Etienne Russeil
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Etienne Russeil has a total of 92 co-authors. Here are the top co-authors by number of shared publications:
- Emille E. O. Ishida
Université Clermont Auvergne, CNRS, LPCA
10 shared publications - Konstantin Malanchev
McWilliams Center for Cosmology and Astrophysics, Department of Physics, Carnegie Mellon University, Pittsburgh, PA 15213, USA
9 shared publications - Maria Victorovna Pruzhinskaya
Lomonosov Moscow State University
9 shared publications - Sreevarsha Sreejith
University of Surrey
8 shared publications - Matwey V. Kornilov
Lomonosov Moscow State University
7 shared publications - Patrick D. Aleo
Department of Astronomy, University of Illinois at Urbana-Champaign, 1002 W. Green St., IL 61801, USA
6 shared publications
Latest Publications
- Open Access
- 11 September 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 713, A141
NOMAI: A real-time photometric classifier for superluminous supernova identification
Superluminous supernovae (SLSNe) are one of the most luminous stellar explosions known, yet they remain poorly understood. Because they are intrinsically rare, efficiently identifying them in the large alert streams produced by modern time-domain surveys is essential for building larger observational samples and enabling spectroscopic follow-up. We present NOMAI, a machine-learning classifier designed to identify SLSN candidates directly from photometric alerts in the ZTF stream, using light curves accumulated over at least 30 days. It does not require any spectroscopic redshift and is running in real time within the Fink broker. ZTF light curves were transformed into a set of physically motivated features derived primarily from model-fitting procedures using SALT2 and Rainbow, a blackbody-based multiband fitting framework. These features were used to train an XGBoost classifier on a curated dataset of labeled ZTF sources constructed using literature samples of SLSNe, along with TNS and internal ZTF labeled sources. The final dataset contains 5280 unique sources, including 225 spectroscopically classified SLSNe. Averaged over 100 bootstrap realizations, the classifier reaches 66% completeness and 58% purity. Deployed within the Fink broker, NOMAI has been running continuously since 18 December 2025 on the ZTF alert stream and publicly reports SLSN candidates every night by automatically posting them to dedicated communication channels. Based on this, we also report the first two-month evaluation period, where the classifier successfully recovered 22 of the 24 active SLSNe reported on the Transient Name Server, providing an upper limit on its real-time completeness. The achieved performances, particularly the high completeness in real-time operations, demonstrate that the classifier provides a valuable tool for experts to efficiently scan the alert stream and identify promising candidates. In the near future, NOMAI is intended to be adapted to operate on the Legacy Survey of Space and Time conducted by the Vera C. Rubin Observatory, which is expected to uncover an unprecedented number of transients, making machine-learning based photometric classification essential.
- 11 September 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 713, A141
Open AccessNOMAI: A real-time photometric classifier for superluminous supernova identification
Superluminous supernovae (SLSNe) are one of the most luminous stellar explosions known, yet they remain poorly understood. Because they are intrinsically rare, efficiently identifying them in the large alert streams produced by modern time-domain surveys is essential for building larger observational samples and enabling spectroscopic follow-up. We present NOMAI, a machine-learning classifier designed to identify SLSN candidates directly from photometric alerts in the ZTF stream, using light curves accumulated over at least 30 days. It does not require any spectroscopic redshift and is running in real time within the Fink broker. ZTF light curves were transformed into a set of physically motivated features derived primarily from model-fitting procedures using SALT2 and Rainbow, a blackbody-based multiband fitting framework. These features were used to train an XGBoost classifier on a curated dataset of labeled ZTF sources constructed using literature samples of SLSNe, along with TNS and internal ZTF labeled sources. The final dataset contains 5280 unique sources, including 225 spectroscopically classified SLSNe. Averaged over 100 bootstrap realizations, the classifier reaches 66% completeness and 58% purity. Deployed within the Fink broker, NOMAI has been running continuously since 18 December 2025 on the ZTF alert stream and publicly reports SLSN candidates every night by automatically posting them to dedicated communication channels. Based on this, we also report the first two-month evaluation period, where the classifier successfully recovered 22 of the 24 active SLSNe reported on the Transient Name Server, providing an upper limit on its real-time completeness. The achieved performances, particularly the high completeness in real-time operations, demonstrate that the classifier provides a valuable tool for experts to efficiently scan the alert stream and identify promising candidates. In the near future, NOMAI is intended to be adapted to operate on the Legacy Survey of Space and Time conducted by the Vera C. Rubin Observatory, which is expected to uncover an unprecedented number of transients, making machine-learning based photometric classification essential.
