correlations between emission line diagnostics and measurable galaxy physical properties. In particular, we examine how surface density of star formation (SFRD), ionization parameter, nitrogen-to-oxygen (N/O) ratio, and stellar mass drive position in two key diagnostic diagrams: O3N2 and [OIII]/Hβ versus [SII]/Hα (O3S2). We show that, at fixed [OIII]/Hβ, galaxies falling closer to the high-redshift locus in O3N2 have higher SFRD, stellar mass and N/O ratios. We also find a tight correspondence in the distributions of stellar mass and N/O in the diagnostic diagrams. This relation, spanning a range of galaxy evolutionary states, suggests that the N/O-M_star relation is more fundamental than the N/O-metallicity relation. We argue that a tight N/O-M_star relation is well-motivated physically, and that the observed correlation of N/O with O/H in the local universe is primarily a reflection of the existence of the mass-metallicity relation. Because the mass-metallicity relation evolves much more rapidly with redshift than N/O-Mstar, the N/O ratios of high redshift galaxies are significantly elevated in comparison with local galaxies with the same gas-phase O/H. The O3N2 shift and elevated N/O ratios observed in high redshift galaxies therefore come about as a natural consequence of the N/O-Mstar relation combined with the evolution of the mass-metallicity relation.

We measure a relation between the depth of four prominent rest-UV absorption complexes and metallicity for local galaxies and verify it up to z~3. We then apply this relation to a sample of 224 galaxies at 3.5 < z < 6.0 (z~4.8) in COSMOS, for which unique UV spectra from DEIMOS and accurate stellar masses from SPLASH are available. The average galaxy population at z~5 and log(M/M_sun) > 9 is characterized by 0.3-0.4 dex (in units of 12+log(O/H)) lower metallicities than at z~2, but comparable to z~3.5. We find galaxies with weak/no Ly-alpha emission to have metallicities comparable to z~2 galaxies and therefore may represent an evolved sub-population of z~5 galaxies. We find a correlation between metallicity and dust in good agreement with local galaxies and an inverse trend between metallicity and star-formation rate (SFR) consistent with observations at z~2. The relation between stellar mass and metallicity (MZ relation) is similar to z~3.5, however, there are indications of it being slightly shallower, in particular for the young, Lyα emitting galaxies. We show that, within a "bathtub" approach, a shallower MZ relation is expected in the case of a fast (exponential) build-up of stellar mass with an e-folding time of 100-200 Myr. Due to this fast evolution, the process of dust production and metal enrichment as a function of mass could be more stochastic in the first billion years of galaxy formation compared to later times.

We measure the H-alpha and [OIII] emission line properties as well as specific star-formation rates (sSFR) of spectroscopically confirmed 3 < z < 6 galaxies in COSMOS from their observed colors vs. redshift evolution. Our model describes consistently the ensemble of galaxies including intrinsic properties (age, metallicity, star-formation history), dust-attenuation, and optical emission lines. We forward-model the measured Hα equivalent-widths (EW) to obtain the sSFR out to z~6 without stellar mass fitting. We find a strongly increasing rest-frame Hα EW that is flattening off above z~2.5 with average EWs of 300-600A at z~6. The sSFR is increasing proportional to (1+z)^2.4 at z < 2.2 and (1+z)^1.5 at higher redshifts, indicative of a fast mass build-up in high-z galaxies within e-folding times of 100-200Myr at z~6. The redshift evolution at z > 3 cannot be fully explained in a picture of cold accretion driven growth. We find a progressively increasing [OIII]5007/Hβ ratio out to z~6, consistent with the ratios in local galaxies selected by increasing H-alpha EW (i.e., sSFR). This demonstrates the potential of using "local high-z analogs" to investigate the spectroscopic properties and relations of galaxies in the re-ionization epoch.

