Search

Since the recent discovery of the directly imaged super-Jovian planet AF Lep b, several studies have been conducted to characterize its atmosphere and constrain its orbital parameters. AF Lep b has a measured dynamical mass of $3.68 \pm 0.48$ MJup, a radius of $1.3 \pm 0.15$ RJup, a nearly circular orbit in spin-orbit alignment with the host star, a relatively high metallicity, and a near-solar to super-solar C/O ratio. However, key parameters such as the rotational velocity and radial velocity could not be estimated as they require high-resolution spectroscopic data that is impossible to obtain with classical spectrographs. AF Lep b was recently observed with the new HiRISE visitor instrument at the VLT, with the goal of obtaining high-resolution (R~140,000) spectroscopic observations to better constrain the orbital and atmospheric parameters of the young giant exoplanet. We compare the extracted spectrum of AF Lep b to self-consistent atmospheric models using ForMoSA. We then use our measurements of the radial velocity of the planet to provide new constraints on the orbit of the planet. From the forward modeling, we find a C/O ratio that aligns with previous low-resolution analyses, and we confirm the super-solar metallicity. We also confirm unambiguously the presence of methane in the atmosphere of the companion. Based on all available relative astrometry and radial velocity measurements of the host star, we show that two distinct orbital populations are possible for the companion. We derive the radial velocity of AF Lep b to be $10.51 \pm 1.03$ km/s, and show that this value agrees well with one of the two orbital solutions, allowing us to rule out an entire family of orbits. Additionally, assuming that the rotation and orbit are coplanar, the derived planet's rotation rate is consistent with the observed trend of increasing spin velocity with higher planet mass., Comment: Accepted for publication in A&A

Measurement of a physical quantity such as light intensity is an integral part of many reconstruction and decision scenarios but can be costly in terms of acquisition time, invasion of or damage to the environment and storage. Data minimisation and compliance with data protection laws is also an important consideration. Where there are a range of measurements that can be made, some may be more informative and compliant with the overall measurement objective than others. We develop an active sequential inference algorithm that uses the low dimensional representational latent space from a variational autoencoder (VAE) to choose which measurement to make next. Our aim is to recover high dimensional data by making as few measurements as possible. We adapt the VAE encoder to map partial data measurements on to the latent space of the complete data. The algorithm draws samples from this latent space and uses the VAE decoder to generate data conditional on the partial measurements. Estimated measurements are made on the generated data and fed back through the partial VAE encoder to the latent space where they can be evaluated prior to making a measurement. Starting from no measurements and a normal prior on the latent space, we consider alternative strategies for choosing the next measurement and updating the predictive posterior prior for the next step. The algorithm is illustrated using the Fashion MNIST dataset and a novel convolutional Hadamard pattern measurement basis. We see that useful patterns are chosen within 10 steps, leading to the convergence of the guiding generative images. Compared with using stochastic variational inference to infer the parameters of the posterior distribution for each generated data point individually, the partial VAE framework can efficiently process batches of generated data and obtains superior results with minimal measurements.

Objectives: To determine the on-field validity of instrumented mouthguards (iMGs) in American football and to quantify head acceleration event (HAE) incidence in NCAA football players. Methods: Instrumented mouthguards were fitted to 35 male NCAA football players. Head kinematic data were collected during 64 player matches. On-field validity was determined through video review with positive predictive value (PPV) and sensitivity values calculated. HAE incidence was calculated as the number of HAEs per player match and stratified by Offense and Defense positions. Results: On-field validity of the Prevent Biometrics iMG in NCAA American Football indicates a sensitivity was 0.89 and PPV ranging from 0.76-0.98 based on false positive definitions. The incidence of PLA and PAA HAEs above a range of thresholds in Defense and Offense appear similar. The incidence of HAEs above 10 g was 11.2 and 11.3 HAEs per player match for Defense and Offense, respectively, while PAA incidence above 1.0 krad/s2 was 5.5 and 6.9 HAEs per player match for Defense and Offense, respectively. Incidence of HAEs above 30 g was 1.6 and 2.6 per player match and 0.9 and 1.4 for HAEs above 2.0 krad/s2 for Defense and Offense, respectively. Conclusion: The Prevent Biometrics iMG appears suitable for measuring HAEs in elite American football. The study provides a benchmark assessment of HAE incidence in elite American Football and lays a foundation for the development of position-specific interventions aimed at reducing HAE exposure.

Comparing composite models for multi-component observational data is a prevalent scientific challenge. When fitting composite models, there exists the potential for systematics from a poor fit of one model component to be absorbed by another, resulting in the composite model providing an accurate fit to the data in aggregate but yielding biased a posteriori estimates for individual components. We begin by defining a classification scheme for composite model comparison scenarios, identifying two categories: category I, where models with accurate and predictive components are separable through Bayesian comparison of the unvalidated composite models, and category II, where models with accurate and predictive components may not be separable due to interactions between components, leading to spurious detections or biased signal estimation. To address the limitations of category II model comparisons, we introduce the Bayesian Null Test Evidence Ratio-based (BaNTER) validation framework. Applying this classification scheme and BaNTER to a composite model comparison problem in 21-cm cosmology, where minor systematics from imperfect foreground modelling can bias global 21-cm signal recovery, we validate six composite models using mock data. We show that incorporating BaNTER alongside Bayes-factor-based comparison reliably ensures unbiased inferences of the signal of interest across both categories, positioning BaNTER as a valuable addition to Bayesian inference workflows with potential applications across diverse fields., Comment: 20 pages, 5 figures; accepted for publication in MNRAS

Based on Irish older adult's perceptions, practices, and challenges regarding password management, the goal of this study was to compile suitable advice that can benefit this demographic. To achieve this, we first conducted semi structured interviews (n=37), we then collated advice based on best practice and what we learned from these interviews. We facilitated two independent focus groups (n=31) to evaluate and adjust this advice and tested the finalized advice through an observational study (n=15). The participants were aged between 59 and 86 and came from various counties in Ireland, both rural and urban. The findings revealed that managing multiple passwords was a significant source of frustration, leading some participants to adopt novel and informal strategies for storing them. A notable hesitation to adopt digital password managers and passphrases was also observed. Participants appreciated guidance on improving their password practices, with many affirming that securely writing down passwords was a practical strategy. Irish older adults demonstrated strong intuition regarding cybersecurity, notably expressing concerns over knowledge-based security checks used by banks and government institutions. This study aims to contribute to the aggregation of practical password advice suited to older adults, making password security more manageable and less burdensome for this demographic.

