Deep Learning for CMB Foreground Removal and Beam Deconvolution: A U-Net GAN Approach

2026-05-10T13:52:49Z

Extracting cosmological information from microwave sky observations requires accurate estimation of the underlying Cosmic Microwave Background (CMB) by removing foreground contamination, instrumental noise, and the effects of beam convolution. In this work, we develop a machine learning-based approach for CMB reconstruction using a generative adversarial network (GAN) architecture, where the generator is modeled as a U-Net-based convolutional neural network. To train the network, we generate realistic microwave sky maps by simulating Planck-like observations: scanning HEALPix-simulated skies with real Planck beam profile, actual scan patterns, and anisotropic noise consistent with Planck data. Our method achieves high-fidelity reconstruction, with the difference between the input and recovered maps being less than $1\%$ (approximately $2μ\mathrm{K}$ for temperature and less than $0.5μ\mathrm{K}$ for polarization) outside the Galactic region. Even within the Galactic plane, the reconstruction error stays below $2$-$3\%$ for temperature maps across most regions, and is even smaller for polarization, apart from a few isolated pixels.. Most importantly, we demonstrate, for the first time, that a GAN-based method can effectively correct for foreground contamination, the systematic effects of non-circular beams and the asymmetric Planck scan pattern for both T and E-mode skymaps. Our results demonstrate the effectiveness of our method for robust and accurate recovery of the CMB signal, even in the presence of strong astrophysical foregrounds and instrumental systematics.

Long-Term Dynamical Evolution and Ejection of Near-Earth Asteroids

2026-05-09T21:30:05Z

Long-term integrations of asteroid orbits with high-accuracy numerical integrators are essential for understanding dynamical evolution and ejection from the Solar System, but are computationally expensive. Here, we investigate the dynamical behaviour of asteroids and explore machine-learning (ML) and deep-learning (DL) approaches as efficient, scalable alternatives for classifying long-term dynamical outcomes. While the ML classifiers are trained on initial orbital elements, the convolutional neural network is trained on recurrence plots derived from short-period numerical integrations generated with the MERCURY integrator. Ensemble tree models perform strongly on the ephemeris input, and the neural network captures temporal signatures of chaotic motion with comparable or slightly improved accuracy. Backward integrations reveal partial overlap between forward- and reverse-ejected sets, illustrating time-asymmetric behaviour in chaotic regions; these backward results are interpreted only as diagnostic probes rather than reconstructions of past histories. Non-ejected asteroids largely correspond to known dynamical groups, underscoring the constraining role of initial orbital configuration. These methods provide scalable frameworks to complement numerical integrations and inform prioritisation for detailed long-term dynamical studies, with implications for planetary-defence analyses.

A new way to explain the 511 keV signal from the center of the Galaxy and experimental search for small hydrogen

2026-05-09T17:44:28Z

The first detected gamma-ray line originating from outside the solar system is the 511 keV emission from the center of our Galaxy. The widely accepted explanation attributes this signal to electron-positron annihilation. However, despite over 30 years of extensive theoretical and observational research, the primary sources of these positrons remain unidentified. In this paper, we propose an alternative explanation: the observed signal arises from atomic transitions involving a small hydrogen atom, where an electron is captured into a tightly bound orbit around a proton. We review the current status of experimental searches for small hydrogen, both in astrophysical data and laboratory experiments, and propose new methods for its direct detection in the lab. Additionally, we explore whether small hydrogen could be a candidate for dark matter.

The new generation lunar gravitational wave detectors: sky map resolution and joint analysis

2026-05-09T12:38:18Z

Lunar-based gravitational-wave interferometry is a fascinating endeavor, and was proposed as a promising approach to bridge the observational gap between space-borne and ground-based detectors. In this work, we adopt the Fisher-matrix method to examine the angular-resolution performance of the newly proposed Crater Interferometry Gravitational-wave Observatory (CIGO) on the lunar crater rim near the north pole, together with TianQin and LISA, for monochromatic sources in the 0.1-10 Hz band. We find that above 0.1 Hz, CIGO achieves better localization accuracy than the other two space-based missions and dominates the combined detector network's performance, provided that lunar noise mitigation is achieved in the 0.1-2.87 Hz frequency range. We further explore an upgraded Tetrahedron configuration, TCIGO, with a fourth station at the bottom of a crater, which forms a regular tetrahedral constellation on the lunar surface. The result shows that TCIGO yields a five-fold improvement in angular-resolution capability over CIGO and gets better sky coverage across the target frequency band.

