Search

In classic time-dependent 1D accretion disk models, the inner radiation pressure dominated regime is viscously unstable. However, late-time observations of accretion disks formed in tidal disruption events (TDEs) do not exhibit evidence of such instabilities. The common theoretical response is to modify the viscosity parametrization, but typically used viscosity parametrization are generally ad hoc. In this study, we take a different approach, and investigate a time-dependent 1D $\alpha$-disk model in which the pressure is dominated by magnetic fields rather than photons. We compare the time evolution of thermally stable, strongly magnetized TDE disks to the simpler linear viscosity model. We find that the light curves of magnetized disks evolve as $L_{\rm UV}\propto t^{-5/6}$ for decades to centuries, and that this same evolution can be reproduced by the linear viscosity model for specific parameter choices. Additionally, we show that TDEs remain UV-bright for many years, suggesting we could possibly find fossil TDEs decades after their bursts. We estimate that ULTRASAT could detect hundreds of such events, providing an opportunity to study late-stage TDE physics and supermassive black hole (SMBH) properties. Finally, we explore the connection between TDE disks and quasi-periodic eruptions (QPEs) suggested by recent observations. One theoretical explanation involves TDE disks expanding to interact with extreme mass ratio inspirals (EMRIs), which produce X-ray flares as the EMRI passes through the disk. Our estimates indicate that magnetized TDE disks should exhibit QPEs earlier than those observed in AT2019qiz, suggesting that the QPEs may have begun before their first detection., Comment: Reader comments very welcome

The ratio between the stellar mass of a galaxy, $M_{*}$, and that of its central supermassive black hole (SMBH), $M_\bullet$, the ``Magorrian'' relationship, traces their coevolution. JWST observations have suggested significant evolution in $M_\bullet/M_{*}$ relative to local scaling relationships both in low-mass galaxies and in quasars at z $\ge$ 4. We test this possibility by (1) determining the preferred $M_\bullet/M_{*}$ scaling relation among those proposed locally; and (2) providing uniform host galaxy stellar mass estimates. These steps reduce the prominence of the reported evolution. We then apply Monte Carlo simulations to account for observational biases. We still find a significant increase over the local scaling relation in $M_\bullet/M_{*}$ for z $\ge$ 4 SMBHs in very low-mass galaxies ($\log(M_*/M_{\odot})<10$). However, similarly high values of $M_\bullet/M_{*}$ are also found in low mass galaxies at $z \sim$ 0.5 to 3 that may be common at cosmic noon. Nonetheless, galaxies with similar behavior are rare locally and not accounted for in the local scaling relations. In contrast, z $\sim$ 6 quasars can have $M_\bullet/M_{*}$ well above the local relation value, but they can be explained as extreme cases still within the scaling relation for their higher mass host galaxies. Black holes in some of them and in the low-mass systems may be undergoing very high accretion episodes that result in high $M_\bullet/M_{*}$ but that will be followed by quiescent periods when growth of the host drives the systems toward more typical $M_\bullet/M_{*}$ values., Comment: 20 pages, 7 figures, resubmitted to ApJ based on reviewer report

Understanding the ages of stars is crucial for unraveling the formation history and evolution of our Galaxy. Traditional methods for estimating stellar ages from spectroscopic data often struggle with providing appropriate uncertainty estimations and are severely constrained by the parameter space. In this work, we introduce a new approach using normalizing flows, a type of deep generative model, to estimate stellar ages for evolved stars with improved accuracy and robust uncertainty characterization. The model is trained on stellar masses for evolved stars derived from asteroseismology and predicts the relationship between the carbon and nitrogen abundances of a given star and its age. Unlike standard neural network techniques, normalizing flows enable the recovery of full likelihood distributions for individual stellar ages, offering a richer and more informative perspective on uncertainties. Our method yields age estimations for 378,720 evolved stars and achieves a typical absolute age uncertainty of approximately 2 Gyr. By intrinsically accounting for the coverage and density of the training data, our model ensures that the resulting uncertainties reflect both the inherent noise in the data and the completeness of the sampled parameter space. Applying this method to data from the SDSS-V Milky Way Mapper, we have produced the largest stellar age catalog for evolved stars to date., Comment: Submitted to AJ. Age data to be released with SDSS DR19

Long-horizon tasks in robotic manipulation present significant challenges in reinforcement learning (RL) due to the difficulty of designing dense reward functions and effectively exploring the expansive state-action space. However, despite a lack of dense rewards, these tasks often have a multi-stage structure, which can be leveraged to decompose the overall objective into manageable subgoals. In this work, we propose DEMO3, a framework that exploits this structure for efficient learning from visual inputs. Specifically, our approach incorporates multi-stage dense reward learning, a bi-phasic training scheme, and world model learning into a carefully designed demonstration-augmented RL framework that strongly mitigates the challenge of exploration in long-horizon tasks. Our evaluations demonstrate that our method improves data-efficiency by an average of 40% and by 70% on particularly difficult tasks compared to state-of-the-art approaches. We validate this across 16 sparse-reward tasks spanning four domains, including challenging humanoid visual control tasks using as few as five demonstrations., Comment: Project page can be found at https://adrialopezescoriza.github.io/demo3/

Scaling up the model size and computation has brought consistent performance improvements in supervised learning. However, this lesson often fails to apply to reinforcement learning (RL) because training the model on non-stationary data easily leads to overfitting and unstable optimization. In response, we introduce SimbaV2, a novel RL architecture designed to stabilize optimization by (i) constraining the growth of weight and feature norm by hyperspherical normalization; and (ii) using a distributional value estimation with reward scaling to maintain stable gradients under varying reward magnitudes. Using the soft actor-critic as a base algorithm, SimbaV2 scales up effectively with larger models and greater compute, achieving state-of-the-art performance on 57 continuous control tasks across 4 domains. The code is available at https://dojeon-ai.github.io/SimbaV2., Comment: 50 pages. Preprint