- Open Access
- 19 June 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 710, A247
Lost and found: A gallery of overlooked optical nuclear transients from the ZTF archive
Context. Tidal disruption events (TDEs) correspond to a star’s destruction by the tidal forces around a black hole, leading to outbursts, which can last from months to years. These transients are rare, and increasing the current sample is paramount for understanding them. As part of the F INK alert broker, we developed an early detection system for TDEs in Zwicky Transient Facility (ZTF) data. Aims. In this paper, we report the optical transients found either during this tool’s development, or through application of the classifier to the existing archive. We used this sample to anticipate the improvements that will need to be implemented in TDE detection systems for future surveys. Methods. For all the transients, we present optical and infrared archival photometry from ZTF, WISE, and Catalina, along with an assessment of each host’s previous nuclear activity. We fitted the ZTF light curves with both a phenomenological and a physically motivated model. Results. We report a total of 19 optical nuclear transients: nine in passive galaxies, eight in active galactic nuclei (AGNs), and two with uncertain host activity. We identify two newly discovered repeating TDE candidates and compare them to the current sample of repeating optical nuclear transients. One transient is exceptionally long-lived (> 5 years) in a previously passive galaxy. Three flares in active galaxies have absolute g -band magnitudes brighter than −24 mag, making them new extreme nuclear transient candidates. One seemingly repeating object was revealed to be two independent supernovae in the same galaxy. In the absence of spectral follow-up, definitive classification of most transients remains difficult, and some may be powered by AGN variability. Conclusions. This sample shows both the potential of our detection system for future discoveries and the value of archival searches in revealing overlooked transients. It also raises several points of concern and avenues of improvement for current and future classifiers.
- 19 June 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 710, A247
Open AccessLost and found: A gallery of overlooked optical nuclear transients from the ZTF archive
Context. Tidal disruption events (TDEs) correspond to a star’s destruction by the tidal forces around a black hole, leading to outbursts, which can last from months to years. These transients are rare, and increasing the current sample is paramount for understanding them. As part of the F INK alert broker, we developed an early detection system for TDEs in Zwicky Transient Facility (ZTF) data. Aims. In this paper, we report the optical transients found either during this tool’s development, or through application of the classifier to the existing archive. We used this sample to anticipate the improvements that will need to be implemented in TDE detection systems for future surveys. Methods. For all the transients, we present optical and infrared archival photometry from ZTF, WISE, and Catalina, along with an assessment of each host’s previous nuclear activity. We fitted the ZTF light curves with both a phenomenological and a physically motivated model. Results. We report a total of 19 optical nuclear transients: nine in passive galaxies, eight in active galactic nuclei (AGNs), and two with uncertain host activity. We identify two newly discovered repeating TDE candidates and compare them to the current sample of repeating optical nuclear transients. One transient is exceptionally long-lived (> 5 years) in a previously passive galaxy. Three flares in active galaxies have absolute g -band magnitudes brighter than −24 mag, making them new extreme nuclear transient candidates. One seemingly repeating object was revealed to be two independent supernovae in the same galaxy. In the absence of spectral follow-up, definitive classification of most transients remains difficult, and some may be powered by AGN variability. Conclusions. This sample shows both the potential of our detection system for future discoveries and the value of archival searches in revealing overlooked transients. It also raises several points of concern and avenues of improvement for current and future classifiers.
