Search

This project is centered around building a neural network that is able to recognize ASL letters in images, particularly within the scope of a live video feed. Initial testing results came up short of expectations when both the convolutional network and VGG16 transfer learning approaches failed to generalize in settings of different backgrounds. The use of a pre-trained hand joint detection model was then adopted with the produced joint locations being fed into a fully-connected neural network. The results of this approach exceeded those of prior methods and generalized well to a live video feed application., Comment: 10 pages, 10 figures

The light from a source at a distance d will arrive at detectors separated by 100 AU at times that differ by as much as 120 (d/100 Mpc)^{-1} nanoseconds because of the curvature of the wavefront. At gigahertz frequencies, the arrival time difference can be determined to better than a nanosecond with interferometry. If the space-time positions of the detectors are known to a few centimeters, comparable to the accuracy to which very long baseline interferometry baselines and global navigation satellite systems (GNSS) geolocations are constrained, nanosecond timing would allow competitive cosmological constraints. We show that a four-detector constellation at Solar radii of >10 AU could measure distances to individual sources with sub-percent precision and, hence, cosmological parameters such as the Hubble constant to this precision. The precision increases quadratically with baseline length. FRBs are the only known bright extragalactic radio source that are sufficiently point-like. Galactic scattering limits the timing precision at <3 GHz, whereas at higher frequencies the precision is set by removing dispersion. Furthermore, for baselines greater than 100 AU, Shapiro time delays limit the precision, but their effect can be cleaned with two additional detectors. Accelerations that result in ~1 cm uncertainty in detector positions (from variations in the Sun's irradiance, dust collisions and gaseous drag) could be corrected for with weekly GNSS-like trilaterations. Gravitational accelerations from asteroids occur over longer timescales, and so a setup with a precise accelerometer and calibrating the detector positions off of distant FRBs may also be sufficient. The proposed interferometer would also resolve the radio emission region of Galactic pulsars, constrain the mass distribution in the outer Solar System, and reach interesting sensitivities to ~0.01-100 micro-Hz gravitational waves., Comment: 34 pages in preprint format; 3 figures; accepted to ApJ Letters

We present a scalable, cloud-based science platform solution designed to enable next-to-the-data analyses of terabyte-scale astronomical tabular datasets. The presented platform is built on Amazon Web Services (over Kubernetes and S3 abstraction layers), utilizes Apache Spark and the Astronomy eXtensions for Spark for parallel data analysis and manipulation, and provides the familiar JupyterHub web-accessible front-end for user access. We outline the architecture of the analysis platform, provide implementation details, rationale for (and against) technology choices, verify scalability through strong and weak scaling tests, and demonstrate usability through an example science analysis of data from the Zwicky Transient Facility's 1Bn+ light-curve catalog. Furthermore, we show how this system enables an end-user to iteratively build analyses (in Python) that transparently scale processing with no need for end-user interaction. The system is designed to be deployable by astronomers with moderate cloud engineering knowledge, or (ideally) IT groups. Over the past three years, it has been utilized to build science platforms for the DiRAC Institute, the ZTF partnership, the LSST Solar System Science Collaboration, the LSST Interdisciplinary Network for Collaboration and Computing, as well as for numerous short-term events (with over 100 simultaneous users). A live demo instance, the deployment scripts, source code, and cost calculators are accessible at http://hub.astronomycommons.org/., Comment: 23 pages, 7 figures, accepted for publication in the Astronomy Journal

The formation of the stellar mass within galaxy cluster cores is a poorly understood process. It features the complicated physics of cooling flows, AGN feedback, star formation and more. Here, we study the growth of the stellar mass in the vicinity of the Brightest Cluster Galaxy (BCG) in a z = 1.7 cluster, SpARCS1049+56. We synthesize a reanalysis of existing HST imaging, a previously published measurement of the star formation rate, and the results of new radio molecular gas spectroscopy. These analyses represent the past, present and future star formation respectively within this system. We show that a large amount of stellar mass -- between $(2.2 \pm 0.5) \times 10^{10} \: M_\odot$ and $(6.6 \pm 1.2) \times 10^{10}\: M_\odot$ depending on the data processing -- exists in a long and clumpy tail-like structure that lies roughly 12 kpc off the BCG. Spatially coincident with this stellar mass is a similarly massive reservoir ($(1.0 \pm 0.7) \times 10^{11} \: M_\odot$) of molecular gas that we suggest is the fuel for the immense star formation rate of $860 \pm 130 \: M_\odot$/yr, as measured by infrared observations. Hlavacek-Larrondo et al. 2021 surmised that massive, runaway cooling of the hot intracluster X-ray gas was feeding this star formation, a process that had not been observed before at high-redshift. We conclude, based on the amount of fuel and current stars, that this event may be rare in the lifetime of a cluster, producing roughly 15 to 21% of the Intracluster Light (ICL) mass in one go, though perhaps a common event for all galaxy clusters., Comment: 11 pages, 5 figures. Submitted to ApJ. Addressed referee report