Topological spin textures, such as magnetic skyrmions, hold great promise for spintronics applications. However, most known skyrmion systems are restricted to a topological charge of one and require an external magnetic field for stabilization. Here, we report the discovery of a skyrmion crystal with a topological number of two in an insulating van der Waals magnet NiI2. We unveil the static and dynamic magnetic correlations across three temperature-driven magnetic phases using neutron scattering measurements, simulations, and model optimisation. By employing a minimal Kitaev-Heisenberg Hamiltonian, we reproduce the experimentally observed excitations and confirm the emergence of the higher-order skyrmion crystal through Monte Carlo simulations. Remarkably, this exotic phase arises without the need for a magnetic field, presenting a novel platform for controlling complex spin textures in an accessible material. These findings establish NiI2 as a pivotal material, bridging fundamental quantum magnetism with advancements in skyrmionics and multiferroic technologies., Comment: 12 pages, 4 figures

Advances in machine learning and the increasing availability of high-dimensional data have led to the proliferation of social science research that uses the predictions of machine learning models as proxies for measures of human activity or environmental outcomes. However, prediction errors from machine learning models can lead to bias in the estimates of regression coefficients. In this paper, we show how this bias can arise, propose a test for detecting bias, and demonstrate the use of an adversarial machine learning algorithm in order to de-bias predictions. These methods are applicable to any setting where machine-learned predictions are the dependent variable in a regression. We conduct simulations and empirical exercises using ground truth and satellite data on forest cover in Africa. Using the predictions from a naive machine learning model leads to biased parameter estimates, while the predictions from the adversarial model recover the true coefficients., Comment: 12 pages, 8 figures

Double perovskites containing heavy transition metal ions are an important family of compounds for the study of the interplay between electron correlation and spin-orbit coupling. Here, by combining magnetic susceptibility, heat capacity, and neutron scattering measurements, we investigate the dipolar and quadrupolar correlations in two prototype rhenium-based double perovskite compounds, Ba$_2$YReO$_6$ and Ba$_2$ScReO$_6$. A type-I dipolar antiferromagnetic ground state with a propagation vector $\mathbf{q} = (0, 0, 1)$ is observed in both compounds. At temperatures above the magnetic transitions, a quadrupolar ordered phase is identified. Weak spin excitations, which are gapped at low temperatures and softened in the correlated paramagnetic phase, are explained using a minimal model that considers both the dipolar and quadrupolar interactions. At larger wavevectors, we observe dominant phonon excitations that are well described by density functional calculations., Comment: 13 pages, 11 figures

Tunable ultrashort laser pulses across the near-ultraviolet to near-infrared with high peak power are crucial for wide-ranging applications in science and industry. Resonant dispersive-wave emission in gas-filled hollow-core fibers is a well-established technique for generating tunable ultrashort pulses from the vacuum ultraviolet to the near-infrared. However, previous demonstrations have relied on complex and expensive laser systems to provide the necessary energetic ultrashort pump pulses. Recent advances in fiber laser technology, particularly gain-managed nonlinear amplification, offer a promising alternative pump source. In this work, we combine gain-managed nonlinear amplification with resonant dispersive-wave emission to demonstrate a compact and tunable source of sub-20 fs pulses at 4.8 MHz. We achieve a tunable output spanning from 400 nm to beyond 700 nm, with energy up to 39 nJ, pulse duration down to 13 fs, and peak power exceeding 2 MW. This compact and efficient laser source opens new avenues for deploying resonant dispersive-wave-based technologies for broader scientific and industrial applications.

Brown dwarfs with known physical properties (e.g., age and mass) are essential for constraining models of the formation and evolution of substellar objects. We present new high-contrast imaging observations of the circumbinary brown dwarf HII 1348B $\mathord{-}$ one of the few known substellar companions in the Pleiades cluster. We observed the system in the infrared (IR) $L^\prime$ band with the Large Binocular Telescope Interferometer (LBTI) in dual-aperture direct imaging mode (i.e., with the two telescope apertures used separately) on 2019 September 18. The observations attained a high signal-to-noise ratio ($\mathrm{SNR} > 150$) photometric detection and relative astrometry with uncertainties of ${\sim}5 \ \mathrm{mas}$. This work presents the first model of the companion's orbital motion using relative astrometry from five epochs across a total baseline of 23 years. Orbital fits to the compiled data show the companion's semimajor axis to be $a = 140 \substack{+130 \\ -30} \ \mathrm{au}$ with an eccentricity of $e = 0.78 \substack{+0.12 \\ -0.29}$. We infer that HII 1348B has a mass of $60\mathord{-}63 \pm 2 \ \mathrm{M_J}$ from comparison to brown dwarf evolutionary models given the well-constrained distance and age of the Pleiades. No other objects were detected in the HII 1348 system, and through synthetic planet injection and retrievals we establish detection limits at a cluster age of $112 \pm 5$ Myr down to ${\sim}10\mathord{-}30 \ \mathrm{M_J}$ for companions with projected separations of $21.5\mathord{-}280 \ \text{au}$. With this work, HII 1348B becomes the second directly imaged substellar companion in the Pleiades with measured orbital motion after HD 23514B., Comment: AASTeX v6.3.1, 29 pages with 8 figures, accepted for publication in The Astronomical Journal

We report the characterisation of anti-resonant hollow core optical fibres guiding at least 50 spatial modes in the infrared. Their propagation losses were measured to be between 0.1 and 0.2 dB/m from 1000 to 1500 nm wavelength, with bend losses of less than 3 dB/turn for bend radii of 7.5 cm despite core radii greater than 60 times the guided wavelengths.

Magnetic heat capacity measurements of a high-quality single crystal of the dipole-octupole pyrochlore Ce$_2$Hf$_2$O$_7$ down to a temperature of $T = 0.02$ K are reported. These show a two-peaked structure, with a Schottky-like peak at $T_1 \sim 0.065$ K, similar to what is observed in its sister Ce-pyrochlores Ce$_2$Zr$_2$O$_7$ and Ce$_2$Sn$_2$O$_7$. However a second, sharper peak is observed at $T_2 \sim 0.025$ K, which signifies the entrance to its ground state, as even the most abrupt low-temperature extrapolation to $C_P=0$ at $T = 0$ K gives a full accounting of $R\ln(2)$ in entropy, associated with the well isolated pseudospin-1/2 doublet for Ce$^{3+}$ in this environment. The ground state could be conventionally ordered, although theory predicts a much larger anomaly in $C_P$, at much higher temperatures than the measured $T_2$, for expectations from an all-in all-out ground state of the nearest-neighbor XYZ Hamiltonian for Ce$_2$Hf$_2$O$_7$. The sharp low-temperature peak could also signify a cross-over from a classical spin liquid to a quantum spin liquid (QSL). The diffuse magnetic neutron scattering observed from Ce$_2$Hf$_2$O$_7$ at low temperatures between $T_2$ and $T_1$ resembles that observed from Ce$_2$Zr$_2$O$_7$, which is well established as a $\pi$-flux quantum spin ice (QSI)., Comment: Main Text (5 pages, 4 figures), Supplemental Material (14 pages, 17 figures)

