arXiv subjects

Xiaoyue Cao

Publications and source records attributed to Xiaoyue Cao.

At least 19 recordsLinked to original sources

arXiv2026

Reducing False Positives in Strong-Lens Searches with Generalized-Mean Consensus of Machine-Learning Ensembles in the Kilo-Degree Survey

Context. In wide-field surveys, the main challenge is not just classifier sensitivity, but the overwhelming number of false positives. Searching for strong lenses among millions to bilions of galaxies produces many contaminants, making the bottleneck for follow-up inspection and building statistically useful lens samples. Aims. We aim to improve the purity of strong-lens candidate selection in KiDS DR4 by combining several classifiers. The objective is to retain high completeness for known candidates while substantially reducing the fraction of non-lenses. Methods. We trained convolutional, Transformer-based, and hybrid classifiers, including Li ResNet+, Swin Transformer variants, Swin-MLP, and DemiLensNet. Their probabilistic outputs were combined at score level using averaging and a generalized mean consensus. The models were tested on simulated KiDS-like lens images and then evaluated on real KiDS DR4 lens candidates embedded in a non-lens sample. Results. On the simulated test set, ensembles show no advantage over the best single models. On the mixed real KiDS test set, the arithmetic mean reduces the false-positive rate at 90% completeness from 0.016-0.020 (the range spanned by the two best individual models) to 0.011 for the seven-model ensemble. The generalized mean reduces it further, to 0.007. Applied to the full LRG and BG samples at the same 90% completeness level, the generalized mean reduces returned candidates by roughly 50% for LRGs and 70% for BGs, relative to the best single model. After visual inspection, we obtain 170 new high-quality candidates (24 Class A and 146 Class B), together with 1706 Class C candidates. Conclusions. Our results demonstrate that the generalized mean consensus of an ML ensemble strategy provides a practical route to reducing the visual inspection workload while preserving a high recovery rate of promising strong-lens candidates.

astro-ph.GA↗

arXiv2026

Identification of gravitational lenses obscured by foreground light in the KiDS dataset using U-Nets and ResNets

*Context.* Many lensing images are often obscured by foreground light from the central galaxies, making them challenging to detect. *Aims.* To address the limitations of previous lens search efforts, particularly for samples with smaller $R_E$ or faint lensed images, we developed a composite convolutional neural network framework that utilizes both U-Net and ResNet architectures for feature extraction and classification. *Methods.* We propose a hybrid search method that combines U-Net and ResNet architectures to enhance the detection of foreground galaxy-obscured lenses. Our approach consists of two main stages: first, the U-Net model separates the foreground galaxy light from potential lensing signals, creating residual images that highlight the lensing features. Next, the ResNet module performs binary classification on these residual images to detect lensing signals. *Results.* We evaluated the hybrid search method with real observational data to demonstrate its effectiveness, achieving a recall of 71.5% and a 4.5% false positive rate at a confidence threshold of 0.6. Applying this method to over 638,398 galaxy samples from the Kilo-Degree Survey Data Release 4 and conducting thorough inspections, we identify 88 Class A, 322 Class B, and 1,758 Class C candidates. *Conclusions.* This hybrid approach significantly enhances the completeness of existing strong gravitational lensing searches and shows great potential for improving future astronomical surveys.

astro-ph.IM↗

arXiv2026

LenNet: Direct Detection and Localization of Strong Gravitational Lenses in Wide-Field Sky Survey Images

Strong gravitational lenses are invaluable tools for addressing fundamental questions in astrophysics, from the nature of dark matter to the expansion of the universe. While current sky surveys have successfully identified thousands of lens candidates, the search methods employed face a critical challenge. The conventional approach relies on a "crop-and-classify" strategy, where small images are first cut out around billions of potential host galaxies before being individually classified. This process creates a significant computational and storage bottleneck that is unsustainable for future large-scale surveys. To overcome this limitation, we propose LenNet, an object detection model that identifies lenses directly within large, original survey images. Our method completely bypasses the inefficient cropping step by framing the problem as a direct detection and localization task. We initially train LenNet on simulated data to learn the complex features of gravitational lenses and then use transfer learning to fine-tune the model on a limited set of real, labeled examples from the Kilo-Degree Survey (KiDS). Our experiments show that LenNet performs remarkably well on real survey data, validating its potential as a highly efficient and scalable solution for lens discovery in massive astronomical surveys.