- Open Access
- 9 April 2026
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 384 (2317)
Exploring multi-view symbolic regression methods in physical sciences
Describing the world’s behaviour through mathematical functions helps scientists to achieve a better understanding of the inner mechanisms of different phenomena. Traditionally this is done by deriving new equations from first principles and careful observations. A modern alternative is to automate part of this process with symbolic regression (SR). The SR algorithms search for a function that adequately fits the observed data while trying to enforce sparsity, in the hopes of generating an interpretable equation. A particularly interesting extension to these algorithms is the multi-view symbolic regression (MvSR). It searches for a parametric function capable of describing multiple datasets generated by the same phenomena, which helps to mitigate the common problems of overfitting and data scarcity. Recently, multiple implementations added support to MvSR with small differences between them. In this paper, we test and compare MvSR as supported in Operon, PySR, ϕ-SO and eggp, in different real-world datasets. We show that they all often achieve good accuracy while proposing solutions with only a few free parameters. However, we find that certain features enable a more frequent generation of better models. We conclude by providing guidelines for future MvSR developments. This article is part of the discussion meeting issue ‘Symbolic regression in the physical sciences’.
- 9 April 2026
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 384 (2317)
Open AccessExploring multi-view symbolic regression methods in physical sciences
Describing the world’s behaviour through mathematical functions helps scientists to achieve a better understanding of the inner mechanisms of different phenomena. Traditionally this is done by deriving new equations from first principles and careful observations. A modern alternative is to automate part of this process with symbolic regression (SR). The SR algorithms search for a function that adequately fits the observed data while trying to enforce sparsity, in the hopes of generating an interpretable equation. A particularly interesting extension to these algorithms is the multi-view symbolic regression (MvSR). It searches for a parametric function capable of describing multiple datasets generated by the same phenomena, which helps to mitigate the common problems of overfitting and data scarcity. Recently, multiple implementations added support to MvSR with small differences between them. In this paper, we test and compare MvSR as supported in Operon, PySR, ϕ-SO and eggp, in different real-world datasets. We show that they all often achieve good accuracy while proposing solutions with only a few free parameters. However, we find that certain features enable a more frequent generation of better models. We conclude by providing guidelines for future MvSR developments. This article is part of the discussion meeting issue ‘Symbolic regression in the physical sciences’.
- Open Access
- 19 March 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 707, A338
Eruptive mass loss less than a year before the explosion of superluminous supernovae
We present X-shooter spectroscopic and photometric observations of a sample of 21 hydrogen-poor superluminous supernovae (SLSNe-I), spanning a redshift range of z = 0.13 − 0.95, aimed at searching for shells of circumstellar material (CSM). Specifically, we focused on identifying broad Mg II absorption features that are blueshifted by several thousand kilometers per second relative to the narrow absorption lines associated with the host galaxy. These broad features have previously been interpreted to arise from resonance line scattering of the SLSN continuum by rapidly expanding CSM ejected shortly before explosion. Utilizing high-quality near-ultraviolet spectra, we modeled the region around 2800 Å to characterize the Mg II line profiles, enabling us to either confirm their presence or place constraints on undetected CSM shells. We identified five objects in our sample that show broad Mg II absorption features consistent with the presence of CSM. While SN 2018ibb, SN 2020xga, and SN 2022xgc have been previously reported, we identified previously undiscovered CSM shells in DES15S2nr and DES16C3ggu. In the case of DES15S2nr, the CSM shell is located at ∼3.4 × 10 15 cm and is moving with a maximum velocity of ∼4800 km s −1 . For DES16C3ggu, the shell lies at ∼4.8 × 10 15 cm and reaches up to ∼4700 km s −1 . These shells were likely expelled approximately two and three months before the explosion of their respective associated SNe on timescales consistent with late-stage eruptive mass-loss episodes. We further found evidence that the velocities of the CSM shells in all objects lie within 3000 − 5000 km s −1 , which may reflect an intrinsic property and could hint at a similar mass-ejection mechanism. We did not find any correlations between the shell properties and the SN properties, except for a marginal correlation between the light curve decline timescale and the shell velocities. This correlation needs further work; however, if it applies, it is a powerful link between the late-time mass ejection and eventual explosion. We further demonstrate that CSM configurations similar to the majority of the detected shells would have been observable in spectra with a signal-to-noise > 5 per resolution element, and that the lines from a shell are, in general, detectable except in cases where the shell is either very geometrically and/or optically thin. Therefore, we conclude that the non-detections are unlikely to arise from selection effects but they may instead point to the existence of a subclass of SLSN-I progenitors undergoing late-stage shell ejections shortly before explosion.