We describe the development of a one-credit course to promote AI literacy at The University of Texas at Austin. In response to a call for the rapid deployment of class to serve a broad audience in Fall of 2023, we designed a 14-week seminar-style course that incorporated an interdisciplinary group of speakers who lectured on topics ranging from the fundamentals of AI to societal concerns including disinformation and employment. University students, faculty, and staff, and even community members outside of the University, were invited to enroll in this online offering: The Essentials of AI for Life and Society. We collected feedback from course participants through weekly reflections and a final survey. Satisfyingly, we found that attendees reported gains in their AI literacy. We sought critical feedback through quantitative and qualitative analysis, which uncovered challenges in designing a course for this general audience. We utilized the course feedback to design a three-credit version of the course that is being offered in Fall of 2024. The lessons we learned and our plans for this new iteration may serve as a guide to instructors designing AI courses for a broad audience., Comment: Accepted to EAAI-25: The 15th Symposium on Educational Advances in Artificial Intelligence, collocated with AAAI-25

Designing a reinforcement learning from human feedback (RLHF) algorithm to approximate a human's unobservable reward function requires assuming, implicitly or explicitly, a model of human preferences. A preference model that poorly describes how humans generate preferences risks learning a poor approximation of the human's reward function. In this paper, we conduct three human studies to asses whether one can influence the expression of real human preferences to more closely conform to a desired preference model. Importantly, our approach does not seek to alter the human's unobserved reward function. Rather, we change how humans use this reward function to generate preferences, such that they better match whatever preference model is assumed by a particular RLHF algorithm. We introduce three interventions: showing humans the quantities that underlie a preference model, which is normally unobservable information derived from the reward function; training people to follow a specific preference model; and modifying the preference elicitation question. All intervention types show significant effects, providing practical tools to improve preference data quality and the resultant alignment of the learned reward functions. Overall we establish a novel research direction in model alignment: designing interfaces and training interventions to increase human conformance with the modeling assumptions of the algorithm that will learn from their input.

While nova eruptions produce some of the most common and dramatic dust formation episodes among astrophysical transients, the demographics of dust-forming novae remain poorly understood. Here, we present a statistical study of dust formation in 40 novae with high-quality optical/IR light curves, quantitatively distinguishing dust-forming from non-dust-forming novae while exploring the properties of the dust events. We find that 50-70% of novae produce dust, significantly higher than previous estimates. Dust-forming novae can be separated from those that do not show dust formation by using the largest redward ($V-K$) colour change from peak visible brightness; ($V-J$) or ($V-H$) offer useful but less sensitive constraints. This makes optical+IR photometry a powerful tool to quantify dust formation in novae. We find that novae detected in GeV $\gamma$-rays by \emph{Fermi}-LAT appear to form dust more often than novae not detected by \emph{Fermi}, implying a possible connection between $\gamma$-ray producing shocks and dust production. We also find that novae that evolve very quickly ($t_2 < 10$ days) are much less likely to form dust, in agreement with previous findings. We confirm a correlation between $t_2$ and the time of the onset of dust formation (which occurs $\sim$1 week--3 months after maximum light), but conclude that it is primarily an observational artifact driven by dust formation determining when a nova drops 2 mag below peak. The significant fraction of novae that form dust make them ideal laboratories in our Galactic backyard to tackle the puzzle of dust formation around explosive transients., Comment: 23 pages, 58 figures, 1 table, submitted to MNRAS

We present multi-wavelength analysis of ZTF23abelseb (AT 2023sva), an optically discovered fast-fading ($\Delta m_r = 2.2$ mag in $\Delta t = 0.74 $ days), luminous ($M_r \sim -30.0$ mag) and red ($g-r = 0.50$ mag) transient at $z = 2.28$ with accompanying luminous radio emission. AT 2023sva does not possess a $\gamma$-ray burst (GRB) counterpart to an isotropic equivalent energy limit of $E_{\rm{\gamma, \, iso}} < 1.6 \times 10^{52}$ erg, determined through searching $\gamma$-ray satellite archives between the last non-detection and first detection, making it the sixth example of an optically-discovered afterglow with a redshift measurement and no detected GRB counterpart. We analyze AT 2023sva's optical, radio, and X-ray observations to characterize the source. From radio analyses, we find the clear presence of strong interstellar scintillation (ISS) 72 days after the initial explosion, allowing us to place constraints on the source's angular size and bulk Lorentz factor. When comparing the source sizes derived from ISS of orphan events to those of the classical GRB population, we find orphan events have statistically smaller source sizes. We also utilize Bayesian techniques to model the multi-wavelength afterglow. Within this framework, we find evidence that AT 2023sva possesses a shallow power-law structured jet viewed slightly off-axis ($\theta_{\rm{v}} = 0.07 \pm 0.02$) just outside of the jet's core opening angle ($\theta_{\rm{c}} = 0.06 \pm 0.02$). We determine this is likely the reason for the lack of a detected GRB counterpart, but also investigate other scenarios. AT 2023sva's evidence for possessing a structured jet stresses the importance of broadening orphan afterglow search strategies to a diverse range of GRB jet angular energy profiles, to maximize the return of future optical surveys., Comment: 22 pages, 14 Figures, Submitted to MNRAS

We investigate the dynamics of close-contact melting (CCM) on gas-trapped hydrophobic surfaces, with specific focus on the effects of geometrical confinement and the liquid-air meniscus below the liquid film. By employing dual-series and perturbation methods, we obtain numerical solutions for the effective slip lengths associated with velocity $\lambda$ and temperature $\lambda_t$ fields, across various values of aspect ratio $\Lambda$ (defined as the ratio of the film thickness $h$ to the structure's periodic length $l$) and gas-liquid fraction $\phi$. Asymptotic solutions of $\lambda$ and $\lambda_t$ for $\Lambda\ll 1$ and $\Lambda \gg 1$ are derived and summarized for different surface structures, interface shapes and $\Lambda$, which reveal a different trend for $\lambda$ and $\Lambda \ll 1$ and the presence of a meniscus. In the context of constant-pressure CCM, our results indicate that transverse-grooves surfaces consistently reduced the heat transfer. However, longitudinal grooves can enhance heat transfer under the effects of confinement and meniscus when $\Lambda \lessapprox 0.1$ and $\phi < 1 - 0.5^{2/3} \approx 0.37$. For gravity-driven CCM, the parameters of $l$ and $\phi$ determine whether the melting rate is enhanced, reduced, or nearly unaffected. We construct a phase diagram based on the parameter matrix $(\log_{10} l, \phi)$to delineate these three regimes. Lastly, we derived two asymptotic solutions for predicting the variation in time of the unmelted solid height.

