Search for: All records

Abstract We present the Online Ranked Astrophysical CLass Estimator (ORACLE), the first hierarchical deep-learning model for real-time, context-aware classification of transient and variable astrophysical phenomena. ORACLE is a recurrent neural network with gated recurrent units, and has been trained using a custom hierarchical cross-entropy loss function to provide high-confidence classifications along an observationally driven taxonomy with as little as a single photometric observation. Contextual information for each object, including host galaxy photometric redshift, offset, ellipticity, and brightness, is concatenated to the light-curve embedding and used to make a final prediction. Training on ∼0.5M events from the Extended LSST Astronomical Time-series Classification Challenge, we achieve a top-level (transient versus variable) macroaveraged precision of 0.96 using only 1 day of photometric observations after the first detection in addition to contextual information, for each event; this increases to >0.99 once 64 days of the light curve has been obtained, and 0.83 at 1024 days after first detection for 19-way classification (including supernova subtypes, active galactic nuclei, variable stars, microlensing events, and kilonovae). We also compare ORACLE with other state-of-the-art classifiers and report comparable performance for the 19-way classification task, in addition to delivering accurate top-level classifications much earlier. The code and model weights used in this work are publicly available at our associated GitHub repository (https://github.com/uiucsn/Astro-ORACLE). 

Abstract We presentFrankenBlast, a customized and improved version of theBlastweb application.FrankenBlastassociates transients to their host galaxies, performs host photometry, and runs a innovative spectral energy distribution fitting code to constrain host stellar population properties—all within minutes per object. We testFrankenBlaston 14,432 supernovae (SNe), ≈half of which are spectroscopically classified, and are able to constrain host properties for 9262 events. When contrasting the host stellar masses (M_*), specific star formation rates (sSFR), and host dust extinction (A_V) between spectroscopically and photometrically classified SNe Ia, Ib/c, II, and IIn, we determine that deviations in these distributions are primarily due to misclassified events contaminating the photometrically classified sample. We further show that the higher redshifts of the photometrically classified sample also force theirM_*and sSFR distributions to deviate from those of the spectroscopically classified sample, as these properties are redshift-dependent. We compare host properties between spectroscopically classified SN populations and determine if they primarily traceM_*or SFR. We find that all SN populations seem to both depend onM_*and SFR, with SNe II and IIn somewhat more SFR-dependent than SNe Ia and Ib/c, and SNe Ia moreM_*-dependent than all other classes. We find the difference in the SNe Ib/c and II hosts to be the most intriguing and speculate that SNe Ib/c must be more dependent on higherM_*and more evolved environments for the right conditions for progenitor formation. All data products andFrankenBlastare publicly available, along with a developingFrankenBlastversion intended for Rubin Observatory science products. 

Abstract With the advent of the Vera C. Rubin Observatory, the discovery rate of supernovae (SNe) will surpass the rate of SNe with real time spectroscopic follow-up by 3 orders of magnitude. Accurate photometric classifiers are essential to both select interesting events for follow-up in real time and for archival population-level studies. In this work, we investigate the impact of observable host-galaxy information on the classification of SNe, both with and without additional light-curve and redshift information. We find that host-galaxy information alone can successfully isolate relatively pure (>90%) samples of Type Ia SNe with or without redshift information. With redshift information, we can additionally produce somewhat pure (>70%) samples of Type II SNe and superluminous SNe. Additionally with redshift information, host-galaxy properties do not significantly improve the accuracy of SN classification when paired with complete light curves. In the absence of redshift information, however, galaxy properties significantly increase the accuracy of photometric classification. As a part of this analysis, we present the first formal application of a new objective function, the weighted hierarchical cross entropy, to the problem of SN classification. This objective function more naturally accounts for the hierarchical nature of SN classes and, more broadly, transients. Finally, we present a new set of SN classifications for the Pan-STARRS Medium Deep Survey of SNe that lack spectroscopic redshift, increasing the full photometric sample to >4400 events. 