- 19 March 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 707, A338
Open AccessEruptive mass loss less than a year before the explosion of superluminous supernovae
We present X-shooter spectroscopic and photometric observations of a sample of 21 hydrogen-poor superluminous supernovae (SLSNe-I), spanning a redshift range of z = 0.13 − 0.95, aimed at searching for shells of circumstellar material (CSM). Specifically, we focused on identifying broad Mg II absorption features that are blueshifted by several thousand kilometers per second relative to the narrow absorption lines associated with the host galaxy. These broad features have previously been interpreted to arise from resonance line scattering of the SLSN continuum by rapidly expanding CSM ejected shortly before explosion. Utilizing high-quality near-ultraviolet spectra, we modeled the region around 2800 Å to characterize the Mg II line profiles, enabling us to either confirm their presence or place constraints on undetected CSM shells. We identified five objects in our sample that show broad Mg II absorption features consistent with the presence of CSM. While SN 2018ibb, SN 2020xga, and SN 2022xgc have been previously reported, we identified previously undiscovered CSM shells in DES15S2nr and DES16C3ggu. In the case of DES15S2nr, the CSM shell is located at ∼3.4 × 10 15 cm and is moving with a maximum velocity of ∼4800 km s −1 . For DES16C3ggu, the shell lies at ∼4.8 × 10 15 cm and reaches up to ∼4700 km s −1 . These shells were likely expelled approximately two and three months before the explosion of their respective associated SNe on timescales consistent with late-stage eruptive mass-loss episodes. We further found evidence that the velocities of the CSM shells in all objects lie within 3000 − 5000 km s −1 , which may reflect an intrinsic property and could hint at a similar mass-ejection mechanism. We did not find any correlations between the shell properties and the SN properties, except for a marginal correlation between the light curve decline timescale and the shell velocities. This correlation needs further work; however, if it applies, it is a powerful link between the late-time mass ejection and eventual explosion. We further demonstrate that CSM configurations similar to the majority of the detected shells would have been observable in spectra with a signal-to-noise > 5 per resolution element, and that the lines from a shell are, in general, detectable except in cases where the shell is either very geometrically and/or optically thin. Therefore, we conclude that the non-detections are unlikely to arise from selection effects but they may instead point to the existence of a subclass of SLSN-I progenitors undergoing late-stage shell ejections shortly before explosion.
- Open Access
- 14 January 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 705, A144
Early identification of optical tidal disruption events
Context. The detection of tidal disruption events (TDEs) is one of the key science goals of large optical time-domain surveys such as the Zwicky Transient Facility (ZTF) and the upcoming Vera C. Rubin Observatory Legacy Survey of Space and Time. Automated and reliable classification pipelines that can select promising candidates in real time are required to identify TDEs in the vast alert streams produced by these surveys, however. Aims. We developed a module within the FINK alert broker to identify TDEs during their rising phase. The module was built to autonomously operate within the ZTF alert stream and to produce a list of candidates every night, enabling spectral and multiwavelength follow-up near peak brightness. Methods. All rising alerts were submitted to selection cuts and feature extraction using the R AINBOW multiband light-curve fit. Best-fit values were used as input to train an XGBoost classifier with the goal of identifying TDEs. The training set was constructed using ZTF observations for objects with available classification in the Transient Name Server. Finally, candidates for which the probability was high enough were inspected visually. Results. The classifier achieved 76% recall, which indicates a strong performance in early-phase identification, despite the limited available information before the peak. Out of the known TDEs that passed the selection cuts, half were flagged as TDEs before they had risen half the way. This proves that an early classification is possible. Additionally, new candidates were identified by applying the classifier on archival data, including a likely repeated TDE and some potential TDEs that occurred in active galaxies. The module is implemented in the FINK alert-processing framework and each night reports a small number of candidates to dedicated communication channels through a user-friendly interface for manual vetting and potential follow-up.