The Weissenberg effect, or rod-climbing phenomenon, occurs in non-Newtonian fluids where the fluid interface ascends along a rotating rod. Despite its prominence, theoretical insights into this phenomenon remain limited. In earlier work, Joseph \& Fosdick (\emph{Arch. Rat. Mech. Anal.}, vol. 49, 1973, pp. 321--380) employed domain perturbation methods for second-order fluids to determine the equilibrium interface height by expanding solutions based on the rotation speed. In this work, we investigate the time-dependent interface height through asymptotic analysis with dimensionless variables and equations using the Oldroyd-B model. We begin by neglecting surface tension and inertia to focus on the interaction between gravity and viscoelasticity. In the small-deformation scenario, the governing equations indicate the presence of a boundary layer in time, where the interface rises rapidly over a short time scale before gradually approaching a steady state. By employing a stretched time variable, we derive the transient velocity field and corresponding interface profile on this short time scale and recover the steady-state profile on a longer time scale. Subsequently, we reintroduce small but finite inertial effects to investigate their interplay with viscoelasticity and propose a criterion for determining the conditions under which rod-climbing occurs., Comment: 24 pages, 6 figures

Tidal disruption events (TDEs) are a class of transients that occur when a star is destroyed by the tides of a massive black hole (MBH). Their rates encode valuable MBH demographic information, but this can only be extracted if accurate TDE rate predictions are available for comparisons with observed rates. In this work, we present a new, observer-friendly Python package called REPTiDE, which implements a standard loss cone model for computing TDE rates given a stellar density distribution and an MBH mass. We apply this software to a representative sample of 91 nearby galaxies over a wide range of stellar masses with high-resolution nuclear density measurements from arXiv:2407.10911. We measure per-galaxy TDE rates ranging between 10$^{-7.7}$ and 10$^{-2.9}$ per year and find that the sample-averaged rates agree well with observations. We find a turnover in the TDE rate as a function of both galaxy stellar mass and black hole mass, with the peak rates being observed in galaxies at a galaxy mass of $10^{9.5}$ M$_\odot$ and a black hole mass of $10^{6.5}$ M$_\odot$. Despite the lower TDE rates inferred for intermediate-mass black holes, we find that they have gained a higher fraction of their mass through TDEs when compared to higher mass black holes. This growth of lower mass black holes through TDEs can enable us to place interesting constraints on their spins; we find maximum spins of $a_\bullet \approx 0.9$ for black holes with masses below $\sim10^{5.5}$ M$_\odot$., Comment: 20 pages, 8 figures, 1 table. Submitted to AAS Journals

Deep surveys reveal tidal debris and associated compact stellar systems. Euclid's unique combination of capabilities (spatial resolution, depth, and wide sky coverage) will make it a groundbreaking tool for galactic archaeology in the local Universe, bringing low surface brightness (LSB) science into the era of large-scale astronomical surveys. Euclid's Early Release Observations (ERO) demonstrate this potential with a field of view that includes several galaxies in the Dorado group. In this paper, we aim to derive from this image a mass assembly scenario for its main galaxies: NGC 1549, NGC 1553, and NGC 1546. We detect internal and external diffuse structures, and identify candidate globular clusters (GCs). By analysing the colours and distributions of the diffuse structures and candidate GCs, we can place constraints on the galaxies' mass assembly and merger histories. The results show that feature morphology, surface brightness, colours, and GC density profiles are consistent with galaxies that have undergone different merger scenarios. We classify NGC 1549 as a pure elliptical galaxy that has undergone a major merger. NGC 1553 appears to have recently transitioned from a late-type galaxy to early type, after a series of radial minor to intermediate mergers. NGC 1546 is a rare specimen of galaxy with an undisturbed disk and a prominent diffuse stellar halo, which we infer has been fed by minor mergers and then disturbed by the tidal effect from NGC 1553. Finally, we identify limitations specific to the observing conditions of this ERO, in particular stray light in the visible and persistence in the near-infrared bands. Once these issues are addressed and the extended emission from LSB objects is preserved by the data-processing pipeline, the Euclid Wide Survey will allow studies of the local Universe to be extended to statistical ensembles over a large part of the extragalactic sky., Comment: 25 pages, 20 figures

The o1 model series is trained with large-scale reinforcement learning to reason using chain of thought. These advanced reasoning capabilities provide new avenues for improving the safety and robustness of our models. In particular, our models can reason about our safety policies in context when responding to potentially unsafe prompts, through deliberative alignment. This leads to state-of-the-art performance on certain benchmarks for risks such as generating illicit advice, choosing stereotyped responses, and succumbing to known jailbreaks. Training models to incorporate a chain of thought before answering has the potential to unlock substantial benefits, while also increasing potential risks that stem from heightened intelligence. Our results underscore the need for building robust alignment methods, extensively stress-testing their efficacy, and maintaining meticulous risk management protocols. This report outlines the safety work carried out for the OpenAI o1 and OpenAI o1-mini models, including safety evaluations, external red teaming, and Preparedness Framework evaluations.