BOCOSUR: An all sky network for fireball detection in Uruguay

2026-05-09T12:36:35Z

Over the past couple of decades, several networks for the automatic detection of fireballs have been deployed. Their primary scientific goal is to facilitate the rapid recovery of meteorites, determine their pre-atmospheric orbits, and look for possible dynamic links with parent bodies. The Bocosur network is a contribution to the global deployment of automated fireball networks and to the increase of the number of recoverable meteorite falls. It is located in Uruguay, South America (Lat: -30$^{\circ}$ to -35$^{\circ}$). Its main scientific goal is the detection of fireballs of asteroidal origin, massive enough to produce meteorites, and also to inspire secondary-level students and teachers through their involvement in this citizen-science oriented project. The deployment of this network started in 2019, and was completed in March, 2023, when we installed 20 stations separated $\sim 120$ km, covering an area of $\sim 180,000$ km$^2$. During this period of time, one major technological upgrade was made when we migrated from a well-known camera to a higher-resolution, more sensitive system. We were able to build a completely autonomous system at an affordable cost that can be replicated in all the stations. A comparison between the astrometric and photometric performance of these two detection systems is reported. Also, a photometric methodology for estimating the brightness of very bright fireballs is presented and validated against the known magnitudes of Jupiter and the full Moon. We obtain mean residuals of the astrometric reduction of $\sim$5', and the discrepancy between the obtained brightness of Jupiter and the Moon average to 0.18 and 1.2 magnitudes, respectively. Results on the processing of a very bright (M$_{peak}\sim$-9.0 mag) fireball detected in four stations are also presented.

Observations of the 2023 February 27 fireball in northern Sweden using the auroral imaging system ALIS_4D

2026-05-09T09:39:31Z

On 2023 February 27 at 18:15:55.77 UT, a bright fireball streaked across the sky above northern Sweden. The event offered a valuable opportunity to study the phenomenon using an optical system primarily designed for auroral studies, the Auroral Large Imaging System (ALIS_4D), that captured the event. In this study we show the capability of ALIS_4D to perform observations in support of meteor event analysis. We estimated the trajectory from the recorded data and computed the orbit. In addition, we investigated the origin of the meteoroid searching for its parent body. Fitting the analytical ablation model known as $α$-$β$ to the trajectory as well as incorporating local wind-field data in Monte-Carlo dark-flight simulations, strewn-fields were computed and physical properties of the meteoroid were estimated. Trajectory analyses delineate a strewn field along the border between Kiruna and Gällivare in northern Sweden. Our findings indicate that the meteoroid's parent body was likely an Apollo family object. We performed an orbital similarity analysis to identify candidate parent bodies of the fireball. Our simulations suggest that close approaches with Earth could have disrupted the meteoroid's orbit, placing it on a collision course.

An approximate Kappa generator for particle simulations

2026-05-09T09:18:55Z

A random number generator for the Kappa velocity distribution in particle simulations is proposed. Approximating the cumulative distribution function with the q-exponential function, an inverse transform procedure is constructed. The proposed method provides practically accurate results, in particular for k<4. It runs fast on graphics processing units (GPUs). The derivation, numerical validation, and relevance to GPU execution models are discussed.

Earth Science Foundation Models: From Perception to Reasoning and Discovery

2026-05-09T08:34:30Z

Large foundation models (FMs) are transforming Earth science by integrating heterogeneous multimodal data, such as multi-platform imagery, gridded reanalysis data, diverse geophysical and geochemical observations, and domain-specific text, to support tasks ranging from basic perception to advanced scientific discovery. This paper provides a unified review of Earth science foundation models (Earth FMs) through two complementary dimensions: depth, which traces the evolution of model capabilities from perception to multimodal reasoning and agentic scientific workflows, and breadth, which summarizes their expanding applications across the atmosphere, hydrosphere, lithosphere, biosphere, anthroposphere, and cryosphere, as well as coupled Earth system processes. Using this framework, we review representative multimodal Earth foundation models and compile more than 200 datasets and benchmarks spanning diverse Earth science tasks and modalities. We further discuss key challenges in multimodal data heterogeneity, scientific reliability and continual updating, scalability and sustainability, and the transition from foundation models to agentic and embodied Earth intelligence, and outline future directions toward more integrated, trustworthy, and actionable AI Earth scientists. Overall, this paper offers a structured roadmap for understanding the development of Earth foundation models from both capability depth and application breadth.