astro-ph.GA↗

arXiv2026

Morphology classification for galaxies in the Kilo Degree Survey using a label-efficient self-supervised learning framework

Galaxy morphology classification is fundamental to understanding galaxy formation and evolution. The advent of large-scale sky surveys has produced an unprecedented volume of galaxy images, making traditional manual classification impractical. Although supervised deep learning can achieve high accuracy, it requires large labeled datasets that are time-consuming to construct. In contrast, unsupervised methods often show limited classification performance. To address this limitation, we propose a label-efficient self-supervised learning framework for galaxy morphology classification. Our method first learns robust morphological representations from 305,583 unlabeled KiDS galaxy images through contrastive learning, and then trains a classifier using only 5,000 human-labeled images. The classifier separates galaxies into five categories: elliptical, spiral, lenticular-disk, irregular, and "other." Using a ResNet-50 model with a crop size of 64x64 pixels, our approach achieves an overall test accuracy of up to 91.0% (90.5% +/- 0.2% on average) on the human-classified catalog. The corresponding F1 scores for elliptical, spiral, irregular, lenticular-disk, and "other" galaxies are 0.96, 0.86, 0.86, 0.95, and 0.92, respectively. We apply this pipeline to the Kilo-Degree Survey Data Release 5 and produce a publicly available morphology catalog of 310,583 galaxies. This is the first morphology catalog for KiDS galaxies and provides a valuable resource for future studies of galaxy evolution. Our results show that self-supervised learning can substantially reduce the need for manual labels while maintaining high classification accuracy, making it a promising and scalable approach for automated galaxy morphology classification in the era of large-scale surveys.

astro-ph.GA↗

arXiv2026

Deep Learning Calibration of the Quasar X-ray/UV Luminosity Relation for Cosmological Applications

Quasars can serve as standard candles through an empirical scaling relation between their ultraviolet (UV) and X-ray luminosities. As high-redshift probes, it is critical to test whether this relation evolves with redshift. In this work, we reconstruct the Hubble diagram of the Pantheon+ sample using the deep learning--based LADDER algorithm and use it as a reference to investigate the quasar scaling relation. Our results, which are consistent with those from Gaussian process regression and narrow-bin analyses, show that the potentially contaminated sample at $z<0.7$ differs significantly from the $z>0.7$ sample; thus, it should be further screened or excluded when quasars are used as cosmological probes. We find that the scaling relation exhibits a non-linear redshift dependence that cannot be accounted for by a simple linear correction, and that this behavior is a feature of the current data sample rather than a consequence of cosmological model misspecification. To use quasars as standardizable candles, further modeling of the scaling relation and intrinsic dispersion, or more advanced data processing techniques, is required.

astro-ph.CO↗

arXiv2026

Detection of a dark matter subhalo in the strongly lensed system PJ011646

We present a strong lensing analysis of the system PJ011646 using high-resolution ($\sim$0.1 arcsec) Atacama Large Millimeter/submillimeter Array (ALMA) dust-continuum observations to test for the presence of dark matter substructures. The lens mass distribution is modelled with an elliptical power law and third- and fourth-order multipoles (PL+MP; $m=3,4$), plus external shear. The multipoles have amplitudes of $\simeq$1.5 per cent of the convergence, consistent with nearby early-type galaxies, and improve the fit by $Δ\ln Z = 52.1$ relative to a pure PL model. Using this best-fitting macromodel, we perform a grid-based subhalo search in the image plane, parametrising the perturber as a spherical NFW. A subhalo in two locations in the image plane improves the fit by $Δ\ln Z>10$. Both correspond to the same location in the source plane, so they are partially degenerate; follow-up analysis suggests that only one is physically real. This is a subhalo of mass $M_{200} = {2.78}_{-0.66}^{+0.43} \times 10^{10} \, M_\odot$ and concentration $c_{200} = 30_{-7}^{+5}$, detected at $\sim$5.8$σ$ significance (relative to the PL+MP). This concentration is consistent with that expected for a typical tidally stripped Navarro-Frenk-White subhalo. The enclosed projected mass is most tightly constrained within a radius of 2 kpc, where we infer $M_{\rm sub} = {3.57}_{-0.14}^{+0.16}\times 10^9 \, M_\odot$. From grid cells consistent with no detection ($Δ\ln Z < 10$), we derive limits on the minimum subhalo mass that could have been detected at $3σ$ significance, finding $M_{200} \approx 8 \times 10^{8} \, M_\odot$ in the most sensitive regions of the lensed arcs. This demonstrates that ALMA continuum imaging at sub-arcsecond resolution can probe dark matter substructure in a mass regime where cold and warm dark matter models predict different abundances of subhalos.

