arXiv subjects

Jorge Martínez-Palomera

Publications and source records attributed to Jorge Martínez-Palomera.

10 recordsLinked to original sources

arXiv2026

TESSExtractor: A web-based interactive tool for light-curve visualisation and period estimation using TESS full-frame images

We present a user-friendly web application that allows users to extract, visualise, and interact with TESS light-curves (LCs) from full-frame images. The service works for any star identified in the TIC database or using stellar coordinates. TESSExtractor performs photometry using a circular aperture centred on the object of interest, with a sky annulus for background subtraction. The application also uses cotrending basis vectors provided by TESS to correct photometry for possible systematic effects. The Lomb-Scargle Periodogram is used to obtain periods from the LC. The TESSExtractor's website offers an intuitive interface for scientific and educational purposes and is publicly available at https://www.tessextractor.app. This tool simplifies the analysis of TESS LCs and provides valuable insight for researchers seeking to extract and analyse TESS data for a variety of scientific goals (e.g., stellar rotation, activity, transient detection, asteroseismology, or simply detection of planetary transits or stellar companions, among others). TESS provides insights into our understanding of time-domain astrophysics, and TESSExtractor provides an accessible and intuitive web interface for scientific exploitation of TESS data. We demonstrate that TESSExtractor produces LCs of sufficient quality for both quantitative and morphological time-domain studies.

astro-ph.IM↗

arXiv2025

A self-regulated convolutional neural network for classifying variable stars

Over the last two decades, machine learning models have been widely applied and have proven effective in classifying variable stars, particularly with the adoption of deep learning architectures such as convolutional neural networks, recurrent neural networks, and transformer models. While these models have achieved high accuracy, they require high-quality, representative data and a large number of labelled samples for each star type to generalise well, which can be challenging in time-domain surveys. This challenge often leads to models learning and reinforcing biases inherent in the training data, an issue that is not easily detectable when validation is performed on subsamples from the same catalogue. The problem of biases in variable star data has been largely overlooked, and a definitive solution has yet to be established. In this paper, we propose a new approach to improve the reliability of classifiers in variable star classification by introducing a self-regulated training process. This process utilises synthetic samples generated by a physics-enhanced latent space variational autoencoder, incorporating six physical parameters from Gaia Data Release 3. Our method features a dynamic interaction between a classifier and a generative model, where the generative model produces ad-hoc synthetic light curves to reduce confusion during classifier training and populate underrepresented regions in the physical parameter space. Experiments conducted under various scenarios demonstrate that our self-regulated training approach outperforms traditional training methods for classifying variable stars on biased datasets, showing statistically significant improvements.

cs.LG↗

arXiv2021

Local Active Galactic Nuclei with Large Broad-Hα Variability Reside in Red Galaxies

Inspired by our serendipitous discovery of six AGNs with varying broad-Halpha fluxes over years out of our searching for intermediate-mass black holes (IMBHs), we conduct a systematic investigation of changing-look (CL) and large-variability AGNs. We collect all the CL AGNs at z<0.15 and the reverberation mapped AGNs with strongly variable broad Halpha, and perform careful decomposition fittings to both their images and spectra. We find two observational facts: (1) The host galaxies of local CL and large-variability AGNs, mainly being Seyferts, are in the red (gas-poor) tail of the general Seyfert galaxy population. (2) In contrast, there is a significant trend that their more luminous counterparts namely CL and extremely variable quasars (CLQs and EVQs) are different: CLQs are generally in blue galaxies; in terms of the diagram of SFR and M* local CL Seyfert galaxies are located in the green valley, whereas CLQ hosts are in the star-forming main sequence. We propose explanations for those strongly variable Seyferts and quasars, respectively, under the thought that accretion disks broadly depend on nuclear fueling modes. Local large-variability and CL Seyferts are in nuclear famine mode, where cold-gas clumps can be formed stochastically in the fueling flow, and their episodic infall produces sharp peaks in the accretion-rate curve. CLQs and EVQs are in feast fueling mode, which may account for both their preference to blue galaxies and their variability pattern (high-amplitude tail of the continuous distribution). Lastly, we propose a new thinking: to search for IMBHs by optical variability in red galaxies.

astro-ph.GA↗

arXiv2020

Deep Generative Modeling of Periodic Variable Stars Using Physical Parameters

The ability to generate physically plausible ensembles of variable sources is critical to the optimization of time-domain survey cadences and the training of classification models on datasets with few to no labels. Traditional data augmentation techniques expand training sets by reenvisioning observed exemplars, seeking to simulate observations of specific training sources under different (exogenous) conditions. Unlike fully theory-driven models, these approaches do not typically allow principled interpolation nor extrapolation. Moreover, the principal drawback of theory-driven models lies in the prohibitive computational cost of simulating source observables from {\it ab initio} parameters. In this work, we propose a computationally tractable machine learning approach to generate realistic light curves of periodic variables capable of integrating physical parameters and variability classes as inputs. Our deep generative model, inspired by the Transparent Latent Space Generative Adversarial Networks (TL-GANs), uses a Variational Autoencoder (VAE) architecture with Temporal Convolutional Network (TCN) layers, trained using the \hbox{OGLE-III} optical light curves and physical characteristics (e.g., effective temperature and absolute magnitude) from Gaia DR2. A test using the temperature-shape relationship of RR\,Lyrae demonstrates the efficacy of our generative "Physics-Enhanced Latent Space VAE" (PELS-VAE) model. Such deep generative models, serving as non-linear non-parametric emulators, present a novel tool for astronomers to create synthetic time series over arbitrary cadences.