Visual imagery does not consist of solitary objects, but instead reflects the composition of a multitude of fluid concepts. While there have been great advances in visual representation learning, such advances have focused on building better representations for a small number of discrete objects bereft of an understanding of how these objects are interacting. One can observe this limitation in representations learned through captions or contrastive learning -- where the learned model treats an image essentially as a bag of words. Several works have attempted to address this limitation through the development of bespoke learned architectures to directly address the shortcomings in compositional learning. In this work, we focus on simple, and scalable approaches. In particular, we demonstrate that by substantially improving weakly labeled data, i.e. captions, we can vastly improve the performance of standard contrastive learning approaches. Previous CLIP models achieved near chance rate on challenging tasks probing compositional learning. However, our simple approach boosts performance of CLIP substantially and surpasses all bespoke architectures. Furthermore, we showcase our results on a relatively new captioning benchmark derived from DOCCI. We demonstrate through a series of ablations that a standard CLIP model trained with enhanced data may demonstrate impressive performance on image retrieval tasks.

We present a systematic study of the $BVRI$ colours of novae over the course of their eruptions. Where possible, interstellar reddening was measured using the equivalent widths of Diffuse Interstellar Bands (DIBs). Some novae lack spectra with sufficient resolution and signal-to-noise ratios; therefore, we supplement as necessary with 3D and 2D dust maps. Utilising only novae with DIB- or 3D-map-based $E(B-V)$, we find an average intrinsic $(B-V)_0$ colour of novae at $V$-band light curve peak of 0.18 with a standard deviation of 0.31, based on a sample of 23 novae. When the light curve has declined by 2 magnitudes ($t_2$), we find an average $(B-V)_0 = -0.02$ with a standard deviation of 0.19. These average colours are consistent with previous findings, although the spreads are larger than previously found due to more accurate reddening estimates. We also examined the intrinsic $(R-I)_0$ and $(V-R)_0$ colours across our sample. These colours behave similarly to $(B-V)_0$, except that the $(V-R)_0$ colour gets redder after peak, likely due to the contributions of emission line flux. We searched for correlations between nova colours and $t_2$, peak $V$-band absolute magnitude, and GeV $\gamma$-ray luminosity, but find no statistically significant correlations. Nova colours can therefore be used as standard "crayons" to estimate interstellar reddening from photometry alone, with 0.2--0.3 mag uncertainty. We present a novel Bayesian strategy for estimating distances to Galactic novae based on these $E(B-V)$ measurements, independent of assumptions about luminosity, built using 3D dust maps and a stellar mass model of the Milky Way., Comment: 19 pages, 21 Figures. Submitted to MNRAS

Recent advancements in robot learning have used imitation learning with large models and extensive demonstrations to develop effective policies. However, these models are often limited by the quantity, quality, and diversity of demonstrations. This paper explores improving offline-trained imitation learning models through online interactions with the environment. We introduce Policy Decorator, which uses a model-agnostic residual policy to refine large imitation learning models during online interactions. By implementing controlled exploration strategies, Policy Decorator enables stable, sample-efficient online learning. Our evaluation spans eight tasks across two benchmarks-ManiSkill and Adroit-and involves two state-of-the-art imitation learning models (Behavior Transformer and Diffusion Policy). The results show Policy Decorator effectively improves the offline-trained policies and preserves the smooth motion of imitation learning models, avoiding the erratic behaviors of pure RL policies. See our project page (https://policydecorator.github.io) for videos., Comment: Explore videos, data, code, and more at https://policydecorator.github.io

Kagome metals with van Hove singularities (VHSs) near the Fermi level can host intriguing quantum phenomena, including chiral loop currents, electronic nematicity, and unconventional superconductivity. However, unconventional magnetic states driven by VHSs, such as spin-density waves (SDWs), have yet to be observed experimentally in kagome metals. Here, we present a comprehensive investigation of the magnetic and electronic structure of the layered kagome metal CeTi3Bi4, where the Ti kagome electronic structure interacts with a magnetic sublattice of Ce3+ Jeff = 1/2 moments. Our neutron diffraction measurements reveal an incommensurate SDW ground state of the Ce3+ Jeff = 1/2 moments, which notably coexists with commensurate antiferromagnetic order across most of the temperature-field phase diagram. The commensurate component is preferentially suppressed by both thermal fluctuations and external magnetic fields, resulting in a rich phase diagram that includes an intermediate single-Q SDW phase. First-principles calculations and angle-resolved photoemission spectroscopy (ARPES) measurements identify VHSs near the Fermi level, with the observed magnetic propagation vectors connecting their high density of states, strongly suggesting a VHS-assisted SDW in CeTi3Bi4. These findings establish the rare-earth Kagome metals LnTi3Bi4 as a model platform where characteristic electronic structure of the kagome lattice plays a pivotal role in magnetic order., Comment: 20 pages, 4 figures, SI not included

Supercontinuum generation in the ultraviolet spectral region is challenging in solid-core optical fibres due to solarization and photodarkening. Antiresonant hollow-core fibres have overcome this limitation and have been shown to guide ultraviolet light at sufficient intensity for ultraviolet spectral broadening through nonlinear optics in the filling gas. However, their ultraviolet guidance is usually limited by discontinuities caused by the presence of high-loss resonance bands. In this paper, we report on resonance-free supercontinuum generation spanning from the deep ultraviolet to the near infrared achieved through modulation instability in an argon-filled antiresonant hollow-core fibre. The fibre was directly fabricated using the stack-and-draw method with a wall thickness of approximately 90 nm, enabling continuous spectral coverage from the deep ultraviolet to the near infrared. We also report numerical simulations to investigate the supercontinuum bandwidth and the factors limiting it, finding that the overall dispersion landscape, and associated group-velocity matching of cross-phase modulation interactions, is the dominant constraint on spectral extension.

Robot decision-making in partially observable, real-time, dynamic, and multi-agent environments remains a difficult and unsolved challenge. Model-free reinforcement learning (RL) is a promising approach to learning decision-making in such domains, however, end-to-end RL in complex environments is often intractable. To address this challenge in the RoboCup Standard Platform League (SPL) domain, we developed a novel architecture integrating RL within a classical robotics stack, while employing a multi-fidelity sim2real approach and decomposing behavior into learned sub-behaviors with heuristic selection. Our architecture led to victory in the 2024 RoboCup SPL Challenge Shield Division. In this work, we fully describe our system's architecture and empirically analyze key design decisions that contributed to its success. Our approach demonstrates how RL-based behaviors can be integrated into complete robot behavior architectures., Comment: ICRA 2025

High-quality benchmarks are the foundation for embodied AI research, enabling significant advancements in long-horizon navigation, manipulation and rearrangement tasks. However, as frontier tasks in robotics get more advanced, they require faster simulation speed, more intricate test environments, and larger demonstration datasets. To this end, we present MS-HAB, a holistic benchmark for low-level manipulation and in-home object rearrangement. First, we provide a GPU-accelerated implementation of the Home Assistant Benchmark (HAB). We support realistic low-level control and achieve over 3x the speed of prior magical grasp implementations at a fraction of the GPU memory usage. Second, we train extensive reinforcement learning (RL) and imitation learning (IL) baselines for future work to compare against. Finally, we develop a rule-based trajectory filtering system to sample specific demonstrations from our RL policies which match predefined criteria for robot behavior and safety. Combining demonstration filtering with our fast environments enables efficient, controlled data generation at scale.