New Methods for Offline GstLAL Analyses

2026-05-09T06:37:12Z

In this work, we present new methods implemented in the GstLAL offline gravitational wave search. These include a technique to reuse the matched filtering data products from a GstLAL online analysis, which hugely reduces the time and computational resources required to obtain offline results; a technique to combine these results with a separate search for heavier black hole mergers, enabling detections from a larger set of gravitational wave sources; changes to the likelihood ratio which increases the sensitivity of the analysis; and two separate changes to the background estimation, allowing more precise significance estimation of gravitational wave candidates. Some of these methods increase the sensitivity of the analysis, whereas others correct previous mis-estimations of sensitivity by eliminating false positives. These methods have been adopted for GstLAL's offline results during the fourth observing run of LIGO, Virgo, and KAGRA (O4). To test these new methods, we perform an offline analysis over one chunk of O3 data, lasting from May 12 19:36:42 UTC 2019 to May 21 14:45:08 UTC 2019, and compare it with previous GstLAL results over the same period of time. We show that cumulatively these methods afford around a 50% - 100% increase in sensitivity in the highest mass space, while simultaneously increasing the reliability of results, and making them more reusable and computationally cheaper.

How Many Times Should We Matched Filter Gravitational Wave Data? A Comparison of GstLAL's Online and Offline Performance

2026-05-09T05:27:19Z

Searches for gravitational waves from compact binary coalescences employ a process called matched filtering, in which gravitational wave strain data is cross-correlated against a bank of waveform templates. Data from every observing run of the LIGO, Virgo, and KAGRA collaboration is typically analyzed in this way twice, first in a low-latency mode in which gravitational wave candidates are identified in near-real time, and later in a high-latency mode. Such high-latency analyses have traditionally been considered more sensitive, since background data from the full observing run is available for assigning significance to all candidates, as well as more robust, since they do not need to worry about keeping up with live data. In this work, we present a novel technique to use the matched filtering data products from a low-latency analysis and re-process them by assigning significances in a high-latency way, effectively removing the need to perform matched filtering a second time. To demonstrate the efficacy of our method, we analyze 38 days of LIGO and Virgo data from the third observing run (O3) using the GstLAL pipeline, and show that our method is as sensitive and reliable as a traditional high-latency analysis. Since matched filtering represents the vast majority of computing time for a traditional analysis, our method greatly reduces the time and computational burden required to produce the same results as a traditional high-latency analysis. Consequently, it has already been adopted by GstLAL for the fourth observing run (O4) of the LIGO, Virgo, and KAGRA collaboration.

Quasiparticle Quality Factors in Superconducting Resonators: Effects of Bath Temperature and Readout Power

2026-05-09T01:15:14Z

The performance of superconducting resonators underpins a wide range of modern quantum technologies, yet their quality factor often deviates at low temperatures from standard Mattis-Bardeen predictions. This discrepancy is often attributed to nonthermal quasiparticles generated by microwave readout power, which limits the sensitivity of superconducting devices. We present a macroscopic model based on modified Rothwarf-Taylor equations that incorporates a power-dependent phonon generation term, providing an explicit relationship between quality factor, bath temperature and readout power. The model shows excellent agreement with temperature sweep measurements of NbN microstrip resonators with \b{eta}-Ta terminations over a wide dynamic range of readout power levels, accurately capturing the transition between thermally-dominated and microwave-induced loss regimes. This framework provides a predictive tool for optimizing superconducting resonators and advancing the design of high-Q devices for quantum sensing and quantum information processing.

Bayesian Doppler Imaging: Simultaneous Inference of Surface Maps and Geometric Parameters