astro-ph.GA↗

arXiv2026

Discovery of a radio jet in the Cloverleaf Quasar at z = 2.56

The fast growth of supermassive black holes and their feedback to the host galaxies play an important role in regulating the evolution of galaxies, especially in the early Universe. However, due to cosmological dimming and the limited angular resolution of most observations, it is difficult to resolve the feedback from the active galactic nuclei (AGNs) to their host galaxies. Gravitational lensing, for its magnification, provides a powerful tool to spatially differentiate emission originating from AGN and host galaxy at high redshifts. Here we report a discovery of a jet-like radio structure in a strongly lensed starburst quasar, H1413+117 or Cloverleaf at redshift z= 2.56, based on observational data at optical, sub-millimetre, and radio wavelengths. With both parametric and non-parametric lens models and with reconstructed images in the source plane, we find a well-separated, kpc-scaled, single-sided radio jet located at projected ~1.2 kpc to the northwest of the host galaxy in the source plane. This could indicate the co-existence of feedback from the AGN by both wind and jet in the Cloverleaf quasar.

astro-ph.GA↗

arXiv2026

Measuring the Stellar-to-Halo Mass Relation at $\sim10^{10}$ Solar masses, using forthcoming space-based imaging of galaxy-galaxy strong lenses

The stellar-to-halo mass relation (SHMR) is central to understanding the co-evolution of galaxies and their host dark matter haloes, yet it remains weakly constrained for dwarf galaxies owing to their faintness, especially beyond the Local Group. Strong gravitational lensing offers a unique probe of the SHMR at sub-galactic scales and cosmological distances, as the masses of subhalos within the main lens can be inferred from the perturbations they imprint on lensed images. Anticipating the discovery of $\sim10^5$ galaxy--galaxy strong lenses by forthcoming facilities such as \textit{Euclid}, we perform an end-to-end simulation to forecast \textit{Euclid}'s constraints on the SHMR at the halo mass scale of $\sim10^{10}\,\mathrm{M}_\odot$. We generate mock \textit{Euclid} VIS images of lens systems hosting a fiducial $3\times10^{10}\,\mathrm{M}_\odot$ subhalo and vary its properties to assess the robustness of mass inference. We find that \textit{Euclid}'s angular resolution cannot break the intrinsic mass--concentration degeneracy of subhaloes, nor deblend the light of satellite galaxies (when present) associated with them, leading to biased inferred halo masses. These limitations are overcome with high-resolution follow-up imaging from facilities such as the \textit{Hubble Space Telescope}, enabling accurate halo-mass measurements. We forecast that a statistical sample of $\sim100$ such systems, combining lensing-derived halo masses with stellar masses from photometric SED fitting, can constrain the SHMR at dwarf-galaxy scales with a precision of $\sim0.05$~dex in halo mass and $\sim0.03$~dex in stellar mass, enabling powerful tests of galaxy formation theories.

astro-ph.GA↗

arXiv2026

Probing Dark Matter Substructures with Free-Form Modelling: A Case Study of the `Jackpot' Strong Lens

Characterising the population and internal structure of sub-galactic halos is critical for constraining the nature of dark matter. These halos can be detected near galaxies that act as strong gravitational lenses with extended arcs, as they perturb the shapes of the arcs. However, this method is subject to false-positive detections and systematic uncertainties, particularly degeneracies between an individual halo and larger-scale asymmetries in the distribution of lens mass. We present a new free-form lens modelling code, developed within the framework of the open-source software \texttt{PyAutoLens}, to address these challenges. Our method models mass perturbations that cannot be captured by parametric models as pixelized potential corrections and suppresses unphysical solutions via a Matérn regularisation scheme that is inspired by Gaussian process regression. This approach enables the recovery of diverse mass perturbations, including subhalos, line-of-sight halos, external shear, and multipole components that represent the complex angular mass distribution of the lens galaxy, such as boxiness/diskiness. Additionally, our fully Bayesian framework objectively infers hyperparameters associated with the regularisation of pixelized sources and potential corrections, eliminating the need for manual fine-tuning. By applying our code to the well-known `Jackpot' lens system, SLACS0946+1006, we robustly detect a highly concentrated subhalo that challenges the standard cold dark matter model. This study represents the first attempt to independently reveal the mass distribution of a subhalo using a fully free-form approach.