- 14 January 2026
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 705, A144
Open AccessEarly identification of optical tidal disruption events
Context. The detection of tidal disruption events (TDEs) is one of the key science goals of large optical time-domain surveys such as the Zwicky Transient Facility (ZTF) and the upcoming Vera C. Rubin Observatory Legacy Survey of Space and Time. Automated and reliable classification pipelines that can select promising candidates in real time are required to identify TDEs in the vast alert streams produced by these surveys, however. Aims. We developed a module within the FINK alert broker to identify TDEs during their rising phase. The module was built to autonomously operate within the ZTF alert stream and to produce a list of candidates every night, enabling spectral and multiwavelength follow-up near peak brightness. Methods. All rising alerts were submitted to selection cuts and feature extraction using the R AINBOW multiband light-curve fit. Best-fit values were used as input to train an XGBoost classifier with the goal of identifying TDEs. The training set was constructed using ZTF observations for objects with available classification in the Transient Name Server. Finally, candidates for which the probability was high enough were inspected visually. Results. The classifier achieved 76% recall, which indicates a strong performance in early-phase identification, despite the limited available information before the peak. Out of the known TDEs that passed the selection cuts, half were flagged as TDEs before they had risen half the way. This proves that an early classification is possible. Additionally, new candidates were identified by applying the classifier on archival data, including a likely repeated TDE and some potential TDEs that occurred in active galaxies. The module is implemented in the FINK alert-processing framework and each night reports a small number of candidates to dedicated communication channels through a user-friendly interface for manual vetting and potential follow-up.
- 9 August 2025
- Published by Springer Nature
- p. 97-104
Signatures to Help Interpretability of Anomalies
- Open Access
- 13 December 2024
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 692, A208
Transient classifiers for Fink
Context. The upcoming Legacy Survey of Space and Time (LSST) at the Vera C. Rubin Observatory is expected to detect a few million transients per night, which will generate a live alert stream during the entire ten years of the survey. This stream will be distributed via community brokers whose task is to select subsets of the stream and direct them to scientific communities. Given the volume and complexity of the anticipated data, machine learning (ML) algorithms will be paramount for this task. Aims. We present the infrastructure tests and classification methods developed within the FINK broker in preparation for LSST. This work aims to provide detailed information regarding the underlying assumptions and methods behind each classifier and enable users to make informed follow-up decisions from FINK photometric classifications. Methods. Using simulated data from the Extended LSST Astronomical Time-series Classification Challenge (ELAsTiCC), we showcase the performance of binary and multi-class ML classifiers available in FINK. These include tree-based classifiers coupled with tailored feature extraction strategies as well as deep learning algorithms. Moreover, we introduce the CBPF (Centro Brasileiro de Pesquisas Físicas) Alert Transient Search (CATS), a deep learning architecture specifically designed for this task. Results. Our results show that FINK classifiers are able to handle the extra complexity that is expected from LSST data. CATS achieved ≥93% precision for all classes except ‘long’ (for which it achieved ∼83%), while our best performing binary classifier achieves ≥98% precision and ≥99% completeness when classifying the periodic class. Conclusions. ELAsTiCC was an important milestone in preparing the FINK infrastructure to deal with LSST-like data. Our results demonstrate that FINK classifiers are well prepared for the arrival of the new stream, but this work also highlights that transitioning from the current infrastructures to Rubin will require significant adaptation of the currently available tools. This work was the first step in the right direction.
