Machine Learning
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Wednesday 2 September
We explore predicting eCommerce user preferences for product aspects such as brand, size, and color - a task we define as Aspect Affinity. Solving this task improves customer understanding and enables fine-grained personalization in recommendation, search, and marketing. We frame Aspect Affinity as a temporal prediction task: forecasting a users future aspect choices from their time-ordered interaction history, capturing long-term preferences that evolve beyond the current session. To this end, we propose DeepAffinity, which leverages Small Language Models (SLMs) with structured prompts and specialized prediction heads fine-tuned for this task. We show DeepAffinity outperforms standard generative fine-tuning methods, while general-purpose open-source LLMs perform poorly without task-specific tuning, highlighting their limits in modeling nuanced behavior. Finally, DeepAffinity enhances recommendation quality on a large-scale multinational eCommerce platform.
High Energy Physics (Phenomenology)
Entanglement Signatures of Kinetic-Mixing Portals in Dark Monopole Scattering
We examine the interaction between the Standard Model (SM) fermions and the topological dark magnetic monopoles mediated via a kinetic-mixing portal to investigate the generation of quantum entanglement in particle-portal scattering. While the conventional phenomenology only takes into account total cross-sections, decay widths, and missing-energy signatures, we employ an information-theoretic approach to explain the scattering event. We quantify the quantum correlations transported over the portal boundary by calculating the Von Neumann entropy ($S_{\text{ent}}$) and the subsystem purity deterioration ($\gamma < 1$) analytically. Our results demonstrate that the non-perturbative core form factor of the dark monopole controls its high-energy momentum transfer and that the Von Neumann entropy increases quadratically with the topological magnetic charge ($g_m^2$) and the kinetic-mixing parameter ($\epsilon^2$). This paradigm provides an information-theoretic "microscope" to restrict portal parameter spaces and explore underlying topological structures without using traditional energy signatures by establishing quantum decoherence and purity loss as new, complementary observables.
Read the full abstract
Strongly Correlated Electrons
Prominent Dimensional Effects on High-order Harmonic Generation in Strongly Correlated Electron Systems
Dimensionality strongly affects elementary excitations in correlated quantum materials, yet its impact on extreme nonlinear optical responses remains largely unexplored. Here, we combine high-harmonic generation (HHG) experiments on quasi-one-dimensional SrCuO$_2$ and quasi-two-dimensional Pr$_2$CuO$_4$ with nonequilibrium simulations of Hubbard models. We find a pronounced dimensional contrast: SrCuO$_2$ exhibits a robust plateau-like high-harmonic spectrum with weak temperature dependence, whereas Pr$_2$CuO$_4$ shows a monotonic decrease in harmonic yield and strong thermal suppression, especially at higher harmonics. The simulations qualitatively reproduce these trends and identify dimensionality-dependent doublon–holon dephasing, governed by spin–charge coupling, as their microscopic origin. These results establish dimensionality and spin–charge coupling as key control parameters for extreme nonlinear optical responses in correlated insulators.
Read the full abstract
Cryptography and Security
Can Risk-Based Alerting Mitigate Cybersecurity Alert Fatigue?
Security operations centers (SOCs) face large numbers of false alerts, making detection of cyberattacks difficult under typical resource constraints. Risk-based alerting (RBA) has been proposed as a means to reduce false alerts and has reportedly succeeded in doing so in various enterprise deployments. However, RBA has not been comprehensively evaluated until now, leaving implementation mostly guesswork based on anecdotal evidence. In this paper, we present the first systematic evaluation of RBA. To this end, we reformulate it as a continuous alert prioritization problem rather than a binary decision problem (i.e., whether an alerting threshold is exceeded), allowing us to evaluate performance across all possible thresholds and thus model SOCs of varying sizes and alert volumes. We distill five fundamental risk hypotheses, formalize them as independently parametrizable modules, and implement them in our novel experimentation suite CATS. We thoroughly assess the hypotheses across eight diverse alert datasets, six of which we created or extended to make such an evaluation possible. Our results show that certain combinations of hypotheses achieve a remarkable alert prioritization performance (AUROC $\mu=0.92$, $\sigma=0.09$ across the eight datasets), outperforming a straightforward prioritization by alert severity level (AUROC $\mu=0.72$, $\sigma=0.21$). We conclude that RBA can substantially reduce the number of false alerts that analysts have to review and thus has the potential to mitigate cybersecurity alert fatigue. In addition, it serves as a strong baseline for more complex, resource-intensive alert triage approaches (e.g., based on large language models).
Read the full abstract
Probability
When does propagation of chaos in the critical Curie-Weiss model stop?