astro-ph.CO↗

arXiv2026

CSST Strong Lensing Preparation: Cosmological constraints from double-source-plane strong lensing systems in era of CSST

Double source plane strong lensing (DSPL) systems offer a robust, independent probe of cosmological parameters. The Chinese Space Station Telescope (CSST) is expected to discover hundreds of DSPLs, yet the survey modes and system configurations that best enable cosmological inference remain uncertain. To investigate the impact of varying signal-to-noise ratios (SNR) and Einstein radius ratios of DSPLs (denoted as $β^{-1}$ parameters) on cosmographic inference under different CSST survey modes (Wide Field (WF), Deep Field (DF), and Ultra-Deep Field (UDF)), we simulate and model mock lenses with Singular Isothermal Ellipsoid (SIE) mass profiles and Sérsic sources whose image properties are tailored to CSST specifications. Assuming a flat $w$CDM universe with fiducial values $Ω_{\rm m} = 0.30966$ and $w = -1$, and uniform priors of $Ω_{\rm m} \in [0, 1]$ and $w \in [-2, -1/3$), we find that the constraining power on cosmological parameters for a given DSPL system increases significantly with survey depth. For a representative DSPL system with two prominent arcs and a moderate $β^{-1}=1.17$, the constraints on ($w, Ω_{\rm m}$) improve from ($-1.28_{-1.00}^{+0.64}, 0.50_{-0.32}^{+0.28}$) in the WF to ($-1.59_{-0.32}^{+0.63}, 0.42_{-0.06}^{+0.15}$) in the UDF. Furthermore, we find that systems with smaller $β$ values yield tighter cosmographic constraints. We conclude that DSPL systems identified in UDF observations, particularly those with small $β$, are the most promising candidates for early-stage cosmological studies with CSST.

astro-ph.CO↗

arXiv2025

The "Little Dark Dot": Evidence for Self-interacting Dark Matter in the Strong Lens SDSS J0946+1006?

Previous studies, based on precise modeling of a gravitationally lensing image, have identified what may be an extremely compact, dark perturber in the well-known lensing system SDSS J0946+1006 (the "Jackpot"). Its remarkable compactness challenges the standard cold dark matter (CDM) paradigm. In this paper, we explore whether such a compact perturber could be explained as a core-collapse halo described by the self-interacting dark matter (SIDM) model. Using the isothermal Jeans method, we compute the density profiles of core-collapse halos across a range of masses. Our comparison with observations indicates that a core-collapse halo has an inner density profile and mass enclosed within 1 kpc that fit the data well, but only if the halo has a total mass $\sim10^{11}~{\rm M_{\odot}}$. While a halo of this mass should host a detectable galaxy, the current observational upper limit on the perturber's luminosity remains uncertain. Resolving whether or not the data support the presence of a core-collapse SIDM halo therefore requires future deep observations to measure its luminosity.

astro-ph.GA↗

arXiv2025

fiDrizzle-MU: A Fast Iterative Drizzle with Multiplicative Updates

We propose fiDrizzleMU, an algorithm for co-adding exposures via iterative multiplicative updates, replacing the additive correction framework. This method achieves superior anti-aliasing and noise reduction in stacked images. When applied to James Webb Space Telescope data, the fiDrizzleMU algorithm reconstructs a gravitational lensing candidate that was significantly blurred by the pipeline's resampling process. This enables the accurate recovery of faint and extended structures in high-resolution astronomical imaging.

astro-ph.IM↗

arXiv2025

CSST Strong Lensing Preparation: Cosmological Constraints Forecast from CSST Galaxy-Scale Strong Lensing