Protein evolution underpins life, and understanding its behavior as a system is of great importance. However, our current models of protein evolution are arguably too simplistic to allow quantitative interpretation and prediction of evolutionary trajectories. Viewing protein evolution as a complex system has the potential to advance our understanding and ability to model protein evolution. In this perspective, we discuss aspects of protein evolution that are typical of complex systems, from nonlinear dynamics, sensitivity to initial conditions, self-organization, and the emergence of order from chaos and disorder. We discuss how the growth in sequence and structural data, insights from laboratory evolution and new machine learning tools can advance the study of protein evolution and that by treating protein evolution as a complex adaptive system, we may gain new insights into the fundamental principles driving biological innovation and adaptation and apply this to protein engineering and design.

Rewards remain an uninterpretable way to specify tasks for Reinforcement Learning, as humans are often unable to predict the optimal behavior of any given reward function, leading to poor reward design and reward hacking. Language presents an appealing way to communicate intent to agents and bypass reward design, but prior efforts to do so have been limited by costly and unscalable labeling efforts. In this work, we propose a method for a completely unsupervised alternative to grounding language instructions in a zero-shot manner to obtain policies. We present a solution that takes the form of imagine, project, and imitate: The agent imagines the observation sequence corresponding to the language description of a task, projects the imagined sequence to our target domain, and grounds it to a policy. Video-language models allow us to imagine task descriptions that leverage knowledge of tasks learned from internet-scale video-text mappings. The challenge remains to ground these generations to a policy. In this work, we show that we can achieve a zero-shot language-to-behavior policy by first grounding the imagined sequences in real observations of an unsupervised RL agent and using a closed-form solution to imitation learning that allows the RL agent to mimic the grounded observations. Our method, RLZero, is the first to our knowledge to show zero-shot language to behavior generation abilities without any supervision on a variety of tasks on simulated domains. We further show that RLZero can also generate policies zero-shot from cross-embodied videos such as those scraped from YouTube., Comment: 27 pages

The majority of most luminous quasars during the epoch of reionization accrete near or above the Eddington limit, marking the vigorous growth of primitive supermassive black holes (SMBHs). However, their subsequent evolution and environmental impact remain poorly characterized. We present JWST/NIRSpec prism IFU observations of HSC J2239+0207, a low-luminosity quasar at $z\sim6.25$ likely in a late stage of mass assembly with an overmassive SMBH relative to its host galaxy. Using H$\beta$ and H$\alpha$ broad emission lines, we estimate an SMBH mass $M_{\rm BH}\sim3\times10^8~M_{\odot}$ and confirm its sub-Eddington accretion at $\lambda_{\rm Edd}\sim0.4$. Strong FeII emission and a proximity zone of typical size suggest a metal-rich, highly evolved system. In the far-UV, this quasar presents strong broad-absorption-line features, indicative of high-velocity winds ($\nu\sim10^4~{\rm km/s}$). Meanwhile, minimal dust reddening is inferred from the quasar continuum and broad-line Balmer decrement, suggesting little dust along the polar direction. Most interestingly, we identify a gas companion $\sim$5 kpc from the quasar with a high [OIII]/H$\beta$ ratio ($\gtrsim10$), likely representing outflowing gas blown away by AGN feedback. These results highlight HSC J2239+0207 as a likely fading quasar in transition, providing rare insights into SMBH evolution, AGN feedback, and AGN-galaxy interactions in the early Universe., Comment: 12 pages, 5 figures; Accepted for publication in ApJL

Kerr-microresonator frequency combs in integrated photonics waveguides are promising technologies for next-generation positioning, navigation, and timing applications, with advantages that include platforms that are mass-producible and CMOS-compatible and spectra that are phase-coherent and octave-spanning. Fundamental thermal noise in the resonator material typically limits the timing and frequency stability of a microcomb. The small optical mode volume of the microresonators exaggerates this effect, as it both increases the magnitude and shortens the timescale of thermodynamic fluctuations. In this work, we investigate thermal instability in silicon nitride microring resonators as well as techniques for reducing their effects on the microcomb light. We characterize the time-dependent thermal response in silicon nitride microring resonators through experimental measurements and finite element method simulations. Through fast control of the pump laser frequency, we reduce thermal recoil due to heating. Finally, we demonstrate the utility of a coupled microresonator system with tunable mode interactions to stabilize a soliton pulse against thermal shifts.

We present the MULTIMODAL UNIVERSE, a large-scale multimodal dataset of scientific astronomical data, compiled specifically to facilitate machine learning research. Overall, the MULTIMODAL UNIVERSE contains hundreds of millions of astronomical observations, constituting 100\,TB of multi-channel and hyper-spectral images, spectra, multivariate time series, as well as a wide variety of associated scientific measurements and "metadata". In addition, we include a range of benchmark tasks representative of standard practices for machine learning methods in astrophysics. This massive dataset will enable the development of large multi-modal models specifically targeted towards scientific applications. All codes used to compile the MULTIMODAL UNIVERSE and a description of how to access the data is available at https://github.com/MultimodalUniverse/MultimodalUniverse, Comment: Accepted at NeurIPS Datasets and Benchmarks track

Activity and autonomous motion are fundamental aspects of many living and engineering systems. Here, the scale of biological agents covers a wide range, from nanomotors, cytoskeleton, and cells, to insects, fish, birds, and people. Inspired by biological active systems, various types of autonomous synthetic nano- and micromachines have been designed, which provide the basis for multifunctional, highly responsive, intelligent active materials. A major challenge for understanding and designing active matter is their inherent non-equilibrium nature due to persistent energy consumption, which invalidates equilibrium concepts such as free energy, detailed balance, and time-reversal symmetry. Furthermore, interactions in ensembles of active agents are often non-additive and non-reciprocal. An important aspect of biological agents is their ability to sense the environment, process this information, and adjust their motion accordingly. It is an important goal for the engineering of micro-robotic systems to achieve similar functionality. With many fundamental properties of motile active matter now reasonably well understood and under control, the ground is prepared for the study of physical aspects and mechanisms of motion in complex environments, of the behavior of systems with new physical features like chirality, of the development of novel micromachines and microbots, of the emergent collective behavior and swarming of intelligent self-propelled particles, and of particular features of microbial systems. The vast complexity of phenomena and mechanisms involved in the self-organization and dynamics of motile active matter poses major challenges, which can only be addressed by a truly interdisciplinary effort involving scientists from biology, chemistry, ecology, engineering, mathematics, and physics.