- 13 December 2024
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 692, A208
Open AccessTransient classifiers for Fink
Context. The upcoming Legacy Survey of Space and Time (LSST) at the Vera C. Rubin Observatory is expected to detect a few million transients per night, which will generate a live alert stream during the entire ten years of the survey. This stream will be distributed via community brokers whose task is to select subsets of the stream and direct them to scientific communities. Given the volume and complexity of the anticipated data, machine learning (ML) algorithms will be paramount for this task. Aims. We present the infrastructure tests and classification methods developed within the FINK broker in preparation for LSST. This work aims to provide detailed information regarding the underlying assumptions and methods behind each classifier and enable users to make informed follow-up decisions from FINK photometric classifications. Methods. Using simulated data from the Extended LSST Astronomical Time-series Classification Challenge (ELAsTiCC), we showcase the performance of binary and multi-class ML classifiers available in FINK. These include tree-based classifiers coupled with tailored feature extraction strategies as well as deep learning algorithms. Moreover, we introduce the CBPF (Centro Brasileiro de Pesquisas Físicas) Alert Transient Search (CATS), a deep learning architecture specifically designed for this task. Results. Our results show that FINK classifiers are able to handle the extra complexity that is expected from LSST data. CATS achieved ≥93% precision for all classes except ‘long’ (for which it achieved ∼83%), while our best performing binary classifier achieves ≥98% precision and ≥99% completeness when classifying the periodic class. Conclusions. ELAsTiCC was an important milestone in preparing the FINK infrastructure to deal with LSST-like data. Our results demonstrate that FINK classifiers are well prepared for the arrival of the new stream, but this work also highlights that transitioning from the current infrastructures to Rubin will require significant adaptation of the currently available tools. This work was the first step in the right direction.
- 1 October 2024
- Published by Springer Nature
- p. 195-208
Exploring the Universe with SNAD: Anomaly Detection in Astronomy
- 1 October 2024
- Published by Springer Nature
- p. 195-208
Exploring the Universe with SNAD: Anomaly Detection in Astronomy
- Open Access
- 14 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 961-970
Multiview Symbolic Regression
Symbolic regression (SR) searches for analytical expressions representing the relationship between explanatory and response variables. Current SR methods assume a single dataset extracted from a single experiment. Nevertheless, frequently, the researcher is confronted with multiple sets of results obtained from experiments conducted with different set-ups. Traditional SR methods may fail to find the underlying expression since the parameters of each experiment can be different. In this work we present Multiview Symbolic Regression (MvSR), which takes into account multiple datasets simultaneously, mimicking experimental environments, and outputs a general parametric solution. This approach fits the evaluated expression to each independent dataset and returns a parametric family of functions f(x; θ) simultaneously capable of accurately fitting all datasets. We demonstrate the effectiveness of MvSR using data generated from known expressions, as well as real-world data from astronomy, chemistry and economy, for which an a priori analytical expression is not available. Results show that MvSR obtains the correct expression more frequently and is robust to hyperparameters change. In real-world data, it is able to grasp the group behaviour, recovering known expressions from the literature as well as promising alternatives, thus enabling the use MvSR to a large range of experimental scenarios.
- 14 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 961-970
Open AccessMultiview Symbolic Regression
Symbolic regression (SR) searches for analytical expressions representing the relationship between explanatory and response variables. Current SR methods assume a single dataset extracted from a single experiment. Nevertheless, frequently, the researcher is confronted with multiple sets of results obtained from experiments conducted with different set-ups. Traditional SR methods may fail to find the underlying expression since the parameters of each experiment can be different. In this work we present Multiview Symbolic Regression (MvSR), which takes into account multiple datasets simultaneously, mimicking experimental environments, and outputs a general parametric solution. This approach fits the evaluated expression to each independent dataset and returns a parametric family of functions f(x; θ) simultaneously capable of accurately fitting all datasets. We demonstrate the effectiveness of MvSR using data generated from known expressions, as well as real-world data from astronomy, chemistry and economy, for which an a priori analytical expression is not available. Results show that MvSR obtains the correct expression more frequently and is robust to hyperparameters change. In real-world data, it is able to grasp the group behaviour, recovering known expressions from the literature as well as promising alternatives, thus enabling the use MvSR to a large range of experimental scenarios.