As digital services increasingly replace traditional analogue systems, ensuring that older adults are not left behind is critical to fostering inclusive access. This study explores how digital educators support older adults in developing essential digital skills, drawing insights from interviews with $34$ educators in Ireland. These educators, both professional and volunteer, offer instruction through a range of formats, including workshops, remote calls, and in-person sessions. Our findings highlight the importance of personalized, step-by-step guidance tailored to older adults' learning needs, as well as fostering confidence through hands-on engagement with technology. Key challenges identified include limited transportation options, poor internet connectivity, outdated devices, and a lack of familial support for learning. To address these barriers, we propose enhanced public funding, expanded access to resources, and sustainable strategies such as providing relevant and practical course materials. Additionally, innovative tools like simulated online platforms for practicing digital transactions can help reduce anxiety and enhance digital literacy among older adults. This study underscores the vital role that digital educators play in bridging the digital divide, creating a more inclusive, human-centered approach to digital learning for older adults.

Self-attention in transformer models is an incremental associative memory that maps key vectors to value vectors. One way to speed up self-attention is to employ GPU-compliant vector search algorithms, yet the standard partitioning methods yield poor results in this context, because (1) keys and queries follow different distributions and (2) the effect of RoPE positional encoding. In this paper, we introduce SAAP (Self-Attention with Asymmetric Partitions), which overcomes these problems. It is an asymmetrical indexing technique that employs distinct partitions for keys and queries, thereby approximating self-attention with a data-adaptive sparsity pattern. It works on pretrained language models without finetuning, as it only requires to train (offline) a small query classifier. On a long context Llama 3.1-8b model, with sequences ranging from 100k to 500k tokens, our method typically reduces by a factor 20 the fraction of memory that needs to be looked-up, which translates to a time saving of 60\% when compared to FlashAttention-v2.

Cluster scaling relations are key ingredients in cluster abundance-based cosmological studies. In optical cluster cosmology, weak gravitational lensing has proven to be a powerful tool to constrain the cluster mass-richness relation. This work is conducted as part of the Dark Energy Science Collaboration (DESC), which aims to analyze the Legacy Survey of Space and Time (LSST) of Vera C. Rubin Observatory, starting in 2026. Weak lensing-inferred cluster properties, such as mass, suffer from several sources of bias. In this paper, we aim to test the impact of modeling choices and observational systematics in cluster lensing on the inference of the mass-richness relation. We constrain the mass-richness relation of 3,600 clusters detected by the redMaPPer algorithm in the cosmoDC2 extra-galactic mock catalog (covering $440$ deg$^2$) of the LSST DESC DC2 simulation, using number count measurements and stacked weak lensing profiles in several intervals of richness ($20 \leq \lambda \leq 200$) and redshift ($0.2 \leq z \leq 1$). By modeling the mean of the scaling relation as $\langle \ln \lambda|M_{\rm 200c}, z\rangle = \ln\lambda_0 + \mu_z\log[(1+z)/(1+0.5)] + \mu_m[\log_{10}(M_{\rm 200c}) - 14.3]$, our baseline constraints are $\ln\lambda_0 = 3.37\pm 0.03$, $\mu_z = 0.08\pm 0.07$ and $\mu_m = 2.18 \pm 0.07$. We have found that, for a LSST-like source galaxy density, our constraints are robust to a change in concentration-mass relation and dark matter density profile modeling choices, when source redshifts and shapes are perfectly known. We have found that photometric redshift uncertainties can introduce bias at the $1\sigma$ level, which can be mitigated by an overall correcting factor, fitted jointly with scaling parameters. We find that including positive shear-richness covariance in the fit shifts the results by up to 0.5$\sigma$., Comment: 26 pages, 15 figures, 3 tables, submitted to A&A (abstract shortened for arXiv)

The tendency of users to anthropomorphise large language models (LLMs) is of growing interest to AI developers, researchers, and policy-makers. Here, we present a novel method for empirically evaluating anthropomorphic LLM behaviours in realistic and varied settings. Going beyond single-turn static benchmarks, we contribute three methodological advances in state-of-the-art (SOTA) LLM evaluation. First, we develop a multi-turn evaluation of 14 anthropomorphic behaviours. Second, we present a scalable, automated approach by employing simulations of user interactions. Third, we conduct an interactive, large-scale human subject study (N=1101) to validate that the model behaviours we measure predict real users' anthropomorphic perceptions. We find that all SOTA LLMs evaluated exhibit similar behaviours, characterised by relationship-building (e.g., empathy and validation) and first-person pronoun use, and that the majority of behaviours only first occur after multiple turns. Our work lays an empirical foundation for investigating how design choices influence anthropomorphic model behaviours and for progressing the ethical debate on the desirability of these behaviours. It also showcases the necessity of multi-turn evaluations for complex social phenomena in human-AI interaction.

The recent development of powerful AI systems has highlighted the need for robust risk management frameworks in the AI industry. Although companies have begun to implement safety frameworks, current approaches often lack the systematic rigor found in other high-risk industries. This paper presents a comprehensive risk management framework for the development of frontier AI that bridges this gap by integrating established risk management principles with emerging AI-specific practices. The framework consists of four key components: (1) risk identification (through literature review, open-ended red-teaming, and risk modeling), (2) risk analysis and evaluation using quantitative metrics and clearly defined thresholds, (3) risk treatment through mitigation measures such as containment, deployment controls, and assurance processes, and (4) risk governance establishing clear organizational structures and accountability. Drawing from best practices in mature industries such as aviation or nuclear power, while accounting for AI's unique challenges, this framework provides AI developers with actionable guidelines for implementing robust risk management. The paper details how each component should be implemented throughout the life-cycle of the AI system - from planning through deployment - and emphasizes the importance and feasibility of conducting risk management work prior to the final training run to minimize the burden associated with it.

Recent radiation-thermochemical-magnetohydrodynamic simulations resolved formation of quasar accretion disks from cosmological scales down to ~300 gravitational radii $R_{g}$, arguing they were 'hyper-magnetized' (plasma $\beta\ll1$ supported by toroidal magnetic fields) and distinct from traditional $\alpha$-disks. We extend these, refining to $\approx 3\,R_{g}$ around a $10^{7}\,{\rm M_{\odot}}$ BH with multi-channel radiation and thermochemistry, and exploring a factor of 1000 range of accretion rates ($\dot{m}\sim0.01-20$). At smaller scales, we see the disks maintain steady accretion, thermalize and self-ionize, and radiation pressure grows in importance, but large deviations from local thermodynamic equilibrium and single-phase equations of state are always present. Trans-Alfvenic and highly-supersonic turbulence persists in all cases, and leads to efficient vertical mixing, so radiation pressure saturates at levels comparable to fluctuating magnetic and turbulent pressures even for $\dot{m}\gg1$. The disks also become radiatively inefficient in the inner regions at high $\dot{m}$. The midplane magnetic field remains primarily toroidal at large radii, but at super-Eddington $\dot{m}$ we see occasional transitions to a poloidal-field dominated state associated with outflows and flares. Large-scale magnetocentrifugal and continuum radiation-pressure-driven outflows are weak at $\dot{m}<1$, but can be strong at $\dot{m}\gtrsim1$. In all cases there is a scattering photosphere above the disk extending to $\gtrsim 1000\,R_{g}$ at large $\dot{m}$, and the disk is thick and flared owing to magnetic support (with $H/R$ nearly independent of $\dot{m}$), so the outer disk is strongly illuminated by the inner disk and most of the inner disk continuum scatters or is reprocessed at larger scales, giving apparent emission region sizes as large as $\gtrsim 10^{16}\,{\rm cm}$., Comment: Submitted to the Open Journal of Astrophysics. 40 pages, 31 figures, high-level overview sections provided. Comments welcome

