arXiv subjects

Louisa Canepa

Publications and source records attributed to Louisa Canepa.

5 recordsLinked to original sources

arXiv2026

Upper Limits on Planet-Induced GHz Radio Emission from Inactive M Dwarfs

Nearby short-period exoplanet systems may produce detectable stellar radio emission due to sub-Alfv\'enic star-planet interaction (SPI), but there are no confirmed cases yet. We targeted five slowly-rotating M dwarfs with transiting terrestrial planets, observing at GHz frequencies throughout their sub-day orbital periods. We did not detect any bursty SPI-like emission, but detected two stars in quiescence: LHS 3844 (unpolarized) and LHS 1678 (circularly polarized). These detections imply persistent magnetic activity at Gyr ages, especially notable for LHS 1678 given its low photometric variability, and can serve as targets for radio transit experiments. Our SPI non-detections may be due to radio beaming geometry, a sub-GHz maximum emission frequency, or undetectable flux density. If the last case applies, then flux density upper limits constrain the exoplanet magnetosphere. GJ 367 b has the tightest constraints -- no extended magnetosphere and an exoplanet field <0.8 G -- although these results depend strongly on unknown stellar wind parameters inferred from stellar rotation period. Due to their small orbital distance, our non-detection systems a priori appear to have more favorable conditions for SPI than most radio-detected SPI candidate systems in the literature, a tension that can either be resolved by favorable wind/geometry conditions on the detected candidates or by a non-SPI (stellar activity) explanation for those candidate detections. Our results favor the approach of sub-GHz searches for radio SPI, especially with the sensitivity of new/upcoming facilities such as MeerKAT, and underscore the need for observational and theoretical work to constrain the magnetized stellar wind parameters.

astro-ph.EP↗

arXiv2025

The dependence of the intracluster light fraction on galaxy cluster properties

We use machine learning to measure the intracluster light (ICL) fractions of 177 galaxy groups and clusters identified from Hyper Suprime-Cam Subaru Strategic Program imaging to explore how the ICL varies with the properties of its host cluster. We study the variation in ICL fraction with host cluster redshift, halo mass, and magnitude gap to investigate how the ICL develops over time, in various cluster environments, and with cluster relaxation. We find that there is a decreasing correlation with redshift (Spearman correlation $r_S=-0.604$, p-value $=9\times10^{-10}$), however this can be plausibly accounted for by including the effects of cosmological surface brightness dimming and the passive aging of stellar populations. There is a weak negative correlation with halo mass ($r_S=-0.330$, p-value $=8\times 10^{-5}$) where ICL fractions are higher in lower halo mass groups than higher halo mass clusters. We also find that there is a marginal positive correlation with magnitude gap ($r_S=0.226$, p-value = 0.01), indicating that relaxed clusters are more likely to host higher ICL fractions. These results are consistent with a scenario where the dominant formation mechanism of the ICL is galaxy-galaxy interactions such as tidal stripping, and demonstrates the capability of the method to easily construct large samples and study large-scale trends in the ICL fraction.

astro-ph.GA↗

arXiv2025

Investigating the imprints of tidal features on simulated galaxy outskirts in LSST-like mock observations

Tidal features provide signatures of recent galaxy mergers, offering insights into the role of mergers in galaxy evolution. The Vera C. Rubin Observatory's upcoming Legacy Survey of Space and Time (LSST) will allow for an unprecedented study of tidal features around millions of galaxies. We use mock images of galaxies at $z\sim0$ ($z\sim0.2$ for \textsc{NewHorizon}) from \textsc{NewHorizon}, \textsc{eagle}, \textsc{IllustrisTNG}, and \textsc{Magneticum Pathfinder} simulations to predict the properties of tidal features in LSST-like images. We find that tidal features are more prevalent around blue galaxies with intrinsic colours $(g-i)\leq0.5$, compared to redder ones, at fixed stellar mass. This trend correlates with elevated specific star formation rates ($\mathrm{sSFR}>10^{-10}\mathrm{\:yr}^{-1}$), suggesting that merger-induced star formation contributes to the bluer colours. Tidal feature hosts in the red sequence appear to exhibit colour profiles offset to bluer colours for galaxies with stellar masses $10^{10}<M_{\star\mathrm{,\:30\:pkpc}}/\mathrm{M}_\odot<10^{11}$, similarly blue cloud tidal feature host galaxies appear to have their colour profiles offset to bluer colours for $10^{9.5}<M_{\star\mathrm{,\:30\:pkpc}}/\mathrm{M}_\odot<10^{10.5}$. However, the differences in colour profiles in either the red sequence or the blue cloud are not statistically robust and larger samples are needed to test if these differences are real. The predictions across the simulations are quantitatively distinct; therefore, LSST observations will allow us to further constrain the differences between different subgrid physics models.

astro-ph.GA↗

arXiv2025

Measuring the intracluster light fraction with machine learning

The intracluster light (ICL) is an important tracer of a galaxy cluster's history and past interactions. However, only small samples have been studied to date due to its very low surface brightness and the heavy manual involvement required for the majority of measurement algorithms. Upcoming large imaging surveys such as the Vera C. Rubin Observatory's Legacy Survey of Space and Time are expected to vastly expand available samples of deep cluster images. However, to process this increased amount of data, we need faster, fully automated methods to streamline the measurement process. This paper presents a machine learning model designed to automatically measure the ICL fraction in large samples of images, with no manual preprocessing required. We train the fully supervised model on a training dataset of 50,000 images with injected artificial ICL profiles. We then transfer its learning onto real data by fine-tuning with a sample of 101 real clusters with their ICL fraction measured manually using the surface brightness threshold method. With this process, the model is able to effectively learn the task and then adapt its learning to real cluster images. Our model can be directly applied to Hyper Suprime-Cam images, processing up to 500 images in a matter of seconds on a single GPU, or fine-tuned for other imaging surveys such as LSST, with the fine-tuning process taking just 3 minutes. The model could also be retrained to match other ICL measurement methods. Our model and the code for training it is made available on GitHub.

astro-ph.GA↗

arXiv2023

Visual Question Answering in the Medical Domain

Medical visual question answering (Med-VQA) is a machine learning task that aims to create a system that can answer natural language questions based on given medical images. Although there has been rapid progress on the general VQA task, less progress has been made on Med-VQA due to the lack of large-scale annotated datasets. In this paper, we present domain-specific pre-training strategies, including a novel contrastive learning pretraining method, to mitigate the problem of small datasets for the Med-VQA task. We find that the model benefits from components that use fewer parameters. We also evaluate and discuss the model's visual reasoning using evidence verification techniques. Our proposed model obtained an accuracy of 60% on the VQA-Med 2019 test set, giving comparable results to other state-of-the-art Med-VQA models.

cs.CV↗