Dan L. Starr

The Palomar Transient Factory: System Overview, Performance, and First Results

The Palomar Transient Factory (PTF) is a fully-automated, wide-field survey aimed at a systematic exploration of the optical transient sky. The transient survey is performed using a new 8.1 square degree camera installed on the 48 inch Samuel Oschin telescope at Palomar Observatory; colors and light curves for detected transients are obtained with the automated Palomar 60 inch telescope. PTF uses 80% of the 1.2 m and 50% of the 1.5 m telescope time. With an exposure of 60 s the survey reaches a depth of m_(g′) ≈ 21.3 and m_R ≈ 20.6 (5σ, median seeing). Four major experiments are planned for the five-year project: (1) a 5 day cadence supernova search; (2) a rapid transient search with cadences between 90 s and 1 day; (3) a search for eclipsing binaries and transiting planets in Orion; and (4) a 3π sr deep H-alpha survey. PTF provides automatic, real-time transient classification and follow-up, as well as a database including every source detected in each frame. This paper summarizes the PTF project, including several months of on-sky performance tests of the new survey camera, the observing plans, and the data reduction strategy. We conclude by detailing the first 51 PTF optical transient detections, found in commissioning data.

FROM SHOCK BREAKOUT TO PEAK AND BEYOND: EXTENSIVE PANCHROMATIC OBSERVATIONS OF THE TYPE Ib SUPERNOVA 2008D ASSOCIATED WITH SWIFT X-RAY TRANSIENT 080109

We present extensive early photometric (ultraviolet through near-infrared) and spectroscopic (optical and near-infrared) data on supernova (SN) 2008D as well as X-ray data analysis on the associated Swift X-ray transient (XRT) 080109. Our data span a time range of 5 hr before the detection of the X-ray transient to 150days after its detection, and a detailed analysis allowed us to derive constraints on the nature of the SN and its progenitor; throughout we draw comparisons with results presented in the literature and find several key aspects that differ. We show that the X-ray spectrum of XRT 080109 can be fit equally well by an absorbed power law or a superposition of about equal parts of both power law and blackbody. Our data first established that SN 2008D is a spectroscopically normal SN Ib (i.e., showing conspicuous He lines) and showed that SN 2008D had a relatively long rise time of 18days and a modest optical peak luminosity. The early-time light curves of the SN are dominated by a cooling stellar envelope (for Δt0.1-4days, most pronounced in the blue bands) followed by 56Ni decay. We construct a reliable measurement of the bolometric output for this stripped-envelope SN, and, combined with estimates of E K and M ej from the literature, estimate the stellar radius R ⊙ of its probable Wolf-Rayet progenitor. According to the model of Waxman etal. and Chevalier & Fransson, we derive R W07⊙ = 1.2 0.7R ⊙ and R CF08⊙ = 12 7 R ⊙, respectively; the latter being more in line with typical WN stars. Spectra obtained at three and four months after maximum light show double-peaked oxygen lines that we associate with departures from spherical symmetry, as has been suggested for the inner ejecta of a number of SN Ib cores.

ON MACHINE-LEARNED CLASSIFICATION OF VARIABLE STARS WITH SPARSE AND NOISY TIME-SERIES DATA

With the coming data deluge from synoptic surveys, there is a need for frameworks that can quickly and automatically produce calibrated classification probabilities for newly observed variables based on small numbers of time-series measurements. In this paper, we introduce a methodology for variable-star classification, drawing from modern machine-learning techniques. We describe how to homogenize the information gleaned from light curves by selection and computation of real-numbered metrics (features), detail methods to robustly estimate periodic features, introduce tree-ensemble methods for accurate variable-star classification, and show how to rigorously evaluate a classifier using cross validation. On a 25-class data set of 1542 well-studied variable stars, we achieve a 22.8% error rate using the random forest (RF) classifier; this represents a 24% improvement over the best previous classifier on these data. This methodology is effective for identifying samples of specific science classes: for pulsational variables used in Milky Way tomography we obtain a discovery efficiency of 98.2% and for eclipsing systems we find an efficiency of 99.1%, both at 95% purity. The RF classifier is superior to other methods in terms of accuracy, speed, and relative immunity to irrelevant features; the RF can also be used to estimate the importance of each feature in classification. Additionally, we present the first astronomical use of hierarchical classification methods to incorporate a known class taxonomy in the classifier, which reduces the catastrophic error rate from 8% to 7.8%. Excluding low-amplitude sources, the overall error rate improves to 14%, with a catastrophic error rate of 3.5%.

