[Paper Review] Gaia GraL: Gaia DR2 Gravitational Lens Systems. V. Doubly-imaged QSOs discovered from entropy and wavelets

A. Krone-Martins, M. J. Graham+23 more|arXiv (Cornell University)|Dec 19, 2019

Stellar, planetary, and galactic studies7 references13 citations

TL;DR

This paper presents a novel method to discover doubly-imaged quasars (QSOs) using low-resolution photometric time-series from the Catalina Survey and wavelet-based image analysis of Pan-STARRS and DECaLS data, combined with Gaia DR2 astrometric anomalies. The approach identifies 7 new doubly-imaged QSOs and 1 likely double quasar, demonstrating that unresolved space astrometry and time-domain data can effectively detect strong gravitational lenses without requiring high-resolution imaging.

ABSTRACT

The discovery of multiply-imaged gravitationally lensed QSOs is fundamental to many astronomical and cosmological studies. However, these objects are rare and challenging to discover due to requirements of high-angular resolution astrometric, multiwavelength photometric and spectroscopic data. This has limited the number of known systems to a few hundred objects. We aim to reduce the constraints on angular resolution and discover multiply-imaged QSO candidates by using new candidate selection principles based on unresolved photometric time-series and ground-based images from public surveys. We selected candidates for multiply-imaged QSOs based on low levels of entropy computed from Catalina unresolved photometric time-series or Euclidean similarity to known lenses in a space defined by the wavelet power spectra of Pan-STARSS DR2 or DECaLS DR7 images, combined with multiple {\it Gaia} DR2 sources or large astrometric errors and supervised and unsupervised learning methods. We then confirmed spectroscopically some candidates with the Palomar Hale, Keck-I, and ESO/NTT telescopes. Here we report the discovery and confirmation of seven doubly-imaged QSOs and one likely double quasar. This demonstrates the potential of combining space-astrometry, even if unresolved, with low spatial-resolution photometric time-series and/or low-spatial resolution multi-band imaging to discover multiply-imaged lensed QSOs.

Motivation & Objective

• To overcome the limitations of high-angular resolution requirements in discovering multiply-imaged quasars.
• To reduce reliance on high-resolution imaging and spectroscopy by leveraging time-domain photometry and low-resolution multi-band imaging.
• To develop and validate new selection criteria based on entropy in light curves and wavelet similarity to known lenses.
• To demonstrate the feasibility of detecting strong gravitational lenses using only unresolved astrometric data from Gaia DR2 and public survey images.
• To expand the known sample of doubly-imaged QSOs using data-driven, scalable methods applicable to upcoming LSST and Gaia data releases.

Proposed method

• Entropy is computed from unresolved photometric light curves of the Catalina Survey to identify variable sources with lens-like behavior.
• Wavelet power spectra are extracted from Pan-STARRS DR2 and DECaLS DR7 images to quantify morphological similarity to known lens systems.
• Candidates are selected based on low entropy in light curves and high wavelet similarity to known lenses, combined with multiple Gaia DR2 sources or large astrometric errors.
• Supervised and unsupervised machine learning techniques are applied to refine candidate selection and prioritize targets.
• Spectroscopic follow-up is conducted using the Palomar Hale, Keck-I, and ESO/NTT telescopes to confirm lensed QSOs.
• The method integrates space astrometry (Gaia DR2) with ground-based time-domain and imaging surveys to enable lens detection without requiring high-resolution imaging.

Experimental results

Research questions

• RQ1 Can low-entropy photometric light curves from unresolved time-series data effectively identify potential lensed QSOs?
• RQ2 To what extent can wavelet-based image analysis of low-resolution survey data detect morphological similarities to known gravitational lens systems?
• RQ3 Can Gaia DR2 astrometric anomalies (e.g., large errors or multiple sources) serve as effective indicators of lensed quasars when combined with photometric and imaging data?
• RQ4 How effective is the combination of entropy, wavelet similarity, and astrometric signatures in reducing false positives in lens candidate selection?
• RQ5 Can this method discover new doubly-imaged QSOs without relying on high-resolution imaging or prior knowledge of lens geometry?

Key findings

• Seven new doubly-imaged quasars were spectroscopically confirmed using the proposed method, significantly expanding the known sample of such systems.
• One additional candidate was identified as a likely double quasar, further validating the method's sensitivity to lensed QSOs.
• The method successfully detected lensed QSOs using only unresolved photometric time-series and low-spatial-resolution imaging, bypassing the need for high-resolution data.
• The combination of low entropy in light curves and high wavelet similarity to known lenses proved effective in distinguishing true lens candidates from false positives.
• The use of Gaia DR2 astrometric anomalies (e.g., multiple sources or large errors) significantly improved candidate prioritization, especially when combined with time-domain and imaging features.
• The approach demonstrates strong potential for application to future large-scale surveys such as LSST and upcoming Gaia data releases, enabling scalable, data-driven lens discovery.