Abstract We present extensive ultraviolet, optical, and near-infrared (NIR) photometric and spectroscopic observations of the nearby hydrogen-poor superluminous supernova (SLSN-I) SN 2024rmj atz= 0.1189. SN 2024rmj reached a peak absolute magnitude ofM_g ≈ −21.9, placing it at the luminous end of the SLSN-I distribution. The light curve exhibits a pronounced prepeak bump (≈60 days before the main peak) and a postpeak bump (≈55 days after the main peak). The bulk of the light curve is otherwise well fit by a magnetar spin-down model, with typical values (spin: ≈2.1 ms; magnetic field: ≈6 × 10¹³G; ejecta mass: ≈12M_⊙). The optical spectra exhibit characteristic SLSN-I features and evolution, but with a relatively high velocity of ≈8000 km s⁻¹postpeak. Most significantly, we find a clear detection of helium in the NIR spectra at Heiλ1.083μm andλ2.058μm, blueshifted by ≈15,000 km s⁻¹(13 days before peak) and ≈13,000 km s⁻¹(40 days after peak), indicating that helium is confined to the outermost ejecta; based on these NIR detections, we also identify likely contribution from Heiλ5876 in the optical spectra on a similar range of timescales. This represents the most definitive detection of helium in a bright SLSN-I to date, and indicates that progenitors with a thin helium layer can still explode as SLSNe. 

Astrophysical transient phenomena are traditionally classified spectroscopically in a hierarchical taxonomy; however, this graph structure is currently not utilized in neural net-based photometric classifiers for time-domain astrophysics. Instead, independent classifiers are trained for different tiers of classified data, and events are excluded if they fall outside of these well-defined but flat classification schemes. Here, we introduce a weighted hierarchical cross-entropy objective function for classification of astrophysical transients. Our method allows users to directly build and use physics- or observationally-motivated tree-based taxonomies. Our weighted hierarchical cross-entropy loss directly uses this graph to accurately classify all targets into any node of the tree, re-weighting imbalanced classes. We test our novel loss on a set of variable stars and extragalactic transients from the Zwicky Transient Facility, showing that we can achieve similar performance to fine-tuned classifiers with the advantage of notably more flexibility in downstream classification tasks. 

Abstract We presentSLIDE, a pipeline that enables transient discovery in data from the Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST), using archival images from the Dark Energy Camera as templates for difference imaging. We apply this pipeline to the recently released Data Preview 1 (DP1; the first public release of Rubin commissioning data) and search for transients in the resulting difference images. The image subtraction, photometry extraction, and transient detection are all performed on the Rubin Science Platform. We demonstrate thatSLIDEeffectively extracts clean photometry by circumventing poor or missing LSST templates. We identified 29 previously unreported transients, 12 of which would not have been detected based on the DP1DiaObjectcatalog.SLIDEwill be especially useful for transient analysis in the early years of LSST, when template coverage will be largely incomplete or when templates may be contaminated by transients present at the time of acquisition. We present multiband light curves for a sample of known transients, along with new transient candidates identified through our search. Finally, we discuss the prospects of applying this pipeline during the main LSST survey. Our pipeline is broadly applicable and will support studies of all transients with slowly evolving phases. 

Abstract We present Young Supernova Experimentgrizyphotometry of SN 2021hpr, the third Type Ia supernova sibling to explode in the Cepheid calibrator galaxy, NGC 3147. Siblings are useful for improving SN-host distance estimates and investigating their contributions toward the SN Ia intrinsic scatter (post-standardization residual scatter in distance estimates). We thus develop a principled Bayesian framework for analyzing SN Ia siblings. At its core is the cosmology-independent relative intrinsic scatter parameter,σ_Rel: the dispersion of siblings distance estimates relative to one another within a galaxy. It quantifies the contribution toward the total intrinsic scatter,σ₀, from within-galaxy variations about the siblings’ common properties. It also affects the combined distance uncertainty. We present analytic formulae for computing aσ_Relposterior from individual siblings distances (estimated using any SN model). Applying a newly trainedBayeSNmodel, we fit the light curves of each sibling in NGC 3147 individually, to yield consistent distance estimates. However, the wideσ_Relposterior meansσ_Rel≈σ₀is not ruled out. We thus combine the distances by marginalizing overσ_Relwith an informative prior:σ_Rel∼U(0,σ₀). Simultaneously fitting the trio’s light curves improves constraints on distanceandeach sibling’s individual dust parameters, compared to individual fits. Higher correlation also tightens dust parameter constraints. Therefore,σ_Relmarginalization yields robust estimates of siblings distances for cosmology, as well as dust parameters for sibling–host correlation studies. Incorporating NGC 3147's Cepheid distance yieldsH₀= 78.4 ± 6.5 km s⁻¹Mpc⁻¹. Our work motivates analyses of homogeneous siblings samples, to constrainσ_Reland its SN-model dependence.