We use observed optical to near-infrared spectral energy distributions (SEDs) of 266 galaxies in the COSMOS survey to derive the wavelength dependence of the dust attenuation at high redshift. All of the galaxies have spectroscopic redshifts in the range z = 2-6.5. The presence of the C IV absorption feature, indicating that the rest-frame UV-optical SED is dominated by OB stars, is used to select objects for which the intrinsic, unattenuated spectrum has a well-established shape. Comparison of this intrinsic spectrum with the observed broadband photometric SED then permits derivation of the wavelength dependence of the dust attenuation. The derived dust attenuation curve is similar in overall shape to the Calzetti curve for local starburst galaxies. We also see the 2175 � bump feature which is present in the Milky Way and Large Magellanic Cloud extinction curves but not seen in the Calzetti curve. The bump feature is commonly attributed to graphite or polycyclic aromatic hydrocarbons. No significant dependence is seen with redshift between sub-samples at z = 2-4 and z = 4-6.5. The "extinction" curve obtained here provides a firm basis for color and extinction corrections of high redshift galaxy photometry.

We present spectroscopic follow-up observations on two bright Lyα emitter (LAE) candidates originally found by Krug et al. at a redshift of z ~ 7.7 using the Multi-Object Spectrometer for Infra-Red Exploration at Keck. We rule out any line emission at the >5? level for both objects, putting on solid ground a previous null result for one of the objects. The limits inferred from the non-detections rule out the previous claim of no or even reversed evolution between 5.7 < z < 7.7 in the Lyα luminosity function (LF) and suggest a drop in the Lyα LF consistent with that seen in Lyman break galaxy (LBG) samples. We model the redshift evolution of the LAE LF using the LBG UV-continuum LF and the observed rest-frame equivalent width distribution. From the comparison of our empirical model with the observed LAE distribution, we estimate lower limits of the neutral hydrogen fraction to be 50% - 70% at z ~ 7.7. Together with this, we find a strong evolution in the Lyα optical depth characterized by (1 + z)^2.2 ± 0.5 beyond z = 6, indicative of a strong evolution of the intergalactic medium. Finally, we extrapolate the LAE LF to z ~ 9 using our model and show that it is unlikely that large area surveys, like UltraVISTA or Euclid, pick up LAEs at this redshift assuming the current depths and area.

We report the results of deep (4.6 hr) H-band spectroscopy of the well studied z ~ 12 H-band dropout galaxy candidate UDFj-39546284 with MOSFIRE on Keck-I. These data reach a sensitivity of 5-10 * 1e-19 erg s^-1 cm^-2 per 4.4A resolution element between sky lines. Previous papers have argued that this source could either be a large equivalent width line emitting galaxy at 2 < z < 3.5 or a luminous galaxy at z ~ 12. We find a 2.2σ peak associated with a line candidate in deep Hubble Space Telescope Wide-Field Camera 3 Infrared grism observations, but at a lower flux than expected. After considering several possibilities, we conclude these data cannot conclusively confirm or reject the previous line detection, and significantly deeper spectroscopic observations are required. We also search for low-redshift emission lines in 10 other 7 < z < 10 z, Y, and J-dropout candidates in our mask and find no significant detections.

We report the detection of [CII]158μm emission from a system of three closely-separated sources in the COSMOS field at z~4.56, as part of the ALMA Large Program to INvestigate CII at Early times (ALPINE). The two dominant sources are closely associated, both spatially (1.6 arcsec ~ 11kpc) and in velocity (~100km/s), while the third source is slightly more distant (2.8 arcsec ~ 18kpc, ~300km/s). The second strongest source features a slight velocity gradient, while no significant velocity gradient is seen in the other two sources. Using the observed [CII] luminosities, we derive a total log(SFR_[CII]/[M_sol/year]) = 2.8 +/-0.2, which may be split into contributions of 59%, 31%, and 10% from the central, east, and west sources, respectively. Comparison of these [CII] detections to recent zoom-in cosmological simulations suggests an ongoing major merger. We are thus witnessing a system in a major phase of mass build-up by merging, including an on-going major merger and an upcoming minor merger, which is expected to end up in a single massive galaxy by z~2.5.