This paper proposes a vision-in-the-loop simulation environment for deep monocular pose estimation of a UAV operating in an ocean environment. Recently, a deep neural network with a transformer architecture has been successfully trained to estimate the pose of a UAV relative to the flight deck of a research vessel, overcoming several limitations of GPS-based approaches. However, validating the deep pose estimation scheme in an actual ocean environment poses significant challenges due to the limited availability of research vessels and the associated operational costs. To address these issues, we present a photo-realistic 3D virtual environment leveraging recent advancements in Gaussian splatting, a novel technique that represents 3D scenes by modeling image pixels as Gaussian distributions in 3D space, creating a lightweight and high-quality visual model from multiple viewpoints. This approach enables the creation of a virtual environment integrating multiple real-world images collected in situ. The resulting simulation enables the indoor testing of flight maneuvers while verifying all aspects of flight software, hardware, and the deep monocular pose estimation scheme. This approach provides a cost-effective solution for testing and validating the autonomous flight of shipboard UAVs, specifically focusing on vision-based control and estimation algorithms., Comment: 8 pages, 15 figures, conference

Acoustic telemetry data plays a vital role in understanding the behaviour and movement of aquatic animals. However, these datasets, which often consist of millions of individual data points, frequently contain anomalous movements that pose significant challenges. Traditionally, anomalous movements are identified either manually or through basic statistical methods, approaches that are time-consuming and prone to high rates of unidentified anomalies in large datasets. This study focuses on the development of automated classifiers for a large telemetry dataset comprising detections from fifty acoustically tagged dusky kob monitored in the Breede Estuary, South Africa. Using an array of 16 acoustic receivers deployed throughout the estuary between 2016 and 2021, we collected over three million individual data points. We present detailed guidelines for data pre-processing, resampling strategies, labelling process, feature engineering, data splitting methodologies, and the selection and interpretation of machine learning and deep learning models for anomaly detection. Among the evaluated models, neural networks autoencoder (NN-AE) demonstrated superior performance, aided by our proposed threshold-finding algorithm. NN-AE achieved a high recall with no false normal (i.e., no misclassifications of anomalous movements as normal patterns), a critical factor in ensuring that no true anomalies are overlooked. In contrast, other models exhibited false normal fractions exceeding 0.9, indicating they failed to detect the majority of true anomalies; a significant limitation for telemetry studies where undetected anomalies can distort interpretations of movement patterns. While the NN-AE's performance highlights its reliability and robustness in detecting anomalies, it faced challenges in accurately learning normal movement patterns when these patterns gradually deviated from anomalous ones.

Computational Fluid Dynamics (CFD) plays a pivotal role in fluid mechanics, enabling precise simulations of fluid behavior through partial differential equations (PDEs). However, traditional CFD methods are resource-intensive, particularly for high-fidelity simulations of complex flows, which are further complicated by high dimensionality, inherent stochasticity, and limited data availability. This paper addresses these challenges by proposing a data-driven approach that leverages a Gaussian Mixture Variational Autoencoder (GMVAE) to encode high-dimensional scientific data into low-dimensional, physically meaningful representations. The GMVAE learns a structured latent space where data can be categorized based on physical properties such as the Reynolds number while maintaining global physical consistency. To assess the interpretability of the learned representations, we introduce a novel metric based on graph spectral theory, quantifying the smoothness of physical quantities along the latent manifold. We validate our approach using 2D Navier-Stokes simulations of flow past a cylinder over a range of Reynolds numbers. Our results demonstrate that the GMVAE provides improved clustering, meaningful latent structure, and robust generative capabilities compared to baseline dimensionality reduction methods. This framework offers a promising direction for data-driven turbulence modeling and broader applications in computational fluid dynamics and engineering systems.

Most of the world's languages and dialects are low-resource, and lack support in mainstream machine translation (MT) models. However, many of them have a closely-related high-resource language (HRL) neighbor, and differ in linguistically regular ways from it. This underscores the importance of model robustness to dialectal variation and cross-lingual generalization to the HRL dialect continuum. We present DialUp, consisting of a training-time technique for adapting a pretrained model to dialectal data (M->D), and an inference-time intervention adapting dialectal data to the model expertise (D->M). M->D induces model robustness to potentially unseen and unknown dialects by exposure to synthetic data exemplifying linguistic mechanisms of dialectal variation, whereas D->M treats dialectal divergence for known target dialects. These methods show considerable performance gains for several dialects from four language families, and modest gains for two other language families. We also conduct feature and error analyses, which show that language varieties with low baseline MT performance are more likely to benefit from these approaches., Comment: 9 pages, 46 incl. appendix

The 2-point correlation function of the galaxy spatial distribution is a major cosmological observable that enables constraints on the dynamics and geometry of the Universe. The Euclid mission aims at performing an extensive spectroscopic survey of approximately 20--30 million H$\alpha$-emitting galaxies up to about redshift two. This ambitious project seeks to elucidate the nature of dark energy by mapping the 3-dimensional clustering of galaxies over a significant portion of the sky. This paper presents the methodology and software developed for estimating the 3-dimensional 2-point correlation function within the Euclid Science Ground Segment. The software is designed to overcome the significant challenges posed by the large and complex Euclid data set, which involves millions of galaxies. Key challenges include efficient pair counting, managing computational resources, and ensuring the accuracy of the correlation function estimation. The software leverages advanced algorithms, including kd-tree, octree, and linked-list data partitioning strategies, to optimise the pair-counting process. The implementation also includes parallel processing capabilities using shared-memory open multi-processing to further enhance performance and reduce computation times. Extensive validation and performance testing of the software are presented. The results indicate that the software is robust and can reliably estimate the 2-point correlation function, which is essential for deriving cosmological parameters with high precision. Furthermore, the paper discusses the expected performance of the software during different stages of the Euclid Wide Survey observations and forecasts how the precision of the correlation function measurements will improve over the mission's timeline, highlighting the software's capability to handle large data sets efficiently., Comment: 17 pages, 13 figures, submitted to A&A