Strong gravitational lensing by galaxies is a powerful tool for studying cosmology and galaxy structure. The China Space Station Telescope (CSST) will revolutionize this field by discovering up to $\sim$100,000 galaxy-scale strong lenses, a huge increase over current samples. To harness the statistical power of this vast dataset, we forecast its cosmological constraining power using the gravitational-dynamical mass combination method. We create a realistic simulated lens sample and test how uncertainties in redshift and velocity dispersion measurements affect results under ideal, optimistic, and pessimistic scenarios. We find that increasing the sample size from 100 to 10,000 systems dramatically improves precision: in the $Λ$CDM model, the uncertainty on the matter density parameter, $Ω_m$, drops from 0.2 to 0.01; in the $w$CDM model, the uncertainty on the dark energy equation of state, $w$, decreases from 0.3 to 0.04. With 10,000 lenses, our constraints on dark energy are twice as tight as those from the latest DESI BAO measurements. We also compare two parameter estimation techniques -- MultiNest sampling and Bayesian Hierarchical Modeling (BHM). While both achieve similar precision, BHM provides more robust estimates of intrinsic lens parameters, whereas MultiNest is about twice as fast. This work establishes an efficient and scalable framework for cosmological analysis with next-generation strong lensing surveys.

astro-ph.CO↗

arXiv2025

Pixel-level modelling of group-scale strong lens CASSOWARY 19

We present the first high-precision model for the group-scale strong lensing system CASSOWARY 19 (CSWA19), utilising images from the Hubble Space Telescope (HST). Sixteen member galaxies identified via the red-sequence method, and the main halo, all modelled as the dual Pseudo Isothermal Elliptical profile (dPIE), are incorporated into a parametric lens model alongside an external shear field. To model the system, we adopt the PyAutoLens software package, employing a progressive search chain strategy for realizing the transition of source model from multiple Sérsic profiles to a brightness-adaptive pixelization, which uses 1000 pixels in the source plane to reconstruct the background source corresponding to 177,144 image pixels in the image plane. Our results indicate that the total mass within the Einstein radius is $M_{θ_\mathrm{E}}$ $\approx 1.41\times10^{13}$M$_{\odot}$ and the average slope of the total mass density $ρ(r)\propto r^{-γ}$ is $\tildeγ=1.33$ within the effective radius. This slope is shallower than those measured in galaxies and groups but is closer to those of galaxy clusters. In addition, our approach successfully resolves the two merging galaxies in the background source and yields a total magnification of $μ=103.18^{+0.23}_{-0.19}$, which is significantly higher than the outcomes from previous studies of CSWA19. In summary, our research demonstrates the effectiveness of the brightness-adaptive pixelization source reconstruction technique for modelling group-scale strong lensing systems. It can serve as a technical reference for future investigations into pixel-level modelling of the group- and cluster-scale strong lensing systems.

astro-ph.GA↗

arXiv2025

CURLING -- II. Improvement on the $H_{0}$ Inference from Pixelized Cluster Strong Lens Modeling

Strongly lensed supernovae (glSNe) provide a powerful, independent method to measure the Hubble constant, $H_{0}$, through time delays between their multiple images. The accuracy of this measurement depends critically on both the precision of time delay estimation and the robustness of lens modeling. In many current cluster-scale modeling algorithms, all multiple images used for modeling are simplified as point sources to reduce computational costs. In the first paper of the CURLING program, we demonstrated that such a point-like approximation can introduce significant uncertainties and biases in both magnification reconstruction and cosmological inference. In this study, we explore how such simplifications affect $H_0$ measurements from glSNe. We simulate a lensed supernova at $z=1.95$, lensed by a galaxy cluster at $z=0.336$, assuming time delays are measured from LSST-like light curves. The lens model is constructed using JWST-like imaging data, utilizing both Lenstool and a pixelated method developed in CURLING. Under a fiducial cosmology with $H_0=70\rm \ km \ s^{-1}\ Mpc^{-1}$, the Lenstool model yields $H_0=69.91^{+6.27}_{-5.50}\rm \ km\ s^{-1}\ Mpc^{-1}$, whereas the pixelated framework improves the precision by over an order of magnitude, $H_0=70.39^{+0.82}_{-0.60}\rm \ km \ s^{-1}\ Mpc^{-1}$. Our results indicate that in the next-generation observations (e.g., JWST), uncertainties from lens modeling dominate the error budget for $H_0$ inference, emphasizing the importance of incorporating the extended surface brightness of multiple images to fully leverage the potential of glSNe for cosmology.