We report the discovery of a mid-infrared variable AGN which is hosted by an ultraluminous infrared galaxy (ULIRG) candidate in the Sloan Stripe 82 field. WISE J030654.88+010833.6 is a red, extended galaxy, which we estimate to be at a photometric redshift of 0.28 < z < 0.31, based on its optical and near-infrared spectral energy distribution (SED). The factor of two variability over 8 years seen in the WISE 3.4 and 4.6 ?m wavelength channels is not clearly correlated with optical variability in archival data. Based on our estimation of the physical parameters of the host galaxy, J030654.88+010833.6 is possibly a composite AGN/starburst ULIRG in a phase where high star formation ~70 M_sun year-1 is occurring. Our estimate of the black hole mass to stellar mass ratio also appears to be consistent with that of broad-line AGN in the local universe. The long-term variability of J030654.88+010833.6 as seen in the WISE W1 and W2 light curves is likely due to variations in the accretion rate, with the energy being reprocessed by dust in the vicinity of the AGN.

We have obtained three-dimensional maps of the universe in ~200x200x80 comoving Mpc³ (cMpc³) volumes each at z=5.7 and 6.6 based on a spectroscopic sample of 179 galaxies that achieves >80% completeness down to the Lyα luminosity of log(L_Lyα/[erg s^-1]) = 43.0, based on our Keck and Gemini observations and the literature. The maps reveal filamentary large-scale structures and two remarkable overdensities made out of at least 44 and 12 galaxies at z=5.692 (z57OD) and z=6.585 (z66OD), respectively, making z66OD the most distant overdensity spectroscopically confirmed to date with >10 spectroscopically confirmed galaxies. We compare spatial distributions of submillimeter galaxies at z=4-6 with our z=5.7 galaxies forming the large-scale structures, and detect a 99.97% signal of cross correlation, indicative of a clear coincidence of dusty star-forming galaxy and dust unobscured galaxy formation at this early epoch. The galaxies in z57OD and z66OD are actively forming stars with star formation rates (SFRs) >5 times higher than the main sequence, and particularly the SFR density in z57OD is 10 times higher than the cosmic average at the redshift (a.k.a. the Madau-Lilly plot). Comparisons with numerical simulations suggest that z57OD and z66OD are protoclusters that are progenitors of the present-day clusters with halo masses of ~10¹⁴ M_sun.

The advent of new sub-millimeter observational facilities has stimulated the desire to model the sub-mm line emission of galaxies within cosmological galaxy formation models. This is typically done in post-processing by applying sub-resolution recipes to describe the properties of the unresolved interstellar medium. At the same time, while there is freedom in how one implements these sorts of recipes, the impact of various choices has yet to be systematically explored in simulations. In this paper, we do just that. We combine a semi-analytic model of galaxy formation with chemical equilibrium networks and numerical radiative transfer models and explore how different choices for the sub-grid modeling affect the predicted CO, [CI], and [CII] emission of galaxies. We find that a key component for a successful model includes a molecular cloud mass-size relation and scaling for the ultraviolet and cosmic ray radiation field that varies based on local ISM properties. Our most successful model adopts a Plummer profile for the radial density distribution of gas within molecular clouds. On the other hand, different assumptions for the clumping of gas within molecular clouds and changes in the slope of the molecular cloud mass distribution function hardly affect the CO, [CI], and [CII] luminosities of galaxies. At fixed star-formation rate the [CII]-SFR ratio of galaxies scales inversely with the pressure acting on molecular clouds, increasing the molecular clouds density and hence decreasing the importance of [CII] line cooling. Overall we find that it is essential that a wide range of sub-mm emission lines arising in vastly different phases of the ISM are used as model constraints in order to limit the freedom in sub-grid choices and we present the details of the most successful model variant.