We study increasing propagation of chaos for the critical Curie- Weiss model (i.e. the mean-field Ising model at critical inverse temperature 1, with no external field). We give a simple way to see that for windows of size k(N) of smaller order than sqrt{N} we still have propagation of chaos (reproving earlier results, see e.g.[1]), while for k(N) of order sqrt{N} the propagation of chaos breaks down. The law of a single spin converges to \pi, the Bernoulli law with parameter 1/2. If k(N) = alpha sqrt{N}, we give an explicit formula for the limiting distance in total variation of the law of the first k(N) spins with respect to the k-fold product of pi, as a function of alpha. For even larger window sizes, the distribution of the spins has, in the thermodynamical limit, maximal distance to the k-fold product of pi. One of the ingredients of the proof is a result about the unimodality/non- unimodality of the law of the number of positive spins among the first k = k(N) spins, which may be of independent interest.
Read the full abstract
Nuclear Theory
Clustered Nature of Hot and Dense Nuclear Matter: A quantum statistical approach
The equilibrium abundances of the light clusters $^2$H, $^3$H, $^3$He, $^4$He in hot nuclear matter at densities near the saturation density are of essential interest for nuclear physics and astrophysical applications, but theoretical approaches give diverging answers. We compare the quantum statistical approach with the recently discussed phase-space excluded-volume approach. We analyze the main ingredients, the Mott momentum, and the momentum distribution functions of light clusters. We observe a sharp decrease in cluster abundances as the density approaches saturation density, that is also seen in a relativistic mean-field calculation. We outline possible improvements in determining the composition of hot, dense matter in thermodynamic equilibrium. Non-equilibrium effects must be taken into account to investigate cluster formation in heavy-ion collisions.
Read the full abstract
Computer Vision
VIPS: Vehicle-Infrastructure Cooperative Planning Benchmark via Pseudo-Simulation
End-to-end autonomous driving in urban environments requires robust decision-making under partial observability and complex multi-agent interactions. Severe occlusions and dense traffic at intersections limit the perception capability of single-agent systems, motivating recent efforts on Vehicle-to-Infrastructure (V2I) cooperation for perception and planning. However, existing evaluation protocols face a fundamental trade-off: open-loop evaluation fails to capture error accumulation and recovery from deviations, while closed-loop evaluation is costly, difficult to scale, and often relies on simulated environments that may suffer from domain gaps. To bridge this gap, we propose VIPS, a benchmark for cooperative autonomous driving in V2I settings based on pseudo-simulation. VIPS extends pseudo-simulation by integrating vehicle and infrastructure observations. This enables scalable yet realistic evaluation of robustness and error propagation without full simulation. We further present CoS-V2X, a cooperative planning framework based on sparse representations. CoS-V2X models vehicle-infrastructure interactions using compact features for efficient communication and robust decision-making under heterogeneous observations. Code and dataset are available at https://vips2026.github.io.
Read the full abstract
Quantum Physics
Accurate theoretical methods for photoionization of H$_2$ molecules
Single-photon ionization cross sections of molecular hydrogen in the electric dipole limit are numerically computed by employing three independent methods. Both time-dependent and -independent approaches within the clamped-nuclei approximation at equilibrium internuclear distance are used, featuring the explicit time-propagation of the time-dependent Schrödinger equation and newly developed multi-channel configuration-interaction free-boundary as well as complex-scaling methods using an explicitly correlated geminal basis set. The found results show convincing mutual agreement despite their entirely different fundamental formulations. They further highlight the challenges in bringing together different theoretical predictions from literature with the experimental data at high photon energies. Overall, the novel CI-based approach demonstrates fast and controllable convergence while being able to provide full channel-resolved information, indicating the need for more accurate experimental data.
Read the full abstract
Neural and Evolutionary Computing
Model-Free Surrogate-Assisted Neural Architecture Search for Evolving Variable-Length Dense Blocks
Neural Architecture Search (NAS) has emerged as a powerful paradigm for automatically designing deep neural networks; however, its practical adoption is often limited by substantial computational cost. To alleviate expensive full-training evaluations, surrogate-based methods have been introduced to estimate network performance efficiently. Nevertheless, existing approaches-particularly model-based surrogates-require training many candidate architectures and involve additional optimization overhead. In this work, we propose a Model-Free Surrogate PSO Network (MFSPNet) for evolving convolutional neural network architectures. The proposed method integrates a lightweight model-free surrogate predictor within a particle swarm optimization (PSO) framework, eliminating the need for pre-trained surrogate models. Specifically, MFSPNet introduces two key contributions: (1) a validation-loss-driven exponential moving average estimator (VLE-EMA) that captures early generalization behavior for reliable architecture ranking; and (2) a block-based dense connection strategy that enables effective stacking of evolved blocks while mitigating vanishing-gradient issues. This design also facilitates transferability of learned blocks across datasets. Extensive experiments demonstrate that MFSPNet achieves competitive performance with reduced computational cost. Under a consistent training protocol with ten independent runs, the proposed method attains error rates of 3.91%, 17.68%, and 1.91% on CIFAR-10, CIFAR-100, and SVHN, respectively, along with top-1/top-5 error rates of 28.29%/12.82% on ImageNet, while requiring less than three GPU days for architecture search. Due to computational constraints, the ImageNet result is based on a single run and should be interpreted as indicative of scalability. Overall, MFSPNet provides an efficient and reliable framework for cost-aware neural architecture search.