astro-ph.CO↗

arXiv2025

CSST Strong Lensing Preparation: Fast Modeling of Galaxy-Galaxy Strong Lenses in the Big Data Era

Galaxy-galaxy strong lensing provides a powerful probe of galaxy formation, evolution, and the properties of dark matter and dark energy. However, conventional lens-modeling approaches are computationally expensive and require fine-tuning to avoid local optima, rendering them impractical for the hundreds of thousands of lenses expected from surveys such as Euclid, CSST, and Roman Space Telescopes. To overcome these challenges, we introduce TinyLensGPU, a GPU-accelerated lens-modeling tool that employs XLA-based acceleration with JAX and a neural-network-enhanced nested sampling algorithm, nautilus-sampler. Tests on 1,000 simulated galaxy-galaxy lenses demonstrate that on an RTX 4060 Ti GPU, TinyLensGPU achieves likelihood evaluations approximately 2,000 times faster than traditional methods. Moreover, the nautilus-sampler reduces the number of likelihood evaluations by a factor of 3, decreasing the overall modeling time per lens from several days to roughly 3 minutes. Application to 63 SLACS lenses observed by the Hubble Space Telescope recovers Einstein radii consistent with the literature values (within $\lesssim 5\%$ deviation), which is within known systematic uncertainties. Catastrophic failures, where the sampler becomes trapped in local optima, occur in approximately 5\% of the simulated cases and 10\% of the SLACS sample. We argue that such issues are inherent to automated lens modeling but can be mitigated by incorporating prior knowledge from machine learning techniques. This work thus marks a promising step toward the efficient analysis of strong lenses in the era of big data. The code and data are available online: https://github.com/caoxiaoyue/TinyLensGpu.

astro-ph.IM↗

arXiv2025

Not so dark, not so dense: an alternative explanation for the lensing subhalo in SDSSJ0946+1006

Previous studies of the strong lens system SDSSJ0946+1006 have reported a dark matter subhalo with an unusually high central density, potentially challenging the standard cold dark matter (CDM) paradigm. However, these analyses assumed the subhalo to be completely dark, neglecting the possibility that it may host a faint galaxy. In this work, we revisit the lensing analysis of SDSSJ0946+1006, explicitly modelling the subhalo as a luminous satellite. Incorporating light from the perturber broadens the range of allowed subhalo properties, revealing solutions with significantly lower central densities that are consistent with CDM expectations. The inferred luminosity of the satellite also aligns with predictions from hydrodynamical simulations. While high-concentration subhaloes remain allowed, they are no longer statistically preferred. The luminous subhalo model yields a better fit to the data, while also offering a more plausible explanation that is in line with theoretical expectations. We validate our methodology using mock data, demonstrating that neglecting subhalo light can lead to inferred mass distributions that are artificially compact.

astro-ph.CO↗

arXiv2025

Galaxy Mass Modelling from Multi-Wavelength JWST Strong Lens Analysis: Dark Matter Substructure, Angular Mass Complexity, or Both?

We analyze two galaxy-scale strong gravitational lenses, SPT0418-47 and SPT2147-50, using JWST NIRCam imaging across multiple filters. To account for angular complexity in the lens mass distribution, we introduce multipole perturbations with orders $m=1, 3, 4$. Our results show strong evidence for angular mass complexity in SPT2147, with multipole strengths of 0.3-1.7 $\%$ for $m=3, 4$ and 2.4-9.5 $\%$ for $m=1$, while SPT0418 shows no such preference. We also test lens models that include a dark matter substructure, finding a strong preference for a substructure in SPT2147-50 with a Bayes factor (log-evidence change) of $\sim 60$ when multipoles are not included. Including multipoles reduces the Bayes factor to $\sim 11$, still corresponding to a $5σ$ detection of a subhalo with an NFW mass of $\log_{10}(M_{200}/M_{\odot}) = 10.87\substack{+0.53\\ -0.71}$. While SPT2147-50 may represent the fourth detection of a dark matter substructure in a strong lens, further analysis is needed to confirm that the signal is not due to systematics associated with the lens mass model.

astro-ph.CO↗