We report the serendipitous discovery of a dusty, starbursting galaxy at z = 5.667 (called CRLE hereafter), in close physical association to the 'normal' Main Sequence galaxy HZ10 at z = 5.654. CRLE was identified by detection of [CII], [NII] and CO(2-1) line emission, making it the highest redshift, most luminous starburst in the COSMOS field. This massive, dusty galaxy appears to be forming stars at a rate of at least 1500 M_☉/yr in a compact region only ~3 kpc in diameter. The dynamical and dust emission properties of CRLE suggest an ongoing merger driving the starburst, in a potentially intermediate stage relative to other known dusty galaxies at the same epoch. The ratio of [CII] to [NII] may suggest that an important contribution to the [CII] emission comes from a diffuse ionized gas component, which could be more extended than the dense, starbursting gas. CRLE appears to be located in a significant galaxy overdensity at the same redshift, potentially associated with a large scale cosmic structure recently identified in a Lyman Alpha Emitter survey. This overdensity suggests that CRLE and HZ10 reside in a protocluster environment, offering the tantalizing opportunity to study the effect of a massive starburst on protocluster star formation. Our findings support the interpretation that a significant fraction of the earliest galaxy formation may occur from the inside-out, within the central regions of the most massive halos, while rapidly evolving into the massive galaxy clusters observed in the local Universe.

We investigate emission lines of Lyα, [OIII]5007, Hα, and [CII]158μm from 1125 galaxies at z=4.9-7.0. Our sample is composed of 1098 Lyα emitters (LAEs) at z=4.9, 5.7, 6.6, and 7.0 identified by Subaru/Hyper Suprime-Cam (HSC) narrowband surveys in the 4deg² area covered by Spitzer large area survey with Subaru/HSC (SPLASH) and 29 galaxies at z=5.148-7.508 with deep [CII] ALMA or PdBI data in the literature. Fluxes of strong rest-frame optical emission lines of [OIII] and Hα (Hβ) are constrained by significant excesses found in the SPLASH 3.6 and 4.5μm photometry. At z=4.9, the rest-frame Ha equivalent width, EW⁰_Ha, and the Lyα escape fraction, f_Lyα, positively correlate with the rest-frame Lyα equivalent width, EW_Lyα. The f_Lyα-EW_Lyα correlation is similarly found at z~0-2, suggesting no evolution of the correlation over z~0-5. The typical ionizing photon production efficiency of LAEs is logξ_ion/[Hz erg^-1]~25.5, significantly (60-100%) higher than those of LBGs at a given UV magnitude. At z=5.7-7.0, there exists an interesting turn-over trend that the [OIII]/Hα flux ratio increases in EW_Lyα~0-30A, where it reaches a maximum, and then decreases out to EW_Lyα~130A. We also identify a negative correlation between a [CII] luminosity to star-formation rate ratio (L_[CII]/SFR) and EW_Lyα at the 4.5σ confidence level. We carefully investigate the physical origins of the trend and the correlations with stellar-synthesis and photoionization models covering the vast parameter space of metallicity, ionization parameter, and stellar age, and find that the [OIII]/Ha turn-over trend is explained by a simple anti-correlation between EW_Lya and metallicity, indicative of the detections of the very metal-poor (~0.04Z_sun) galaxies with EW_Lyα~200A. Moreover, this simple anti-correlation explains self-consistently all of the correlations of Lyα, Hα, [OIII]/Hα, and [CII] identified in our study.