Type Ia Supernovae Are Good Standard Candles in the Near Infrared: Evidence from PAIRITEL

We have obtained 1087 NIR (JHKs) measurements of 21 SNe Ia using PAIRITEL, nearly doubling the number of well-sampled NIR SN Ia light curves. These data strengthen the evidence that SNe Ia are excellent standard candles in the NIR, even without correction for optical light-curve shape. We construct fiducial NIR templates for normal SNe Ia from our sample, excluding only the three known peculiar SNe Ia: SN 2005bl, SN 2005hk, and SN 2005ke. The H-band absolute magnitudes in this sample of 18 SNe Ia have an intrinsic rms of only 0.15 mag with no correction for light-curve shape. We found a relationship between the H-band extinction and optical color excess of AH = 0.2E(B − V) . This variation is as small as the scatter in distance modulus measurements currently used for cosmology based on optical light curves after corrections for light-curve shape. Combining the homogeneous PAIRITEL measurements with 23 SNe Ia from the literature, these 41 SNe Ia have standard H-band magnitudes with an rms scatter of 0.16 mag. The good match of our sample with the literature sample suggests there are few systematic problems with the photometry. We present a nearby NIR Hubble diagram that shows no correlation of the residuals from the Hubble line with light-curve properties. Future samples that account for optical and NIR light-curve shapes, absorption, spectroscopic variation, or host-galaxy properties may reveal effective ways to improve the use of SNe Ia as distance indicators. Since systematic errors due to dust absorption in optical bands remain the leading difficulty in the cosmological use of supernovae, the good behavior of SN Ia NIR light curves and their relative insensitivity to reddening make these objects attractive candidates for future cosmological work.

The troublesome broadband evolution of GRB 061126: Does a gray burst imply gray dust?

We report on observations of a gamma-ray burst (GRB 061126) with an extremely bright (R ≈ 12 mag at peak) early-time optical afterglow. The optical afterglow is already fading as a power law 22 s after the trigger, with no detectable prompt contribution in our first exposure, which was coincident with a large prompt-emission gamma-ray pulse. The optical-infrared photometric SED is an excellent fit to a power law, but it exhibits a moderate red-to-blue evolution in the spectral index at about 500 s after the burst. This color change is contemporaneous with a switch from a relatively fast decay to slower decay. The rapidly decaying early afterglow is broadly consistent with synchrotron emission from a reverse shock, but a bright forward-shock component predicted by the intermediate- to late-time X-ray observations under the assumptions of standard afterglow models is not observed. Indeed, despite its remarkable early-time brightness, this burst would qualify as a dark burst at later times on the basis of its nearly flat optical-to-X-ray spectral index. Our photometric SED provides no evidence of host galaxy extinction, requiring either large quantities of gray dust in the host system (at redshift 1.1588 ± 0.0006, based on our late-time Keck spectroscopy) or separate physical origins for the X-ray and optical afterglows.

Automating Discovery and Classification of Transients and Variable Stars in the Synoptic Survey Era

The rate of image acquisition in modern synoptic imaging surveys has already begun to outpace the feasibility of keeping astronomers in the real-time discovery and classification loop. Here we present the inner workings of a framework, based on machine-learning algorithms, that captures expert training and ground-truth knowledge about the variable and transient sky to automate (1) the process of discovery on image differences, and (2) the generation of preliminary science-type classifications of discovered sources. Since follow-up resources for extracting novel science from fast-changing transients are precious, self-calibrating classification probabilities must be couched in terms of efficiencies for discovery and purity of the samples generated. We estimate the purity and efficiency in identifying real sources with a two-epoch image-difference discovery algorithm for the Palomar Transient Factory (PTF) survey. Once given a source discovery, using machine-learned classification trained on PTF data, we distinguish between transients and variable stars with a 3.8% overall error rate (with 1.7% errors for imaging within the Sloan Digital Sky Survey footprint). At >96% classification efficiency, the samples achieve 90% purity. Initial classifications are shown to rely primarily on context-based features, determined from the data itself and external archival databases. In the first year of autonomous operations of PTF, this discovery and classification framework led to several significant science results, from outbursting young stars to subluminous Type IIP supernovae to candidate tidal disruption events. We discuss future directions of this approach, including the possible roles of crowdsourcing and the scalability of machine learning to future surveys such as the Large Synoptic Survey Telescope (LSST).

A Putative Early-Type Host Galaxy for GRB 060502B: Implications for the Progenitors of Short-Duration Hard-Spectrum Bursts

Starting with the first detection of an afterglow from a short-duration hard-spectrum γ-ray burst (SHB) by Swift last year, a growing body of evidence has suggested that SHBs are associated with an older and lower redshift galactic population than long-soft GRBs and, in a few cases, with large (≳10 kpc) projected offsets from the centers of their putative host galaxies. Here we present observations of the field of ORB 060502B, a SHB detected by Swift and localized by the X-Ray Telescope (XRT). We find a massive red galaxy at a redshift of z = 0.287 at an angular distance of 17.1″ from our revised XRT position. Using associative and probabilistic arguments, we suggest that this galaxy hosted the progenitor of GRB 060502B. If true, this offset would correspond to a physical displacement of 73 ± 19 kpc in projection , about twice the largest offset inferred for any SHB to date and almost an order of magnitude larger than a typical long-soft burst offset. Spectra and modeling of the star formation history of this possible host show it to have undergone a large ancient starburst. If the progenitor of GRB 060502B was formed in this starburst episode, the time of the GRB explosion since birth is τ ≈ 1.3 ± 0.2 Gyr and the minimum kick velocity of the SHB progenitor is vkick,min = 55 ± 15 km s-1.