Design educators are finding ways to support students in skillfully using GenAI tools in their practices while encouraging the critical scrutiny of the ethical and social issues around these technologies. However, the issue of environmental sustainability remains unaddressed. There is a lack of both resources to grasp the environmental costs of genAI in education and a lack of shared practices for engaging with the issue. This paper critically reflects on the energy costs of using genAI in design education, using a workshop held in 2023 with 49 students as a motivating example. Through this reflection, we develop a set of five alternative stances, with related actions, that support the conscious use of genAI in design education. The work contributes to the field of design and HCI by bringing together ways for educators to reflect on their practices, informing the future development of educational programs around genAI., Comment: 25 pages, 3 figures, 1 table

In this work we study non-Hermitian extensions of the paradigmatic spin-1/2 XY chain in a magnetic field. Using the mapping of the model to free fermion form, we provide analytical insights into the energy spectrum of the non-Hermitian model and establish an intrinsic connection between the quasienergies and topological invariants. We also use exact diagonalization as a supplementary method to examine the performance of biorthogonal-based expectation values. Our results confirm that the theoretical analysis is consistent with the numerical results, with the extended phase diagram determined via the analytical solution and the critical behavior of the fidelity and entanglement. The entanglement transition goes hand in hand with the non-Hermitian topological phase transition. Like the Hermitian case, we analyze the critical behavior using finite-size scaling. Our results show that non-Hermiticity can induce the system into a new universality class with unusual critical exponent. We also emphasize the ability of the Loschmidt echo to characterize potential phase transitions and introduce the average of the Loschmidt echo to describe phase transitions in non-Hermitian systems., Comment: 13 pages, 5 figures

The half-life of $^{39}$Ar is measured using the DEAP-3600 detector located 2 km underground at SNOLAB. In 2016-2020, DEAP-3600 used a target mass of (3269 $\pm$ 24) kg of liquid argon distilled from the atmosphere in a direct-detection dark matter search. Such an argon mass also enables direct measurements of argon isotope properties. The decay of $^{39}$Ar in DEAP-3600 is the dominant source of triggers by two orders of magnitude, ensuring high statistics and making DEAP-3600 well-suited for measuring this isotope's half-life. Use of the pulse-shape discrimination technique in DEAP-3600 allows for powerful discrimination between nuclear recoils and electron recoils, resulting in the selection of a clean sample of $^{39}$Ar decays. Observing over a period of 3.4 years, the $^{39}$Ar half-life is measured to be $(302 \pm 8_{\rm stat} \pm 6_{\rm sys})$ years. This new direct measurement suggests that the half-life of $^{39}$Ar may be significantly longer than the accepted value, with potential implications for measurements using this isotope's half-life as input., Comment: 11 pages, 8 figures

We use the Milky Way neutral hydrogen (HI) absorption and emission spectra from the Galactic Australian Square Kilometre Array Pathfinder (GASKAP) Phase II Pilot survey along with toy models to investigate the effects of stacking multicomponent spectra on measurements of peak optical depth and spin temperature. Shifting spectra by the peak in emission, 'primary' components shifted to 0 km s$^{-1}$ are correctly averaged. Additional components on individual sightlines are averaged with non-centred velocities, producing a broader and shallower 'secondary' component in the resulting stack. Peak optical depths and brightness temperatures of the secondary components from stacks are lower limits of their true average values due to the velocity offset of each component. The spin temperature however is well correlated with the truth since the velocity offset of components affects the emission and absorption spectra equally. Stacking 462 GASKAP absorption-emission spectral pairs, we detect a component with a spin temperature of 1320 $\pm$ 263 K, consistent with gas from the unstable neutral medium and higher than any previous GASKAP detection in this region. We also stack 2240 pilot survey spectra containing no Milky Way absorption, revealing a primary narrow and secondary broad component, with spin temperatures belonging to the cold neutral medium (CNM). Spatially binning and stacking the non-detections across the plane-of-sky by their distance from CNM absorption detections, the primary component's optical depth decreases with distance from known locations of cold gas. The spin temperature however remains stable in both components, over an approximate physical plane-of-sky distance of $\sim$ 100 pc., Comment: This paper has 16 pages, 17 figures. This paper has been accepted for publication in Monthly Notices of the Royal Astronomical Society Main Journal

We show that if a locally finite Borel graph with quasitransitive amenable components admits a fractional perfect matching, it will admit a Borel fractional perfect matching. In particular, if a countable amenable quasitransitive graph admits a fractional perfect matching then its Bernoulli graph admits a Borel fractional perfect matching., Comment: 18 pages, 5 figures; corrected reference

We present a catalog of individual stellar and dust extinction properties along close to 500,000 sight lines in the southwest bar of the Small Magellanic Cloud (SMC). The catalog is based on multiband Hubble Space Telescope photometric data spanning near-ultraviolet to near-infrared wavelengths from the Small Magellanic Cloud Investigation of Dust and Gas Evolution survey (SMIDGE) covering a 100 x 200 pc area. We use the probabilistic technique of the Bayesian Extinction And Stellar Tool (BEAST) to model the spectral energy distributions of individual stars in SMIDGE and include the effects of observational uncertainties in the data. We compare BEAST-derived dust extinction properties with tracers of the interstellar medium, such as the emission from the 12CO (2-1) transition (I(CO)), the dust mass surface density ({\Sigma}dust) from far-IR emission, the H I column density (N(HI)) from the 21cm transition, and the mass fraction of polycyclic aromatic hydrocarbons (PAHs; qPAH, derived from IR emission). We find that the dust extinction (A(V )) in the SMIDGE field is strongly correlated with {\Sigma}dust and I(CO), and less so with N(HI) and qPAH, and suggest potential explanations. Our extinction measurements are also sensitive to the presence of the 2175 {\AA} bump in the extinction curve toward UV bright stars. While most do not show evidence for the bump, we identify ~200 lines of sight that are 2175 {\AA} bump candidates. Furthermore, we find distinct structures in the dust extinction-distance distributions that provide insights into the 3D geometry of the SMC., Comment: 20 pages, 13 figures, 4 tables

With the rise and ubiquity of larger deep learning models, the need for high-quality compression techniques is growing in order to deploy these models widely. The sheer parameter count of these models makes it difficult to fit them into the memory constraints of different hardware. In this work, we present a novel approach to model compression by merging similar parameter groups within a model, rather than pruning away less important parameters. Specifically, we select, align, and merge separate feed-forward sublayers in Transformer models, and test our method on language modeling, image classification, and machine translation. With our method, we demonstrate performance comparable to the original models while combining more than a third of model feed-forward sublayers, and demonstrate improved performance over a strong layer-pruning baseline. For instance, we can remove over 21% of total parameters from a Vision Transformer, while maintaining 99% of its original performance. Additionally, we observe that some groups of feed-forward sublayers exhibit high activation similarity, which may help explain their surprising mergeability.