We present a catalog of 10718 objects in the COSMOS field observed through multi-slit spectroscopy with the Deep Imaging Multi-Object Spectrograph (DEIMOS) on the Keck II telescope in the wavelength range ~5500-9800Å. The catalog contains 6617 objects with high-quality spectra (two or more spectral features), and 1798 objects with a single spectroscopic feature confirmed by the photometric redshift. For 2024 typically faint objects we could not obtain reliable redshifts. The objects have been selected from a variety of input catalogs based on multi-wavelength observations in the field, and thus have a diverse selection function, which enables the study of the diversity in the galaxy population. The magnitude distribution of our objects is peaked at I_AB ~ 23 and K_AB ~ 21, with a secondary peak at K_AB ~ 24. We sample a broad redshift distribution in the range 0 < z < 6, with one peak at z ~ 1, and another one around z ~ 4. We have identified 13 redshift spikes at z > 0.65 with chance probabilities < 4 x 10^-4, some of which are clearly related to protocluster structures of sizes > 10 Mpc. An object-to-object comparison with a multitude of other spectroscopic samples in the same field shows that our DEIMOS sample is among the best in terms of fraction of spectroscopic failures and relative redshift accuracy. We have determined the fraction of spectroscopic blends to about 0.8% in our sample. This is likely a lower limit and at any rate well below the most pessimistic expectations. Interestingly, we find evidence for strong lensing of Lyα background emitters within the slits of 12 of our target galaxies, increasing their apparent density by about a factor of 4.

We present a multi-wavelength catalog in the Subaru-XMM Deep Field (SXDF) as part of the Spitzer Large Area Survey with Hyper-Suprime-Cam (SPLASH). We include the newly acquired optical data from the Hyper-Suprime Cam Subaru Strategic Program, accompanied by IRAC coverage from the SPLASH survey. All available optical and near-infrared data is homogenized and resampled on a common astrometric reference frame. Source detection is done using a multi-wavelength detection image including the u-band to recover the bluest objects. We measure multi-wavelength photometry and compute photometric redshifts as well as stellar masses for ~800,000 objects over ~4.3 deg². Using the available spectroscopic redshifts from various surveys over the range of 0 < z < 6, we verify the performance of the photometric redshifts and we find a normalized median absolute deviation of 0.023 and outlier fraction of 3.4%. The SPLASH-SXDF catalog is a valuable, publicly available resource that is perfectly suited for studying galaxies in the early universe and tracing their evolution through cosmic time. The catalog is available for download from https://z.umn.edu/SXDF.

Knowledge of the number density of Hα emitting galaxies is vital for assessing the scientific impact of the Euclid and WFIRST missions. In this work we present predictions from a galaxy formation model, Galacticus, for the cumulative number counts of Hα-emitting galaxies. We couple Galacticus to three different dust attenuation methods and examine the counts using each method. A χ² minimisation approach is used to compare the model predictions to observed galaxy counts and calibrate the dust parameters. We find that weak dust attenuation is required for the Galacticus counts to be broadly consistent with the observations, though the optimum dust parameters return large values for χ², suggesting that further calibration of Galacticus is necessary. The model predictions are also consistent with observed estimates for the optical depth and the Hα luminosity function. Finally we present forecasts for the redshift distributions and number counts for two Euclid-like and one WFIRST-like survey. For a Euclid-like survey with redshift range 0.9 < z < 1.8 and Hα+[NII] blended flux limit of 2 x 10^-16erg/s/cm² we predict a number density between 3900-4800 galaxies per square degree. For a WFIRST-like survey with redshift range 1 < z < 2 and blended flux limit of 1 x 10^-16erg/s/cm² we predict a number density between 10400-15200 galaxies per square degree.