We present a novel method to produce empirical generative models of all kinds of astronomical transients from datasets of unlabeled light curves. Our hybrid model, that we call ParSNIP, uses a neural network to model the unknown intrinsic diversity of different transients and an explicit physics-based model of how light from the transient propagates through the universe and is observed. The ParSNIP model predicts the time-varying spectra of transients despite only being trained on photometric observations. With a three-dimensional intrinsic model, we are able to fit out-of-sample multiband light curves of many different kinds of transients with model uncertainties of 0.04-0.06 mag. The representation learned by the ParSNIP model is invariant to redshift, so it can be used to perform photometric classification of transients even with heavily biased training sets. Our classification techniques significantly outperform state-of-the-art methods on both simulated (PLAsTiCC) and real (PS1) datasets with 2.3$\times$ and 2$\times$ less contamination respectively for classification of Type~Ia supernovae. We demonstrate how our model can identify previously-unobserved kinds of transients and produce a sample that is 90% pure. The ParSNIP model can also estimate distances to Type Ia supernovae in the PS1 dataset with an RMS of 0.150 $\pm$ 0.007 mag compared to 0.155 $\pm$ 0.008 mag for the SALT2 model on the same sample. We discuss how our model could be used to produce distance estimates for supernova cosmology without the need for explicit classification., Comment: 27 pages, 17 figures, accepted for publication in AJ

The See Change survey was designed to make $z>1$ cosmological measurements by efficiently discovering high-redshift Type Ia supernovae (SNe Ia) and improving cluster mass measurements through weak lensing. This survey observed twelve galaxy clusters with the Hubble Space Telescope spanning the redshift range $z=1.13$ to $1.75$, discovering 57 likely transients and 27 likely SNe Ia at $z\sim 0.8-2.3$. As in similar previous surveys (Dawson et al. 2009), this proved to be a highly efficient use of HST for SN observations; the See Change survey additionally tested the feasibility of maintaining, or further increasing, the efficiency at yet higher redshifts, where we have less detailed information on the expected cluster masses and star-formation rates. We find that the resulting number of SNe Ia per orbit is a factor of $\sim 8$ higher than for a field search, and 45% of our orbits contained an active SN Ia within 22 rest-frame days of peak, with one of the clusters by itself yielding 6 of the SNe Ia. We present the survey design, pipeline, and SN discoveries. Novel features include fully blinded SN searches, the first random forest candidate classifier for undersampled IR data (with a 50% detection threshold within 0.05 magnitudes of human searchers), real-time forward-modeling photometry of candidates, and semi-automated photometric classifications and follow-up forecasts. We also describe the spectroscopic follow-up, instrumental in measuring host-galaxy redshifts. The cosmology analysis of our sample will be presented in a companion paper., Comment: ApJ preprint

In preparation for photometric classification of transients from the Legacy Survey of Space and Time (LSST) we run tests with different training data sets. Using estimates of the depth to which the 4-metre Multi-Object Spectroscopic Telescope (4MOST) Time Domain Extragalactic Survey (TiDES) can classify transients, we simulate a magnitude-limited sample reaching $r_{\textrm{AB}} \approx$ 22.5 mag. We run our simulations with the software snmachine, a photometric classification pipeline using machine learning. The machine-learning algorithms struggle to classify supernovae when the training sample is magnitude-limited, in contrast to representative training samples. Classification performance noticeably improves when we combine the magnitude-limited training sample with a simulated realistic sample of faint, high-redshift supernovae observed from larger spectroscopic facilities; the algorithms' range of average area under ROC curve (AUC) scores over 10 runs increases from 0.547-0.628 to 0.946-0.969 and purity of the classified sample reaches 95 per cent in all runs for 2 of the 4 algorithms. By creating new, artificial light curves using the augmentation software avocado, we achieve a purity in our classified sample of 95 per cent in all 10 runs performed for all machine-learning algorithms considered. We also reach a highest average AUC score of 0.986 with the artificial neural network algorithm. Having `true' faint supernovae to complement our magnitude-limited sample is a crucial requirement in optimisation of a 4MOST spectroscopic sample. However, our results are a proof of concept that augmentation is also necessary to achieve the best classification results., Comment: MNRAS accepted version Aug 2021