Friedl and L\"oh (2021, Confl. Math.) prove that testing whether or not there is an epimorphism from a finitely presented group to a virtually cyclic group, or to the direct product of an abelian and a finite group, is decidable. Here we prove that these problems are $\mathsf{NP}$-complete. We also show that testing epimorphism is $\mathsf{NP}$-complete when the target is a restricted type of semi-direct product of a finitely generated free abelian group and a finite group, thus extending the class of virtually abelian target groups for which decidability of epimorphism is known. Lastly, we consider epimorphism onto a fixed finite group. We show the problem is $\mathsf{NP}$-complete when the target is a dihedral groups of order that is not a power of 2, complementing the work on Kuperberg and Samperton (2018, Geom. Topol.) who showed the same result when the target is non-abelian finite simple., Comment: 41 pages, 2 tables

While machine learning offers diverse techniques suitable for exploring various medical research questions, a cohesive synergistic framework can facilitate the integration and understanding of new approaches within unified model development and interpretation. We therefore introduce the Medical Artificial Intelligence Toolbox (MAIT), an explainable, open-source Python pipeline for developing and evaluating binary classification, regression, and survival models on tabular datasets. MAIT addresses key challenges (e.g., high dimensionality, class imbalance, mixed variable types, and missingness) while promoting transparency in reporting (TRIPOD+AI compliant). Offering automated configurations for beginners and customizable source code for experts, MAIT streamlines two primary use cases: Discovery (feature importance via unified scoring, e.g., SHapley Additive exPlanations - SHAP) and Prediction (model development and deployment with optimized solutions). Moreover, MAIT proposes new techniques including fine-tuning of probability threshold in binary classification, translation of cumulative hazard curves to binary classification, enhanced visualizations for model interpretation for mixed data types, and handling censoring through semi-supervised learning, to adapt to a wide set of data constraints and study designs. We provide detailed tutorials on GitHub, using four open-access data sets, to demonstrate how MAIT can be used to improve implementation and interpretation of ML models in medical research., Comment: 14 pages, 2 figures, 1 table

With the advent of ever more intense and focused X-ray sources, including in laboratories, at synchrotrons, and at X-ray free electron lasers, radiation-induced sample change and damage are becoming increasingly challenging. Therefore, the exploration of possible mitigation strategies is crucial to continue to allow the collection of robust and repeatable data. One mitigation approach is the introduction of short, X-ray-free ``dark'' periods. However, it is unclear whether this strategy minimises damage or, in actuality, promotes it through a phenomenon called ``dark progression'', i.e. the increase or progression of radiation damage that occurs after the X-ray beam is turned off. This work discusses the influence of introducing dark periods and their duration on the radiation-induced changes in two model small-molecule catalysts, [Ir(COD)Cl]$_2$ and [Rh(COD)Cl]$_2$, exposed to X-ray radiation in powder diffraction (PXRD) and photoelectron spectroscopy (XPS) experiments. This provides, for the first time, insights into how damage progresses under varying radiation regimes and allows the distinction between the processes that affect the unit cell itself, the individual molecular units, and the respective atomic chemical environments. Furthermore, it provides the basis for informed decision-making in the design of future experiments where the need to minimise radiation-induced damage is crucial.

This paper explores entropy-controlled dithering techniques in audio compression, examining the application of standard and modified TPDFs, combined with noise shaping and entropy-controlled parameters, across various audio contexts, including pitch, loudness, rhythm, and instrumentation variations. Perceptual quality metrics such as VISQOL and STOI were used to evaluate performance. The results demonstrate that TPDF-based dithering consistently outperforms RPDF, particularly under optimal alpha conditions, while highlighting performance variability based on signal characteristics. These findings suggest the situational appropriateness of using various TPDF distributions. This work emphasizes the trade-off between entropy and perceptual fidelity, offering insights into the potential of entropy-controlled dithering as a foundation for enhanced audio compression algorithms. A practical implementation as a Digital Audio Workstation plugin introduces customizable dithering controls, laying the groundwork for future advancements in audio compression algorithms., Comment: 19 pages, 15 figures

Accurate depth estimation is crucial for many fields, including robotics, navigation, and medical imaging. However, conventional depth sensors often produce low-resolution (LR) depth maps, making detailed scene perception challenging. To address this, enhancing LR depth maps to high-resolution (HR) ones has become essential, guided by HR-structured inputs like RGB or grayscale images. We propose a novel sensor fusion methodology for guided depth super-resolution (GDSR), a technique that combines LR depth maps with HR images to estimate detailed HR depth maps. Our key contribution is the Incremental guided attention fusion (IGAF) module, which effectively learns to fuse features from RGB images and LR depth maps, producing accurate HR depth maps. Using IGAF, we build a robust super-resolution model and evaluate it on multiple benchmark datasets. Our model achieves state-of-the-art results compared to all baseline models on the NYU v2 dataset for $\times 4$, $\times 8$, and $\times 16$ upsampling. It also outperforms all baselines in a zero-shot setting on the Middlebury, Lu, and RGB-D-D datasets. Code, environments, and models are available on GitHub.

Complex materials with internal microstructure such as suspensions and emulsions exhibit time-dependent rheology characterized by viscoelasticity and thixotropy. In many large-scale applications such as turbulent pipe flow, the elastic response occurs on a much shorter timescale than the thixotropy, hence these flows are purely thixotropic. The fundamental dynamics of thixotropic turbulence is poorly understood, particularly the interplay between microstructural state, rheology, and turbulence structure. To address this gap, we conduct direct numerical simulations (DNS) of fully developed turbulent pipe flow of a model thixotropic (Moore) fluid over a range of thixoviscous numbers $\Lambda$ from slow ($\Lambda\ll 1$) to fast ($\Lambda\gg 1$) thixotropic kinetics relative to the eddy turnover time. Analysis of DNS results in the Lagrangian frame shows that, as expected, in the limits of slow and fast kinetics, these time-dependent flows behave as time-independent purely viscous (generalized Newtonian) analogues. For intermediate kinetics ($\Lambda\sim 1$), the rheology is governed by a \emph{path integral} of the thixotropic fading memory kernel over the distribution of Lagrangian shear history, the latter of which is modelled via a simple stochastic model for the radially non-stationary pipe flow. DNS computations based on this effective viscosity closure exhibit excellent agreement (within 2.4\% error) with the fully thixotropic model for $\Lambda=1$, indicating that the purely viscous (generalized Newtonian) analogue persists for arbitrary values of $\Lambda\in[0,\infty^+)$ and across nonlinear rheology models. These results uncover the feedback mechanisms between microstructure, rheology, and turbulence and offer fundamental insights into the structure of thixotropic turbulence., Comment: 31 pages