astro-ph.IM↗

arXiv2019

Introducing the Search for Intermediate-mass Black-hole In Nearby Galaxy (SIBLING) Survey

Intermediate mass black holes (IMBHs) have masses between the $10^2\!-\!10^6$ M$_\odot$ and are key to our understanding of the formation of massive black holes. The known population of IMBH remains small, with a few hundred candidates and only a handful of them confirmed as bona-fide IMBHs. Until now, the most widely used selection method is based on spectral analysis. Here we present a methodology to select IMBH candidates via optical variability analysis of the nuclear region of local galaxies ($z \leqslant 0.35$). Active IMBH accreting at low rates show small amplitude variability with time scales of hours, as it is seen in one of the known IMBH NGC4395. We found a sample of $\sim \!500$ galaxies evidencing fast and small amplitude variation in their weekly based light curves. We estimate an average occupancy fraction of 4\% and a surface density of $\sim \!3$ deg$^{-2}$, which represent an increase by a factor of $\sim\!40$ compared to previous searches. A large fraction ($78\%$) of the candidates are in spiral galaxies. We preliminary confirm the AGN nature of 22 sources via BPT diagrams using SDSS legacy spectra. Further confirmation of these candidates will require multiwavelength observations, especially in X-ray and radio bands.

astro-ph.GA↗

arXiv2019

GROWTH on S190425z: Searching thousands of square degrees to identify an optical or infrared counterpart to a binary neutron star merger with the Zwicky Transient Facility and Palomar Gattini IR

The third observing run by LVC has brought the discovery of many compact binary coalescences. Following the detection of the first binary neutron star merger in this run (LIGO/Virgo S190425z), we performed a dedicated follow-up campaign with the Zwicky Transient Facility (ZTF) and Palomar Gattini-IR telescopes. The initial skymap of this single-detector gravitational wave (GW) trigger spanned most of the sky observable from Palomar Observatory. Covering 8000 deg$^2$ of the initial skymap over the next two nights, corresponding to 46\% integrated probability, ZTF system achieved a depth of $\approx$\,21 $m_\textrm{AB}$ in $g$- and $r$-bands. Palomar Gattini-IR covered 2200 square degrees in $J$-band to a depth of 15.5\,mag, including 32\% integrated probability based on the initial sky map. The revised skymap issued the following day reduced these numbers to 21\% for the Zwicky Transient Facility and 19\% for Palomar Gattini-IR. We narrowed 338,646 ZTF transient "alerts" over the first two nights of observations to 15 candidate counterparts. Two candidates, ZTF19aarykkb and ZTF19aarzaod, were particularly compelling given that their location, distance, and age were consistent with the GW event, and their early optical lightcurves were photometrically consistent with that of kilonovae. These two candidates were spectroscopically classified as young core-collapse supernovae. The remaining candidates were ruled-out as supernovae. Palomar Gattini-IR did not identify any viable candidates with multiple detections only after merger time. We demonstrate that even with single-detector GW events localized to thousands of square degrees, systematic kilonova discovery is feasible.

astro-ph.HE↗

arXiv2019

GROWTH on S190510g: DECam Observation Planning and Follow-Up of a Distant Binary Neutron Star Merger Candidate

The first two months of the third Advanced LIGO and Virgo observing run (2019 April-May) showed that distant gravitational wave (GW) events can now be readily detected. Three candidate mergers containing neutron stars (NS) were reported in a span of 15 days, all likely located more than 100 Mpc away. However, distant events such as the three new NS mergers are likely to be coarsely localized, which highlights the importance of facilities and scheduling systems that enable deep observations over hundreds to thousands of square degrees to detect the electromagnetic counterparts. On 2019-05-10 02:59:39.292 UT the GW candidate S190510g was discovered and initially classified as a BNS merger with 98% probability. The GW event was localized within an area of 3462 deg2, later refined to 1166 deg2 (90%) at a distance of 227 +- 92 Mpc. We triggered Target of Opportunity observations with the Dark Energy Camera (DECam), a wide-field optical imager mounted at the prime focus of the 4m Blanco Telescope at CTIO in Chile. This Letter describes our DECam observations and our real-time analysis results, focusing in particular on the design and implementation of the observing strategy. Within 24 hours of the merger time, we observed 65% of the total enclosed probability of the final skymap with an observing efficiency of 94%. We identified and publicly announced 13 candidate counterparts. S190510g was re-classified 1.7 days after the merger, after our observations were completed, with a "binary neutron star merger" probability reduced from 98% to 42% in favor of a "terrestrial" classification.