CoSn and FeSn, two kagome-lattice metals, have recently attracted significant attention as hosts of electronic flat bands and emergent physical properties. However, current understandings of their physical properties are limited to the knowledge of the average crystal structure. Here, we report the Fe-doping induced co-emergence of the antiferromagentic (AFM) order and local symmetry breaking in (Co0.45Fe0.55)Sn. Rietveld analysis on the neutron and synchrotron x-ray diffraction data indicates A-type antiferromagnetic order with the moment pointing perpendicular to the kagome layers, associated with the anomaly in the MSn(1)2Sn(2)4 (M = Co/Fe) octahedral distortion and the lattice constant c. Reverse Monte Carlo (RMC) modeling of the synchrotron x-ray total scattering results captured the subtle local orthorhombic distortion involving off-axis displacements of Sn2. Our results indicate that the stable hexagonal lattice above TN becomes unstable once the A-type AFM order is formed below TN. We argue that the local symmetry breaking has a magnetic origin and is driven by the out-of-plane magnetic exchange coupling. Our study provides comprehensive information on the crystal structure in both long-range scale and local scale, unveiling unique coupling between AFM order, octahedral distortion, and hidden local symmetry breaking., Comment: 28 pages, 7 figures, Accepted in Journal of the American Chemical Society

Having explored an environment, intelligent agents should be able to transfer their knowledge to most downstream tasks within that environment. Referred to as "zero-shot learning," this ability remains elusive for general-purpose reinforcement learning algorithms. While recent works have attempted to produce zero-shot RL agents, they make assumptions about the nature of the tasks or the structure of the MDP. We present \emph{Proto Successor Measure}: the basis set for all possible solutions of Reinforcement Learning in a dynamical system. We provably show that any possible policy can be represented using an affine combination of these policy independent basis functions. Given a reward function at test time, we simply need to find the right set of linear weights to combine these basis corresponding to the optimal policy. We derive a practical algorithm to learn these basis functions using only interaction data from the environment and show that our approach can produce the optimal policy at test time for any given reward function without additional environmental interactions. Project page: https://agarwalsiddhant10.github.io/projects/psm.html., Comment: Under submission, 23 pages

Triangular lattice antiferromagnets are prototypes for frustrated magnetism and may potentially realize novel quantum magnetic states such as a quantum spin liquid ground state. A recent work suggests NdTa$_7$O$_{19}$ with rare-earth triangular lattice is a quantum spin liquid candidate and highlights the large family of rare-earth heptatantalates as a framework for quantum magnetism investigation. In this paper, we report the structural and magnetic characterization of CeTa$_7$O$_{19}$ and YbTa$_7$O$_{19}$. Both compounds are isostructural to NdTa$_7$O$_{19}$ with no detectable structural disorder. For CeTa$_7$O$_{19}$, the crystal field energy levels and parameters are determined by inelastic neutron scattering measurements. Based on the crystal field result, the magnetic susceptibility data could be well fitted and explained, which reveals that CeTa$_7$O$_{19}$ is a highly anisotropic Ising triangular-lattice antiferromagnet ($g_z$/$g_{xy}$$\sim$3) with very weak exchange interaction (J$\sim$0.22~K). For YbTa$_7$O$_{19}$, millimeter sized single crystals could be grown. The anisotropic magnetization and electron spin resonance data show that YbTa$_7$O$_{19}$ has a contrasting in-plane magnetic anisotropy with $g_z$/$g_{xy}$$\sim$0.67 similar as that of YbMgGaO$_4$. The above results indicate that CeTa$_7$O$_{19}$ and YbTa$_7$O$_{19}$ with pseudospin-1/2 ground states might either be quantum spin liquid candidate materials or find applications in adiabatic demagnetization refrigeration due to the weak exchange interaction., Comment: 10 pages, 7 figures

A pulsar's scintillation bandwidth is inversely proportional to the scattering delay, making accurate measurements of scintillation bandwidth critical to characterize unmitigated delays in efforts to measure low-frequency gravitational waves with pulsar timing arrays. In this pilot work, we searched for a subset of known pulsars within $\sim$97% of the data taken with the PUPPI instrument for the AO327 survey with the Arecibo telescope, attempting to measure the scintillation bandwidths in the dataset by fitting to the 2D autocorrelation function of their dynamic spectra. We successfully measured 38 bandwidths from 23 pulsars (six without prior literature values), finding that: almost all of the measurements are larger than the predictions from NE2001 and YMW16 (two popular galactic models); NE2001 is more consistent with our measurements than YMW16; Gaussian fits to the bandwidth are more consistent with both electron density models than Lorentzian ones; and for the 17 pulsars with prior literature values, the measurements between various sources often vary by factors of a few. The success of Gaussian fits may be due to the use of Gaussian fits to train models in previous work. The variance of literature values over time could relate to the scaling factor used to compare measurements, but also seems consistent with time-varying interstellar medium parameters. This work can be extended to the rest of AO327 to further investigate these trends, highlighting the continuing importance of large archival datasets for projects beyond their initial conception., Comment: 21 pages, 8 figures, 2 tables