2026-05-08T23:10:07Z

We present a fully Bayesian, pixel-based Doppler imaging framework that enables the simultaneous inference of surface brightness maps and geometric parameters, including the inclination $i$ and equatorial rotation velocity $v_{\mathrm{rot}}$, from high-resolution spectral time series. We treat the inference as a Bayesian linear inverse problem conditioned on nonlinear geometric parameters. The surface map is modeled as a Gaussian Process prior over pixel intensities, introducing a characteristic spatial scale that sets the map resolution. This allows analytical marginalization of the linear coefficients and efficient sampling of the nonlinear parameters with Hamiltonian Monte Carlo. {Validation with synthetic data demonstrates that our method recovers the longitudes of large-scale surface inhomogeneities and constrains $v_{\mathrm{rot}}$ and $i$ under the adopted model assumptions, while also revealing the limited latitudinal sensitivity intrinsic to Doppler imaging.} We applied this framework to high-resolution VLT/CRIRES observations of the brown dwarf Luhman 16B. Our analysis reveals a large-scale dark region at mid-latitudes, consistent with previous studies but now with spatially resolved uncertainty estimates. Furthermore, we successfully constrained the geometric parameters without fixing $v_{\mathrm{rot}}\sin i$ or $i$ to literature values, deriving an inclination of $i = 61.0_{-12.3}^{+14.3}$ degrees and an equatorial rotation velocity of $v_{\mathrm{rot}} = 31.2_{-3.1}^{+5.3}~\mathrm{km\,s^{-1}}$. These results indicate a radius broadly consistent with evolutionary models and suggest a possible spin-axis misalignment under the assumption of comparable equatorial rotation velocities for the two components. Our code is publicly available under the MIT license.

Wideband RFI Monitor Requirements, Design, and Commissioning at DRAO

2026-05-08T20:55:23Z

In this paper, we introduce the radio frequency interference monitor deployed at the Dominion Radio Astrophysical Observatory. It provides 2 GHz of instantaneous bandwidth, supporting channel bandwidths as fine as ~100 Hz for 1 s integrations, or integration times as low as ~50 ms for the standard 3.33 kHz channel bandwidth. After operating as a prototype instrument for several years, the monitor was commissioned to improve the calibration method, analog section temperature, and gain stability. It now operates both as a transient detector and as a long-term radio environment characterization tool. We introduce novel applications for the monitor and derive a new method for calculating the effect of gain drift on integrated data.

Robust 1-norm periodograms for analysis of noisy non-Gaussian time series with irregular cadences: Application to VLBI astrometry of quasars

2026-05-08T19:53:40Z

Astronomical time series often have non-uniform sampling in time, or irregular cadences, with long gaps separating clusters of observations. Some of these data sets are also explicitly non-Gaussian with respect to the expected model fit, or the simple mean. The standard Lomb-Scargle periodogram is based on the least squares solution for a set of test periods and, therefore, is easily corrupted by a subset of statistical outliers or an intrinsically non-Gaussian population. It can produce completely misleading results for heavy-tailed distribution of residuals. We propose a robust 1-norm periodogram technique, which is based on the principles of robust statistical estimation. This technique can be implemented in weighted or unweighted options. The method is described in detail and compared with the classical least squares periodogram on a set of astrometric VLBI measurements of the ICRF quasar IERS B0642+449. It is uniformly applied to a collection of 259 ICRF3 quasars each with more than 200 epoch VLBI measurements, resulting in a list of 49 objects with quasi-periodic position changes above the $3σ$ level, which warrant further investigation.

A SPHEREx Pipeline and Spectral Library for Ultracool Dwarfs

2026-05-08T18:17:06Z

We present a Python spectrophotometry extraction tool tailored for fast-moving point sources detected in the SPHEREx mission, and use it to construct a set of 0.75-5.0 $μ$m low-resolution ($λ/Δλ\sim 50$) spectrophotometry data products based on the SPHEREx Quick Release 2 (QR2) for a set of 6003 L0-Y1 ultracool dwarfs: 2050 known ultracool dwarfs, 3008 known photometric ultracool dwarf candidates, and 947 newly identified ultracool dwarfs. This work more than doubles the number of ultracool dwarfs with spectroscopy, from 3449 to 7402. We provide SPHEREx templates for each spectral subtype and a set of tools to assign automated spectral types. The QR2 data release generates spectrophotometry with an average signal-to-noise per spectral channel above $\sim$10 for most objects with WISE W2 magnitudes of 14.0 mag and brighter. The compiled data set is made available publicly at https://mocadb.ca, where new spectral compilations from future data releases will also be made available as they are published. These new data provide a significant increase in the number of substellar objects for which the 2.4-5.0 $μ$m window is now accessible, making it possible to probe important molecular chemistry of key CNOS-bearing species for the coolest brown dwarfs. We flag 2668 ultracool dwarfs as candidate young brown dwarfs, 250 as candidate subdwarfs, and 865 as possibly otherwise peculiar for future investigation. The SPIFF library presented here opens the doors to efficient confirmation of candidate substellar objects and follow-up studies of population-level atmospheric properties of cold brown dwarfs.