GJ 3236: A NEW BRIGHT, VERY LOW MASS ECLIPSING BINARY SYSTEM DISCOVERED BY THE MEARTH OBSERVATORY

We report the detection of eclipses in GJ 3236, a bright (I = 11.6), very low mass binary system with an orbital period of 0.77 days. Analysis of light and radial velocity curves of the system yielded component masses of 0.38 {+-} 0.02 M{sub sun} and 0.28 {+-} 0.02 M{sub sun}. The central values for the stellar radii are larger than the theoretical models predict for these masses, in agreement with the results for existing eclipsing binaries, although the present 5% observational uncertainties limit the significance of the larger radii to approximately 1{sigma}. Degeneracies in the light curve models resulting from the unknown configuration of surface spots on the components of GJ 3236 currently dominate the uncertainties in the radii, and could be reduced by obtaining precise, multiband photometry covering the full orbital period. The system appears to be tidally synchronized and shows signs of high activity levels as expected for such a short orbital period, evidenced by strong H{alpha} emission lines in the spectra of both components. These observations probe an important region of mass-radius parameter space around the predicted transition to fully convective stellar interiors, where there are a limited number of precise measurements available in the literature.

Construction of a Calibrated Probabilistic Classification Catalog: Application to 50k Variable Sources in the All-Sky Automated Survey

With growing data volumes from synoptic surveys, astronomers necessarily must become more abstracted from the discovery and introspection processes. Given the scarcity of follow-up resources, there is a particularly sharp onus on the frameworks that replace these human roles to provide accurate and well-calibrated probabilistic classification catalogs. Such catalogs inform the subsequent follow-up, allowing consumers to optimize the selection of specific sources for further study and permitting rigorous treatment of classification purities and efficiencies for population studies. Here, we describe a process to produce a probabilistic classification catalog of variability with machine learning from a multi-epoch photometric survey. In addition to producing accurate classifications, we show how to estimate calibrated class probabilities and motivate the importance of probability calibration. We also introduce a methodology for feature-based anomaly detection, which allows discovery of objects in the survey that do not fit within the predefined class taxonomy. Finally, we apply these methods to sources observed by the All-Sky Automated Survey (ASAS), and release the Machine-learned ASAS Classification Catalog (MACC), a 28 class probabilistic classification catalog of 50,124 ASAS sources in the ASAS Catalog of Variable Stars. We estimate that MACC achieves a sub-20% classification error rate and demonstrate that the class posterior probabilities are reasonably calibrated. MACC classifications compare favorably to the classifications of several previous domain-specific ASAS papers and to the ASAS Catalog of Variable Stars, which had classified only 24% of those sources into one of 12 science classes.

ACTIVE LEARNING TO OVERCOME SAMPLE SELECTION BIAS: APPLICATION TO PHOTOMETRIC VARIABLE STAR CLASSIFICATION

Despite the great promise of machine-learning algorithms to classify and predict astrophysical parameters for the vast numbers of astrophysical sources and transients observed in large-scale surveys, the peculiarities of the training data often manifest as strongly biased predictions on the data of interest. Typically, training sets are derived from historical surveys of brighter, more nearby objects than those from more extensive, deeper surveys (testing data). This sample selection bias can cause catastrophic errors in predictions on the testing data because (1) standard assumptions for machine-learned model selection procedures break down and (2) dense regions of testing space might be completely devoid of training data. We explore possible remedies to sample selection bias, including importance weighting, co-training, and active learning (AL). We argue that AL—where the data whose inclusion in the training set would most improve predictions on the testing set are queried for manual follow-up—is an effective approach and is appropriate for many astronomical applications. For a variable star classification problem on a well-studied set of stars from Hipparcos and Optical Gravitational Lensing Experiment, AL is the optimal method in terms of error rate on the testing data, beating the off-the-shelf classifier by 3.4% and the other proposed methods by at least 3.0%. To aid with manual labeling of variable stars, we developed a Web interface which allows for easy light curve visualization and querying of external databases. Finally, we apply AL to classify variable stars in the All Sky Automated Survey, finding dramatic improvement in our agreement with the ASAS Catalog of Variable Stars, from 65.5% to 79.5%, and a significant increase in the classifiers average confidence for the testing set, from 14.6% to 42.9%, after a few AL iterations.