In kinetic theory, the classic $n \Sigma v$ approach calculates the rate of particle interactions from local quantities: the number density of particles $n$, the cross-section $\Sigma$, and the average relative speed $v$. In stellar dynamics, this formula is often applied to problems in collisional (i.e. dense) environments such as globular and nuclear star clusters, where blue stragglers, tidal capture binaries, binary ionizations, and micro-tidal disruptions arise from rare close encounters. The local $n \Sigma v$ approach implicitly assumes the ergodic hypothesis, which is not well motivated for the densest star systems in the Universe. In the centers of globular and nuclear star clusters, orbits close into 1D ellipses because of the degeneracy of the potential (either Keplerian or harmonic). We find that the interaction rate in perfectly Keplerian or harmonic potentials is determined by a global quantity -- the number of orbital intersections -- and that this rate can be far lower or higher than the ergodic $n \Sigma v$ estimate. However, we find that in most astrophysical systems, deviations from a perfectly Keplerian or harmonic potential (due to e.g. granularity or extended mass) trigger sufficient orbital precession to recover the $n \Sigma v$ interaction rate. Astrophysically relevant failures of the $n \Sigma v$ approach only seem to occur for tightly bound stars orbiting intermediate-mass black holes, or for the high-mass end of collisional cascades in certain debris disks., Comment: Submitted to ApJ

The Euclid Wide Survey (EWS) is predicted to find approximately 170 000 galaxy-galaxy strong lenses from its lifetime observation of 14 000 deg^2 of the sky. Detecting this many lenses by visual inspection with professional astronomers and citizen scientists alone is infeasible. Machine learning algorithms, particularly convolutional neural networks (CNNs), have been used as an automated method of detecting strong lenses, and have proven fruitful in finding galaxy-galaxy strong lens candidates. We identify the major challenge to be the automatic detection of galaxy-galaxy strong lenses while simultaneously maintaining a low false positive rate. One aim of this research is to have a quantified starting point on the achieved purity and completeness with our current version of CNN-based detection pipelines for the VIS images of EWS. We select all sources with VIS IE < 23 mag from the Euclid Early Release Observation imaging of the Perseus field. We apply a range of CNN architectures to detect strong lenses in these cutouts. All our networks perform extremely well on simulated data sets and their respective validation sets. However, when applied to real Euclid imaging, the highest lens purity is just 11%. Among all our networks, the false positives are typically identifiable by human volunteers as, for example, spiral galaxies, multiple sources, and artefacts, implying that improvements are still possible, perhaps via a second, more interpretable lens selection filtering stage. There is currently no alternative to human classification of CNN-selected lens candidates. Given the expected 10^5 lensing systems in Euclid, this implies 10^6 objects for human classification, which while very large is not in principle intractable and not without precedent., Comment: 22 pages, 11 figures, Euclid consortium paper, A&A submitted

This paper starts by introducing results from geometric measure theory to prove symmetric decreasing rearrangement inequalities on $\mathbb{R}^n$, which give multiple proofs of the isoperimetric and P\'{o}lya-Szeg\H{o} inequalities. Then we consider smooth oriented Riemannian manifolds of the form $M^n = (0,\infty)\times \Sigma^{n-1}$, and test what results carry over from the $\mathbb{R}^n$ setting or what assumptions about $M^n$ need to be added. Of particular interest was proving the smooth co-area formula in the Riemannian manifolds setting and re-formulating particular geometric inequalities., Comment: This work was conducted as part of an undergraduate research program (URSS) at the University of Warwick, under the supervision of Dr Maxwell Stolarski. It applies geometric measure theory to study how symmetric rearrangement inequalities in $\mathbb{R}^n$ generalize to a certain type of Riemannian manifold with symmetry. The work is a preliminary investigation into these geometric inequalities

We analyze 99 photometrically selected Little Red Dots (LRDs) at z ~ 4-8 in the GOODS fields, leveraging ultra-deep JADES NIRCam short-wavelength (SW) data. We examine the morphology of 30 LRDs; the remaining 69 are predominantly compact, i.e. are strongly dominated by sources < 400 pc in diameter and lack extended components even in stacked SW band images. Among the LRDs selected for morphological analysis, 50% show at least two distinct, associated sources or galaxy components, while the others appear as single sources with highly asymmetric structures. We find median stellar masses of log10(M*/Msun) = 9.07(-0.08)(+0.11) for pure stellar models with Av ~ 1.16(+0.11)(-0.21) mag, and log10(M*/Msun) = 9.67(+0.17)(-0.27) for models including AGNs, where Av ~ 2.74(+0.55)(-0.71) mag, consistent with recent results showing LRDs tend to have high stellar masses and dust content when fitted with AGN models. NIRSpec spectra are available for 15 sources, 6 of which fall within the morphological analysis sample and show multiple components. Among these 15, broad H-alpha emission is detected in 40%, with full-width half-maximum (FWHM) ranging from 1200 to 2900 km/s. One source exhibits broad H-beta emission with FWHM = 2000 +/- 500 km/s. Analysis of line ratios probing the interstellar medium (ISM) reveals a composite nature, indicating AGN activity combined with stellar processes. These findings suggest LRDs have a mixed nature, with AGN signatures in some cases linked to disturbed morphologies observed at rest-frame UV wavelengths., Comment: 28 pages, 9 figures, 3 tables. Submitted to ApJ. Comments are welcome!

We investigate the evolution of supernova remnants (SNRs) in a two-phase cloudy medium by performing a series of high-resolution (up to $\Delta x\approx0.01\,\mathrm{pc}$), 3D hydrodynamical simulations including radiative cooling and thermal conduction. We aim to reach a resolution that directly captures the shock-cloud interactions for the majority of the clouds initialized by the saturation of thermal instability. In comparison to the SNR in a uniform medium with the volume filling warm medium, the SNR expands similarly (following $\propto t^{2/5}$) but sweeps up more mass as the cold clouds contribute before shocks in the warm medium become radiative. However, the SNR in a cloudy medium continuously loses energy after shocks toward the cold clouds cool, resulting in less hot gas mass, thermal energy, and terminal momentum. Thermal conduction has little effect on the dynamics of the SNR but smooths the morphology and modifies the internal structure by increasing the density of hot gas by a factor of $\sim 3-5$. The simulation results are not fully consistent with many previous 1D models describing the SNR in a cloudy medium including a mass loading term. By direct measurement in the simulations, we find that, apart from the mass source, the energy sink is also important with a spatially flat cooling rate $\dot{e}\propto t^{-11/5}$. As an illustration, we show an example 1D model including both mass source and energy sink terms (in addition to the radiative cooling in the volume filling component) that better describes the structure of the simulated SNR., Comment: 24 pages, 14 figures, submitted to ApJ