astro-ph.HE↗

arXiv2019

GROWTH on S190426c. II. Real-Time Search for a Counterpart to the Probable Neutron Star-Black Hole Merger using an Automated Difference Imaging Pipeline for DECam

The discovery of a transient kilonova following the gravitational-wave event GW170817 highlighted the critical need for coordinated rapid and wide-field observations, inference, and follow-up across the electromagnetic spectrum. In the Southern hemisphere, the Dark Energy Camera (DECam) on the Blanco 4-m telescope is well-suited to this task, as it is able to cover wide-fields quickly while still achieving the depths required to find kilonovae like the one accompanying GW170817 to $\sim$500 Mpc, the binary neutron star horizon distance for current generation of LIGO/Virgo collaboration (LVC) interferometers. Here, as part of the multi-facility followup by the Global Relay of Observatories Watching Transients Happen (GROWTH) collaboration, we describe the observations and automated data movement, data reduction, candidate discovery, and vetting pipeline of our target-of-opportunity DECam observations of S190426c, the first possible neutron star--black hole merger detected via gravitational waves. Starting 7.5hr after S190426c, over 11.28\,hr of observations, we imaged an area of 525\,deg$^2$ ($r$-band) and 437\,deg$^2$ ($z$-band); this was 16.3\% of the total original localization probability and nearly all of the probability density visible from the Southern hemisphere. The machine-learning based pipeline was optimized for fast turnaround, delivering transient candidates for human vetting within 17 minutes, on average, of shutter closure. We reported nine promising counterpart candidates 2.5 hours before the end of our observations. Our observations yielded no detection of a bona fide counterpart to $m_z = 22.5$ and $m_r = 22.9$ at the 5$σ$ level of significance, consistent with the refined LVC positioning. We view these observations and rapid inferencing as an important real-world test for this novel end-to-end wide-field pipeline.

astro-ph.HE↗

arXiv2018

Deep Learning for Image Sequence Classification of Astronomical Events

We propose a new sequential classification model for astronomical objects based on a recurrent convolutional neural network (RCNN) which uses sequences of images as inputs. This approach avoids the computation of light curves or difference images. This is the first time that sequences of images are used directly for the classification of variable objects in astronomy. The second contribution of this work is the image simulation process. We generate synthetic image sequences that take into account the instrumental and observing conditions, obtaining a realistic, set of movies for each astronomical object. The simulated dataset is used to train our RCNN classifier. This approach allows us to generate datasets to train and test our RCNN model for different astronomical surveys and telescopes. We aim at building a simulated dataset whose distribution is close enough to the real dataset, so that a fine tuning could match the distributions between real and simulated dataset. To test the RCNN classifier trained with the synthetic dataset, we used real-world data from the High cadence Transient Survey (HiTS) obtaining an average recall of 85%, improved to 94% after performing fine tuning with 10 real samples per class. We compare the results of our model with those of a light curve random forest classifier. The proposed RCNN with fine tuning has a similar performance on the HiTS dataset compared to the light curve classifier, trained on an augmented training set with 10 real samples per class. The RCNN approach presents several advantages in an alert stream classification scenario, such as a reduction of the data pre-processing, faster online evaluation and easier performance improvement using a few real data samples. These results encourage us to use this method for alert brokers systems that will process alert streams generated by new telescopes such as the Large Synoptic Survey Telescope.

astro-ph.IM↗

arXiv2018

The High Cadence Transient Survey (HITS): Compilation and characterization of light-curve catalogs

The High Cadence Transient Survey (HiTS) aims to discover and study transient objects with characteristic timescales between hours and days, such as pulsating, eclipsing and exploding stars. This survey represents a unique laboratory to explore large etendue observations from cadences of about 0.1 days and to test new computational tools for the analysis of large data. This work follows a fully \textit{Data Science} approach: from the raw data to the analysis and classification of variable sources. We compile a catalog of ${\sim}15$ million object detections and a catalog of ${\sim}2.5$ million light-curves classified by variability. The typical depth of the survey is $24.2$, $24.3$, $24.1$ and $23.8$ in $u$, $g$, $r$ and $i$ bands, respectively. We classified all point-like non-moving sources by first extracting features from their light-curves and then applying a Random Forest classifier. For the classification, we used a training set constructed using a combination of cross-matched catalogs, visual inspection, transfer/active learning and data augmentation. The classification model consists of several Random Forest classifiers organized in a hierarchical scheme. The classifier accuracy estimated on a test set is approximately $97\%$. In the unlabeled data, $3\,485$ sources were classified as variables, of which $1\,321$ were classified as periodic. Among the periodic classes we discovered with high confidence, 1 $δ$-scutti, 39 eclipsing binaries, 48 rotational variables and 90 RR-Lyrae and for the non-periodic classes we discovered 1 cataclysmic variables, 630 QSO, and 1 supernova candidates. The first data release can be accessed in the project archive of HiTS.

astro-ph.IM↗