- Open Access
- 29 March 2024
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 683, A251
RAINBOW: A colorful approach to multipassband light-curve estimation
Context. Time series generated by repeatedly observing astronomical transients are generally sparse, irregularly sampled, noisy, and multidimensional (obtained through a set of broad-band filters). In order to fully exploit their scientific potential, it is necessary to use this incomplete information to estimate a continuous light-curve behavior. Traditional approaches use ad hoc functional forms to approximate the light curve in each filter independently (hereafter, the MONOCHROMATIC method). Aims. We present RAINBOW, a physically motivated framework that enables simultaneous multiband light-curve fitting. It allows the user to construct a 2D continuous surface across wavelength and time, even when the number of observations in each filter is significantly limited. Methods. Assuming the electromagnetic radiation emission from the transient can be approximated by a blackbody, we combined an expected temperature evolution and a parametric function describing its bolometric light curve. These three ingredients allow the information available in one passband to guide the reconstruction in the others, thus enabling a proper use of multisurvey data. We demonstrate the effectiveness of our method by applying it to simulated data from the Photometric LSST Astronomical Time-series Classification Challenge (PLAsTiCC) as well as to real data from the Young Supernova Experiment (YSE DR1). Results. We evaluate the quality of the estimated light curves according to three different tests: goodness of fit, peak-time prediction, and ability to transfer information to machine-learning (ML) based classifiers. The results confirm that RAINBOW leads to an equivalent goodness of fit (supernovae II) or to a goodness of fit that is better by up to 75% (supernovae Ibc) than the MONOCHROMATIC approach. Similarly, the accuracy improves for all classes in our sample when the RAINBOW best-fit values are used as a parameter space in a multiclass ML classification. Conclusions. Our approach enables a straightforward light-curve estimation for objects with observations in multiple filters and from multiple experiments. It is particularly well suited when the light-curve sampling is sparse. We demonstrate its potential for characterizing supernova-like events here, but the same approach can be used for other classes by changing the function describing the light-curve behavior and temperature representation. In the context of the upcoming large-scale sky surveys and their potential for multisurvey analysis, this represents an important milestone in the path to enable population studies of photometric transients.
- 29 March 2024
- Published by EDP Sciences in Astronomy & Astrophysics
- Vol. 683, A251
Open AccessRAINBOW: A colorful approach to multipassband light-curve estimation
Context. Time series generated by repeatedly observing astronomical transients are generally sparse, irregularly sampled, noisy, and multidimensional (obtained through a set of broad-band filters). In order to fully exploit their scientific potential, it is necessary to use this incomplete information to estimate a continuous light-curve behavior. Traditional approaches use ad hoc functional forms to approximate the light curve in each filter independently (hereafter, the MONOCHROMATIC method). Aims. We present RAINBOW, a physically motivated framework that enables simultaneous multiband light-curve fitting. It allows the user to construct a 2D continuous surface across wavelength and time, even when the number of observations in each filter is significantly limited. Methods. Assuming the electromagnetic radiation emission from the transient can be approximated by a blackbody, we combined an expected temperature evolution and a parametric function describing its bolometric light curve. These three ingredients allow the information available in one passband to guide the reconstruction in the others, thus enabling a proper use of multisurvey data. We demonstrate the effectiveness of our method by applying it to simulated data from the Photometric LSST Astronomical Time-series Classification Challenge (PLAsTiCC) as well as to real data from the Young Supernova Experiment (YSE DR1). Results. We evaluate the quality of the estimated light curves according to three different tests: goodness of fit, peak-time prediction, and ability to transfer information to machine-learning (ML) based classifiers. The results confirm that RAINBOW leads to an equivalent goodness of fit (supernovae II) or to a goodness of fit that is better by up to 75% (supernovae Ibc) than the MONOCHROMATIC approach. Similarly, the accuracy improves for all classes in our sample when the RAINBOW best-fit values are used as a parameter space in a multiclass ML classification. Conclusions. Our approach enables a straightforward light-curve estimation for objects with observations in multiple filters and from multiple experiments. It is particularly well suited when the light-curve sampling is sparse. We demonstrate its potential for characterizing supernova-like events here, but the same approach can be used for other classes by changing the function describing the light-curve behavior and temperature representation. In the context of the upcoming large-scale sky surveys and their potential for multisurvey analysis, this represents an important milestone in the path to enable population studies of photometric transients.