We measure the stellar mass function (SMF) of galaxies in the COSMOS field up to z~6. We select them in the near-IR bands of the COSMOS2015 catalogue, which includes ultra-deep photometry from UltraVISTA-DR2, SPLASH, and Subaru/Hyper-SuprimeCam. At z>2.5 we use new precise photometric redshifts with error σ_z=0.03(1+z) and an outlier fraction of 12%, estimated by means of the unique spectroscopic sample of COSMOS. The increased exposure time in the DR2, along with our panchromatic detection strategy, allow us to improve the stellar mass completeness at high z with respect to previous UltraVISTA catalogues. We also identify passive galaxies through a robust colour-colour selection, extending their SMF estimate up to z=4. Our work provides a comprehensive view of galaxy stellar mass assembly between z=0.1 and 6, for the first time using consistent estimates across the entire redshift range. We fit these measurements with a Schechter function, correcting for Eddington bias. We compare the SMF fit with the halo mass function predicted from ΛCDM simulations. We find that at z>3 both functions decline with a similar slope in the high-mass end. This feature could be explained assuming that the mechanisms that quench star formation in massive haloes become less effective at high redshifts; however further work needs to be done to confirm this scenario. Concerning the SMF low-mass end, it shows a progressive steepening as moving towards higher redshifts, with α decreasing from -1.47_{-0.02}^{+0.02} at z~0.1 to -2.11_{-0.13}^{+0.30} at z~5. This slope depends on the characterisation of the observational uncertainties, which is crucial to properly remove the Eddington bias. We show that there is currently no consensus on the method to model such errors: different descriptions result in different best-fit Schechter parameters.

We present, for the first time, the local [CII] 158um emission line luminosity function measured using a sample of more than 500 galaxies from the Revised Bright Galaxy Sample (RBGS). [CII] luminosities are measured from the Herschel PACS observations of the Luminous Infrared Galaxies in the Great Observatories All-sky LIRG Survey (GOALS) and estimated for the rest of the sample based on the far-IR luminosity and color. The sample covers 91.3% of the sky and is complete at S_60μm > 5.24 Jy. We calculated the completeness as a function of [CII] line luminosity and distance, based on the far-IR color and flux densities. The [CII] luminosity function is constrained in the range ~10^(7-9) (Lsun) from both the 1/V_max and a maximum likelihood methods. The shape of our derived [CII] emission line luminosity function agrees well with the IR luminosity function. For the CO(1-0) and [CII] luminosity functions to agree, we propose a varying ratio of [CII]/CO(1-0) as a function of CO luminosity, with larger ratios for fainter CO luminosities. Limited [CII] high redshift observations as well as estimates based on the IR and UV luminosity functions are suggestive of an evolution in the [CII] luminosity function similar to the evolution trend of the cosmic star formation rate density. Deep surveys using ALMA with full capability will be able to confirm this prediction.

We study the effects of local environment and stellar mass on galaxy properties using a mass complete sample of quiescent and star-forming systems in the COSMOS field at z < 3. We show that at z < 1, the median star-formation rate (SFR) and specific SFR (sSFR) of all galaxies depend on environment, but they become independent of environment at z > 1. However, we find that only for star-forming galaxies, the median SFR and sSFR are similar in different environments, regardless of redshift and stellar mass. We find that the quiescent fraction depends on environment at z < 1, and on stellar mass out to z ~ 3. We show that at z < 1, galaxies become quiescent faster in denser environments and that the overall environmental quenching efficiency increases with cosmic time. Environmental and mass quenching processes depend on each other. At z < 1, denser environments more efficiently quench galaxies with higher masses (log(M/M_sol) > 10.7), possibly due to a higher merger rate of massive galaxies in denser environments, and that mass quenching is more efficient in denser regions. We show that the overall mass quenching efficiency (ε_mass) for more massive galaxies (log(M/M_sol) > 10.2) rises with cosmic time until z ~ 1 and flattens out since then. However, for less massive galaxies, the rise in ε_mass continues to the present time. Our results suggest that environmental quenching is only relevant at z < 1, likely a fast process, whereas mass quenching is the dominant mechanism at z > 1, with a possible stellar feedback physics.