Continuous gravitational waves (CWs) emission from neutron stars carries information about their internal structure and equation of state, and it can provide tests of General Relativity. We present a search for CWs from a set of 45 known pulsars in the first part of the fourth LIGO--Virgo--KAGRA observing run, known as O4a. We conducted a targeted search for each pulsar using three independent analysis methods considering the single-harmonic and the dual-harmonic emission models. We find no evidence of a CW signal in O4a data for both models and set upper limits on the signal amplitude and on the ellipticity, which quantifies the asymmetry in the neutron star mass distribution. For the single-harmonic emission model, 29 targets have the upper limit on the amplitude below the theoretical spin-down limit. The lowest upper limit on the amplitude is $6.4\!\times\!10^{-27}$ for the young energetic pulsar J0537-6910, while the lowest constraint on the ellipticity is $8.8\!\times\!10^{-9}$ for the bright nearby millisecond pulsar J0437-4715. Additionally, for a subset of 16 targets we performed a narrowband search that is more robust regarding the emission model, with no evidence of a signal. We also found no evidence of non-standard polarizations as predicted by the Brans-Dicke theory., Comment: main paper: 12 pages, 6 figures, 4 tables

We present OLMo 2, the next generation of our fully open language models. OLMo 2 includes dense autoregressive models with improved architecture and training recipe, pretraining data mixtures, and instruction tuning recipes. Our modified model architecture and training recipe achieve both better training stability and improved per-token efficiency. Our updated pretraining data mixture introduces a new, specialized data mix called Dolmino Mix 1124, which significantly improves model capabilities across many downstream task benchmarks when introduced via late-stage curriculum training (i.e. specialized data during the annealing phase of pretraining). Finally, we incorporate best practices from T\"ulu 3 to develop OLMo 2-Instruct, focusing on permissive data and extending our final-stage reinforcement learning with verifiable rewards (RLVR). Our OLMo 2 base models sit at the Pareto frontier of performance to compute, often matching or outperforming open-weight only models like Llama 3.1 and Qwen 2.5 while using fewer FLOPs and with fully transparent training data, code, and recipe. Our fully open OLMo 2-Instruct models are competitive with or surpassing open-weight only models of comparable size, including Qwen 2.5, Llama 3.1 and Gemma 2. We release all OLMo 2 artifacts openly -- models at 7B and 13B scales, both pretrained and post-trained, including their full training data, training code and recipes, training logs and thousands of intermediate checkpoints. The final instruction model is available on the Ai2 Playground as a free research demo., Comment: Model demo available at playground.allenai.org

In the realm of data analysis and bioinformatics, representing time series data in a manner akin to biological sequences offers a novel approach to leverage sequence analysis techniques. Transforming time series signals into molecular sequence-type representations allows us to enhance pattern recognition by applying sophisticated sequence analysis techniques (e.g. $k$-mers based representation) developed in bioinformatics, uncovering hidden patterns and relationships in complex, non-linear time series data. This paper proposes a method to transform time series signals into biological/molecular sequence-type representations using a unique alphabetic mapping technique. By generating 26 ranges corresponding to the 26 letters of the English alphabet, each value within the time series is mapped to a specific character based on its range. This conversion facilitates the application of sequence analysis algorithms, typically used in bioinformatics, to analyze time series data. We demonstrate the effectiveness of this approach by converting real-world time series signals into character sequences and performing sequence classification. The resulting sequences can be utilized for various sequence-based analysis techniques, offering a new perspective on time series data representation and analysis.

Accurate molecular sequence analysis is a key task in the field of bioinformatics. To apply molecular sequence classification algorithms, we first need to generate the appropriate representations of the sequences. Traditional numeric sequence representation techniques are mostly based on sequence alignment that faces limitations in the form of lack of accuracy. Although several alignment-free techniques have also been introduced, their tabular data form results in low performance when used with Deep Learning (DL) models compared to the competitive performance observed in the case of image-based data. To find a solution to this problem and to make Deep Learning (DL) models function to their maximum potential while capturing the important spatial information in the sequence data, we propose a universal Hibert curve-based Chaos Game Representation (CGR) method. This method is a transformative function that involves a novel Alphabetic index mapping technique used in constructing Hilbert curve-based image representation from molecular sequences. Our method can be globally applied to any type of molecular sequence data. The Hilbert curve-based image representations can be used as input to sophisticated vision DL models for sequence classification. The proposed method shows promising results as it outperforms current state-of-the-art methods by achieving a high accuracy of $94.5$\% and an F1 score of $93.9\%$ when tested with the CNN model on the lung cancer dataset. This approach opens up a new horizon for exploring molecular sequence analysis using image classification methods.

This position paper argues that the majority of theory of mind benchmarks are broken because of their inability to directly test how large language models (LLMs) adapt to new partners. This problem stems from the fact that theory of mind benchmarks for LLMs are overwhelmingly inspired by the methods used to test theory of mind in humans and fall victim to a fallacy of attributing human-like qualities to AI agents. We expect that humans will engage in a consistent reasoning process across various questions about a situation, but this is known to not be the case for current LLMs. Most theory of mind benchmarks only measure what we call literal theory of mind: the ability to predict the behavior of others. Measuring this kind of reasoning is very informative in testing the ability of agents with self-consistent reasoning. However, it is important to note the distinction between this and what we actually care about when this self-consistency cannot be taken for granted. We call this functional theory of mind: the ability to adapt to agents in-context following a rational response to predictions about their behavior. We find that top performing open source LLMs may display strong capabilities in literal theory of mind, depending on how they are prompted, but seem to struggle with functional theory of mind -- even when partner policies are exceedingly simple. Simply put, strong literal theory of mind performance does not necessarily imply strong functional theory of mind performance. Achieving functional theory of mind, particularly over long interaction horizons with a partner, is a significant challenge deserving a prominent role in any meaningful LLM theory of mind evaluation.

Randomized benchmarking (RB) comprises a set of mature and widely used techniques for assessing the quality of operations on a quantum information-processing system. Modern RB protocols for multiqubit systems extract physically relevant error rates by exploiting the structure of the group representation generated by the set of benchmarked operations. However, existing techniques become prohibitively inefficient for representations that are highly reducible yet decompose into irreducible subspaces of high dimension. These situations prevail when benchmarking high-dimensional systems such as qudits or bosonic modes, where experimental control is limited to implementing a small subset of all possible unitary operations. In this work, we introduce a broad framework for enhancing the sample efficiency of RB that is sufficiently powerful to extend the practical reach of RB beyond the multiqubit setting. Our strategy, which applies to any benchmarking group, uses "synthetic" quantum circuits with classical post-processing of both input and output data to leverage the full structure of reducible superoperator representations. To demonstrate the efficacy of our approach, we develop a detailed theory of RB for systems with rotational symmetry. Such systems carry a natural action of the group $\text{SU}(2)$, and they form the basis for several novel quantum error-correcting codes. We show that, for experimentally accessible high-spin systems, synthetic RB protocols can reduce the complexity of measuring rotationally invariant error rates by more than two orders of magnitude relative to standard approaches such as character RB., Comment: 68 pages, 9 figures