The morphology-density relationship states that dense cosmic environments such as galaxy clusters have an overabundance of quiescent elliptical galaxies, but it is unclear at which redshift this relationship is first established. We study the morphology of 4 clusters with $1.2

Upcoming astronomical surveys such as the Large Synoptic Survey Telescope (LSST) will rely on photometric classification to identify the majority of the transients and variables that they discover. We present a set of techniques for photometric classification that can be applied even when the training set of spectroscopically-confirmed objects is heavily biased towards bright, low-redshift objects. Using Gaussian process regression to model arbitrary light curves in all bands simultaneously, we "augment" the training set by generating new versions of the original light curves covering a range of redshifts and observing conditions. We train a boosted decision tree classifier on features extracted from the augmented light curves, and we show how such a classifier can be designed to produce classifications that are independent of the redshift distributions of objects in the training sample. Our classification algorithm was the best-performing among the 1,094 models considered in the blinded phase of the Photometric LSST Astronomical Time-Series Classification Challenge (PLAsTiCC), scoring 0.468 on the organizers' logarithmic-loss metric with flat weights for all object classes in the training set, and achieving an AUC of 0.957 for classification of Type Ia supernovae. Our results suggest that spectroscopic campaigns used for training photometric classifiers should focus on typing large numbers of well-observed, intermediate redshift transients instead of attempting to type a sample of transients that is directly representative of the full dataset being classified. All of the algorithms described in this paper are implemented in the avocado software package., Comment: 24 pages, 9 figures

We present measurements of the stellar mass fractions ($f_\star$) for a sample of high-redshift ($0.93 \le z \le 1.32$) infrared-selected galaxy clusters from the Massive and Distant Clusters of WISE Survey (MaDCoWS) and compare them to the stellar mass fractions of Sunyaev-Zel'dovich (SZ) effect-selected clusters in a similar mass and redshift range from the South Pole Telescope (SPT)-SZ Survey. We do not find a significant difference in mean $f_\star$ between the two selection methods, though we do find an unexpectedly large range in $f_\star$ for the SZ-selected clusters. In addition, we measure the luminosity function of the MaDCoWS clusters and find $m^*= 19.41\pm0.07$, similar to other studies of clusters at or near our redshift range. Finally, we present SZ detections and masses for seven MaDCoWS clusters and new spectroscopic redshifts for five MaDCoWS clusters. One of these new clusters, MOO J1521+0452 at $z=1.31$, is the most distant MaDCoWS cluster confirmed to date., Comment: Accepted to ApJ

We present the discovery and measurements of a gravitationally lensed supernova (SN) behind the galaxy cluster MOO J1014+0038. Based on multi-band Hubble Space Telescope and Very Large Telescope (VLT) photometry of the supernova, and VLT spectroscopy of the host galaxy, we find a 97.5% probability that this SN is a SN Ia, and a 2.5% chance of a CC SN. Our typing algorithm combines the shape and color of the light curve with the expected rates of each SN type in the host galaxy. With a redshift of 2.2216, this is the highest redshift SN Ia discovered with a spectroscopic host-galaxy redshift. A further distinguishing feature is that the lensing cluster, at redshift 1.23, is the most distant to date to have an amplified SN. The SN lies in the middle of the color and light-curve shape distributions found at lower redshift, disfavoring strong evolution to z = 2.22. We estimate an amplification due to gravitational lensing of 2.8+0.6-0.5 (1.10 +- 0.23 mag)---compatible with the value estimated from the weak-lensing-derived mass and the mass-concentration relation from LambdaCDM simulations---making it the most amplified SN Ia discovered behind a galaxy cluster., Comment: Accepted for publication in ApJ