We present new Keck/MOSFIRE K-band spectroscopy for a sample of 14 faint, X-ray selected Active Galactic Nuclei (AGNs) in the COSMOS field. The data covers the spectral region surrounding the broad Balmer emission lines, which enables the estimation of black hole masses (M_BH) and accretion rates (in terms of L/L_Edd). We focus on ten AGN at z~3.3, where the we observe the Hbeta spectral region, while for the other four z~2.4 sources we use the Halpha broad emission line. Compared with previous detailed studies of unobscured AGNs at these high redshifts, our sources are fainter by an order of magnitude, corresponding to number densities of order ~10^-6 - 10^-5 Mpc^-3. The lower luminosities also allow for a robust identification of the host galaxies emission, necessary to obtain reliable intrinsic AGN luminosities, BH masses and accretion rates. We find the AGNs in our sample to be powered by SMBHs with a typical mass of M_BH~6*10⁸ M_Sol - significantly lower than the higher-luminosity, rarer quasars reported in previous studies. The accretion rates are in the range of L/L_Edd~0.1-0.5, with an evident lack of lower-L/L_Edd (and higher M_BH) sources, as found in several studies of faint AGNs at intermediate redshifts. Based on the early growth expected for the SMBHs in our sample, we argue that a significant population of faint z~6 AGNs, with M_BH~10⁶ M_Sol, should be detectable in the deepest X-ray surveys available, which is however not observed. We discuss several possible explanations for the apparent absence of such a population, concluding that the most probable scenario involves an evolution in source obscuration and/or radiative efficiencies.

Calibrating the photometric redshifts of >10⁹ galaxies for upcoming weak lensing cosmology experiments is a major challenge for the astrophysics community. The path to obtaining the required spectroscopic redshifts for training and calibration is daunting, given the anticipated depths of the surveys and the difficulty in obtaining secure redshifts for some faint galaxy populations. Here we present an analysis of the problem based on the self-organizing map, a method of mapping the distribution of data in a high-dimensional space and projecting it onto a lower-dimensional representation. We apply this method to existing photometric data from the COSMOS survey selected to approximate the anticipated Euclid weak lensing sample, enabling us to robustly map the empirical distribution of galaxies in the multidimensional color space defined by the expected Euclid filters. Mapping this multicolor distribution lets us determine where - in galaxy color space - redshifts from current spectroscopic surveys exist and where they are systematically missing. Crucially, the method lets us determine whether a spectroscopic training sample is representative of the full photometric space occupied by the galaxies in a survey. We explore optimal sampling techniques and estimate the additional spectroscopy needed to map out the color-redshift relation, finding that sampling the galaxy distribution in color space in a systematic way can efficiently meet the calibration requirements. While the analysis presented here focuses on the Euclid survey, similar analysis can be applied to other surveys facing the same calibration challenge, such as DES, LSST, and WFIRST.

Large-scale structures (LSSs) out to z < 3.0 are measured in the Cosmic Evolution Survey (COSMOS) using extremely accurate photometric redshifts (photoz). The Ks-band-selected sample (from Ultra-Vista) is comprised of 155,954 galaxies. Two techniques - adaptive smoothing and Voronoi tessellation - are used to estimate the environmental densities within 127 redshift slices. Approximately 250 statistically significant overdense structures are identified out to z = 3.0 with shapes varying from elongated filamentary structures to more circularly symmetric concentrations. We also compare the densities derived for COSMOS with those based on semi-analytic predictions for a ΛCDM simulation and find excellent overall agreement between the mean densities as a function of redshift and the range of densities. The galaxy properties (stellar mass, spectral energy distributions (SEDs), and star formation rates (SFRs)) are strongly correlated with environmental density and redshift, particularly at z < 1.0-1.2. Classifying the spectral type of each galaxy using the rest-frame b - i color (from the photoz SED fitting), we find a strong correlation of early-type galaxies (E-Sa) with high-density environments, while the degree of environmental segregation varies systematically with redshift out to z ~ 1.3. In the highest density regions, 80% of the galaxies are early types at z = 0.2 compared to only 20% at z = 1.5. The SFRs and the star formation timescales exhibit clear environmental correlations. At z > 0.8, the SFR density is uniformly distributed over all environmental density percentiles, while at lower redshifts the dominant contribution is shifted to galaxies in lower density environments.