Context. The Galactic ASKAP collaboration (GASKAP) is undertaking an HI emission survey of the 21cm line to map the Magellanic system and the Galactic plane with the Australian Square Kilometre Array Pathfinder (ASKAP). One of the first areas observed in the Pilot Phase I of the survey was the Small Magellanic Cloud (SMC). Previous surveys of the SMC have uncovered new structures in the periphery of the SMC, along relatively low column density lines of sight. Aims. In this work we aimed to uncover the phase distribution of three distinct structures in the periphery of the SMC. This work will add to the constraints we have on the existence and survival of the cold neutral medium (CNM) in the SMC. Methods. We used ROHSA, a Gaussian decomposition algorithm, to model the emission across each cloud and classify the HI emission into their respective phases based on the linewidths of the fitted Gaussians. We created maps of velocity and column density of each phase of the HI across these three clouds. We measured the HI mass and CNM number density for each cloud. We also compared the HI results across the different phases with other gas tracers. Results. We find that in two clouds, the ends of each cloud are almost completely CNM dominated. Analysis of these two clouds indicates they are experiencing a compressive force from the direction of the SMC main body. In the third cloud we find a uniform CNM distribution along one wall of what is likely a supershell structure. Comparison with previous measurements of CO clumps in two of the clouds show the CO and HI are co-moving within a few km/s in regions of high HI column density, particularly when considering just the CNM., Comment: 16 pages, 11 figures, accepted for publication in Astronomy & Astrophysics

The availability of SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) virus data post-COVID has reached exponentially to an enormous magnitude, opening research doors to analyze its behavior. Various studies are conducted by researchers to gain a deeper understanding of the virus, like genomic surveillance, etc, so that efficient prevention mechanisms can be developed. However, the unstable nature of the virus (rapid mutations, multiple hosts, etc) creates challenges in designing analytical systems for it. Therefore, we propose a neural network-based (NN) mechanism to perform an efficient analysis of the SARS-CoV-2 data, as NN portrays generalized behavior upon training. Moreover, rather than using the full-length genome of the virus, we apply our method to its spike region, as this region is known to have predominant mutations and is used to attach to the host cell membrane. In this paper, we introduce a pipeline that first converts the spike protein sequences into a fixed-length numerical representation and then uses Neuromorphic Spiking Neural Network to classify those sequences. We compare the performance of our method with various baselines using real-world SARS-CoV-2 spike sequence data and show that our method is able to achieve higher predictive accuracy compared to the recent baselines.

Active Inference is a closed-loop computational theoretical basis for understanding behaviour, based on agents with internal probabilistic generative models that encode their beliefs about how hidden states in their environment cause their sensations. We review Active Inference and how it could be applied to model the human-computer interaction loop. Active Inference provides a coherent framework for managing generative models of humans, their environments, sensors and interface components. It informs off-line design and supports real-time, online adaptation. It provides model-based explanations for behaviours observed in HCI, and new tools to measure important concepts such as agency and engagement. We discuss how Active Inference offers a new basis for a theory of interaction in HCI, tools for design of modern, complex sensor-based systems, and integration of artificial intelligence technologies, enabling it to cope with diversity in human users and contexts. We discuss the practical challenges in implementing such Active Inference-based systems.

In molecular structure data, SMILES (Simplified Molecular Input Line Entry System) strings are used to analyze molecular structure design. Numerical feature representation of SMILES strings is a challenging task. This work proposes a kernel-based approach for encoding and analyzing molecular structures from SMILES strings. The proposed approach involves computing a kernel matrix using the Sinkhorn-Knopp algorithm while using kernel principal component analysis (PCA) for dimensionality reduction. The resulting low-dimensional embeddings are then used for classification and regression analysis. The kernel matrix is computed by converting the SMILES strings into molecular structures using the Morgan Fingerprint, which computes a fingerprint for each molecule. The distance matrix is computed using the pairwise kernels function. The Sinkhorn-Knopp algorithm is used to compute the final kernel matrix that satisfies the constraints of a probability distribution. This is achieved by iteratively adjusting the kernel matrix until the marginal distributions of the rows and columns match the desired marginal distributions. We provided a comprehensive empirical analysis of the proposed kernel method to evaluate its goodness with greater depth. The suggested method is assessed for drug subcategory prediction (classification task) and solubility AlogPS ``Aqueous solubility and Octanol/Water partition coefficient" (regression task) using the benchmark SMILES string dataset. The outcomes show the proposed method outperforms several baseline methods in terms of supervised analysis and has potential uses in molecular design and drug discovery. Overall, the suggested method is a promising avenue for kernel methods-based molecular structure analysis and design.

A remarkable characteristic of quantum computing is the potential for reliable computation despite faulty qubits. This can be achieved through quantum error correction, which is typically implemented by repeatedly applying static syndrome checks, permitting correction of logical information. Recently, the development of time-dynamic approaches to error correction has uncovered new codes and new code implementations. In this work, we experimentally demonstrate three time-dynamic implementations of the surface code, each offering a unique solution to hardware design challenges and introducing flexibility in surface code realization. First, we embed the surface code on a hexagonal lattice, reducing the necessary couplings per qubit from four to three. Second, we walk a surface code, swapping the role of data and measure qubits each round, achieving error correction with built-in removal of accumulated non-computational errors. Finally, we realize the surface code using iSWAP gates instead of the traditional CNOT, extending the set of viable gates for error correction without additional overhead. We measure the error suppression factor when scaling from distance-3 to distance-5 codes of $\Lambda_{35,\text{hex}} = 2.15(2)$, $\Lambda_{35,\text{walk}} = 1.69(6)$, and $\Lambda_{35,\text{iSWAP}} = 1.56(2)$, achieving state-of-the-art error suppression for each. With detailed error budgeting, we explore their performance trade-offs and implications for hardware design. This work demonstrates that dynamic circuit approaches satisfy the demands for fault-tolerance and opens new alternative avenues for scalable hardware design., Comment: 11 pages, 5 figures, Supplementary Information

Quantum error correction is essential for bridging the gap between the error rates of physical devices and the extremely low logical error rates required for quantum algorithms. Recent error-correction demonstrations on superconducting processors have focused primarily on the surface code, which offers a high error threshold but poses limitations for logical operations. In contrast, the color code enables much more efficient logic, although it requires more complex stabilizer measurements and decoding techniques. Measuring these stabilizers in planar architectures such as superconducting qubits is challenging, and so far, realizations of color codes have not addressed performance scaling with code size on any platform. Here, we present a comprehensive demonstration of the color code on a superconducting processor, achieving logical error suppression and performing logical operations. Scaling the code distance from three to five suppresses logical errors by a factor of $\Lambda_{3/5}$ = 1.56(4). Simulations indicate this performance is below the threshold of the color code, and furthermore that the color code may be more efficient than the surface code with modest device improvements. Using logical randomized benchmarking, we find that transversal Clifford gates add an error of only 0.0027(3), which is substantially less than the error of an idling error correction cycle. We inject magic states, a key resource for universal computation, achieving fidelities exceeding 99% with post-selection (retaining about 75% of the data). Finally, we successfully teleport logical states between distance-three color codes using lattice surgery, with teleported state fidelities between 86.5(1)% and 90.7(1)%. This work establishes the color code as a compelling research direction to realize fault-tolerant quantum computation on superconducting processors in the near future.