Hot, Dust-Obscured Galaxies, or "Hot DOGs", are a rare, dusty, hyperluminous galaxy population discovered by the WISE mission. Predominantly at redshifts 2-3, they include the most luminous known galaxies in the universe. Their high luminosities likely come from accretion onto highly obscured super massive black holes (SMBHs). We have conducted a pilot survey to measure the SMBH masses of five z~2 Hot DOGs via broad H_alpha emission lines, using Keck/MOSFIRE and Gemini/FLAMINGOS-2. We detect broad H_alpha emission in all five Hot DOGs. We find substantial corresponding SMBH masses for these Hot DOGs (~ 10^{9} M_sun), and their derived Eddington ratios are close to unity. These z~2 Hot DOGs are the most luminous AGNs at given BH masses, suggesting they are accreting at the maximum rates for their BHs. A similar property is found for known z~6 quasars. Our results are consistent with scenarios in which Hot DOGs represent a transitional, high-accretion phase between obscured and unobscured quasars. Hot DOGs may mark a special evolutionary stage before the red quasar and optical quasar phases, and they may be present at other cosmic epochs., Comment: 15 pages, 9 figures. Accepted by ApJ

Hamamatsu Photonics introduced a new generation of their Multi-Pixel Photon Counters in 2013 with significantly reduced after-pulsing rate. In this paper, we investigate the causes of after-pulsing by testing pre-2013 and post-2013 devices using laser light ranging from 405 to 820nm. Doing so we investigate the possibility that afterpulsing is also due to optical photons produced in the avalanche rather than to impurities trapping charged carriers produced in the avalanches and releasing them at a later time. For pre-2013 devices, we observe avalanches delayed by ns to several 100~ns at 637, 777nm and 820 nm demonstrating that holes created in the zero field region of the silicon bulk can diffuse back to the high field region triggering delayed avalanches. On the other hand post-2013 exhibit no delayed avalanches beyond 100~ns at 777nm. We also confirm that post-2013 devices exhibit about 25 times lower after-pulsing. Taken together, our measurements show that the absorption of photons from the avalanche in the bulk of the silicon and the subsequent hole diffusion back to the junction was a significant source of after-pulse for the pre-2013 devices. Hamamatsu appears to have fixed this problem in 2013 following the preliminary release of our results. We also show that even at short wavelength the timing distribution exhibit tails in the sub-nanosecond range that may impair the MPPC timing performances., Comment: Submitted to JINST, 14 pages, 16 figures

We present confirmation of the cluster MOO J1142+1527, a massive galaxy cluster discovered as part of the Massive and Distant Clusters of WISE Survey. The cluster is confirmed to lie at $z=1.19$, and using the Combined Array for Research in Millimeter-wave Astronomy we robustly detect the Sunyaev-Zel'dovich (SZ) decrement at 13.2$\sigma$. The SZ data imply a mass of $\mathrm{M}_{200m}=(1.1\pm0.2)\times10^{15}$ $\mathrm{M}_\odot$, making MOO J1142+1527 the most massive galaxy cluster known at $z>1.15$ and the second most massive cluster known at $z>1$. For a standard $\Lambda$CDM cosmology it is further expected to be one of the $\sim 5$ most massive clusters expected to exist at $z\ge1.19$ over the entire sky. Our ongoing Spitzer program targeting $\sim1750$ additional candidate clusters will identify comparably rich galaxy clusters over the full extragalactic sky., Comment: Submitted to ApJ Letters, 6 pages, 4 figures

While recent supernova cosmology research has benefited from improved measurements, current analysis approaches are not statistically optimal and will prove insufficient for future surveys. This paper discusses the limitations of current supernova cosmological analyses in treating outliers, selection effects, shape- and color-standardization relations, unexplained dispersion, and heterogeneous observations. We present a new Bayesian framework, called UNITY (Unified Nonlinear Inference for Type-Ia cosmologY), that incorporates significant improvements in our ability to confront these effects. We apply the framework to real supernova observations and demonstrate smaller statistical and systematic uncertainties. We verify earlier results that SNe Ia require nonlinear shape and color standardizations, but we now include these nonlinear relations in a statistically well-justified way. This analysis was primarily performed blinded, in that the basic framework was first validated on simulated data before transitioning to real data. We also discuss possible extensions of the method., Comment: Minor fix in PGM