Next-generation heat-assisted magnetic recording (HAMR) relies on fast, localized heating of the magnetic medium during the write process. Au plasmonic near-field transducers are an attractive solution to this challenge, but increased thermal stability of Au films is required to improve long-term reliability. This work compares the effect of nanoscale Al, AlOx, and Ta capping films on Au thin films with Ti or Ta adhesion layers for use in HAMR and other high-temperature plasmonic applications. Thermal stability is investigated using a bespoke laser dewetting system, and SEM and AFM are extensively used to interrogate the resulting dewet areas. The most effective capping layers are found to be 0.5-1 nm of Al or AlOx, which can eliminate dewetting under certain conditions. Even one monolayer of AlOx is shown to be highly effective in reducing dewetting. In the case of thicker capping layers of Ta and AlOx, the Au film can easily dewet underneath, leaving an intact capping layer. It is concluded that thinner capping layers are most effective against dewetting as the Au cannot dewet without breaking them and pulling them apart during the dewetting process. A simple model based on energetics considerations is developed, which explains how thinner capping layers can more effectively protect the metal from pore or fissure creation. The model provides some convenient guidelines for choosing both the substrate and capping layer, for a given metal, to maximize the resistance to laser-induced damage., Comment: 14 pages, 12 figures, 3 tables. See publisher's website for Supporting Information

Recent studies in interpretability have explored the inner workings of transformer models trained on tasks across various domains, often discovering that these networks naturally develop surprisingly structured representations. When such representations comprehensively reflect the task domain's structure, they are commonly referred to as ``World Models'' (WMs). In this work, we discover such WMs in transformers trained on maze tasks. In particular, by employing Sparse Autoencoders (SAEs) and analysing attention patterns, we examine the construction of WMs and demonstrate consistency between the circuit analysis and the SAE feature-based analysis. We intervene upon the isolated features to confirm their causal role and, in doing so, find asymmetries between certain types of interventions. Surprisingly, we find that models are able to reason with respect to a greater number of active features than they see during training, even if attempting to specify these in the input token sequence would lead the model to fail. Futhermore, we observe that varying positional encodings can alter how WMs are encoded in a model's residual stream. By analyzing the causal role of these WMs in a toy domain we hope to make progress toward an understanding of emergent structure in the representations acquired by Transformers, leading to the development of more interpretable and controllable AI systems., Comment: Main paper: 9 pages, 9 figures. Supplementary material: 10 pages, 17 additional figures. Code and data will be available upon publication. Corresponding author: A. F. Spies (afspies@imperial.ac.uk)

Large language models (LLMs) offer the potential to automate a large number of tasks that previously have not been possible to automate, including some in science. There is considerable interest in whether LLMs can automate the process of causal inference by providing the information about causal links necessary to build a structural model. We use the case of confounding in the Coronary Drug Project (CDP), for which there are several studies listing expert-selected confounders that can serve as a ground truth. LLMs exhibit mediocre performance in identifying confounders in this setting, even though text about the ground truth is in their training data. Variables that experts identify as confounders are only slightly more likely to be labeled as confounders by LLMs compared to variables that experts consider non-confounders. Further, LLM judgment on confounder status is highly inconsistent across models, prompts, and irrelevant concerns like multiple-choice option ordering. LLMs do not yet have the ability to automate the reporting of causal links., Comment: 8 figures, 27 pages

My paper "Talking About Large Language Models" has more than once been interpreted as advocating a reductionist stance towards large language models. But the paper was not intended that way, and I do not endorse such positions. This short note situates the paper in the context of a larger philosophical project that is concerned with the (mis)use of words rather than metaphysics, in the spirit of Wittgenstein's later writing.

A generic scheme is proposed to investigate the entanglement entropy for a type of scale-invariant states, valid for orthonormal basis states in the ground state subspace of quantum many-body systems undergoing spontaneous symmetry breaking with type-B Goldstone modes in two spatial dimensions and beyond. It is argued that a contribution from the area law to the entanglement entropy is absent, since the closeness to the boundary between a subsystem and its environment is not well-defined, given that a permutation symmetry group with respect to the unit cells of degenerate ground state wave functions emerges. Three physical constraints imposed lead to a universal finite-system size scaling function in the dominant logarithmic contribution to the entanglement entropy. As a result, an abstract fractal underlying the ground state subspace is revealed, characterized by the fractal dimension. The latter in turn is identical to the number of type-B Goldstone modes for the orthonormal basis states. The prediction is numerically confirmed for the ${\rm SU}(2)$ spin-$s$ ferromagnetic Heisenberg model, the ${\rm SU}(2s+1)$ ferromagnetic model, and the staggered ${\rm SU}(3)$ spin-1 ferromagnetic biquadratic model., Comment: 16 pages, 5 figures

This study investigates the imaging ability of the time-domain linear sampling method (TLSM) when applied to laser ultrasonic (LU) tomography of subsurface defects from limited-aperture measurements. In this vein, the TLSM indicator and it spectral counterpart known as the multifrequency LSM are formulated within the context of LU testing. The affiliated imaging functionals are then computed using synthetic and experimental data germane to LU inspection of aluminum alloy specimens with manufactured defects. Hyperparameters of inversion are computationally analyzed. We demonstrate using synthetic data that the TLSM indicator has the unique ability to recover weak (or hard-to-reach) scatterers and has the potential to generate higher quality images compared to LSM. Provided high-SNR measurements, this advantage may be preserved in reconstructions from LU test data.

The ability of language models to learn a task from a few examples in context has generated substantial interest. Here, we provide a perspective that situates this type of supervised few-shot learning within a much broader spectrum of meta-learned in-context learning. Indeed, we suggest that any distribution of sequences in which context non-trivially decreases loss on subsequent predictions can be interpreted as eliciting a kind of in-context learning. We suggest that this perspective helps to unify the broad set of in-context abilities that language models exhibit $\unicode{x2014}$ such as adapting to tasks from instructions or role play, or extrapolating time series. This perspective also sheds light on potential roots of in-context learning in lower-level processing of linguistic dependencies (e.g. coreference or parallel structures). Finally, taking this perspective highlights the importance of generalization, which we suggest can be studied along several dimensions: not only the ability to learn something novel, but also flexibility in learning from different presentations, and in applying what is learned. We discuss broader connections to past literature in meta-learning and goal-conditioned agents, and other perspectives on learning and adaptation. We close by suggesting that research on in-context learning should consider this broader spectrum of in-context capabilities and types of generalization.