Read the full abstract
Artificial Intelligence
CivBench: A Long-Horizon Benchmark for Tool-Mediated Agents in Civilization VI
We present CivBench, an open-source benchmark for evaluating language model agents in long-horizon, tool-mediated environments through the Model Context Protocol (MCP). A single episode can span over 300 turns and produces thousands of tool calls over a large action space, requiring sustained planning, state monitoring, and execution under partial observability. The environment exposes 76 MCP tools and a deterministic narration layer that retrieves and formats game state as structured text. We use CivBench to characterise agent behaviour across four model families in 23 admissible runs. The sample is a pilot, not a model ranking: aggregate outcomes do not reliably discriminate models at this scale. Instead, we introduce two interface-level metrics that the environment makes measurable: Proactive Monitoring Rate (PMR), capturing whether agents actively query latent strategic state, and RAG@10, capturing whether commitments stated in structured planning reflections are executed within ten subsequent turns. Across runs we observe two consistent patterns under a shared playbook protocol. Agents under-monitor strategically relevant state that is available but requires explicit querying: despite playbook guidance to query victory progress every 20 turns, agents do so only every 30-75 turns, and in 8 of 20 detectable defeats they failed to query within the 20-turn warning window before game end. Agents also frequently fail to execute near-term commitments stated in their own planning reflections (RAG@10 between 48.2% and 65.8% across models). Both patterns arise despite tool access and explicit guidance, and we interpret them as deviations under instruction rather than absences of capability. We release the environment, scenarios, logs, metrics, and analysis pipeline at https://github.com/lmwilki/civ6-mcp.
Read the full abstract
Statistical Mechanics
Impedance in Periodically Driven Stochastic Systems
We study the time-dependent currents arising in periodically driven stochastic systems. In the linear regime of small driving, a closed expression is obtained for the impedance/admittance associated with these currents. This expression leads directly to an interpretation in terms of an equivalent electrical circuit. For a stochastic system with $N$ states, the electrical circuit consists of precisely $N$ parallel branches, with each branch comprising a resistor and a capacitor in series. The low- and high-frequency limits of these currents are calculated, and the results are generalized to more general current expressions. We demonstrate our findings across several archetypal settings, illustrating the potential to detect specific broken symmetries or the graph topology associated with the stochastic process.
Read the full abstract
Logic in Computer Science
Specification-Guided Path Shortcutting for Efficient Probabilistic Model Checking
Given the safety-critical nature of many embedded systems, their safety assurance is essential. Because such systems are typically stochastic, probabilistic model checking is a particularly important technique. However, there is a well-known scalability issue due to state-space explosion, especially when verifying complex properties. To mitigate this issue, we propose specification-guided path shortcutting for probabilistic systems, focusing on Markov chains (MCs) and $\omega$-regular properties. The key idea is that, when the verified property is fixed, certain sequences of transitions in an MC can be replaced with a single transition without changing the satisfaction probability, and thus, we can reduce the state space of the MC. We implement the proposed path shortcutting and evaluate its contribution to the performance of probabilistic model checking, using Storm as the baseline model checker. The results suggest that our approach often outperforms the baseline, particularly on benchmark instances with complex specifications.
Read the full abstract
General Relativity and Quantum Cosmology
Horizon energy fluctuations beyond Einstein's gravity: Gauss-Bonnet, Lovelock and Quantum deformed frameworks
We study thermal energy fluctuations of cosmological horizons within the canonical ensemble framework, treating the horizon as a thermodynamic system characterized by temperature and entropy. The analysis is performed for a class of gravitational theories, including (n+1)-dimensional Einstein gravity, Gauss-Bonnet gravity, Lovelock gravity, and models incorporating quantum deformed entropy corrections. We find that horizon energy fluctuations stabilize to a constant value in the asymptotic de Sitter limit, independent of the underlying gravity theory and robust against higher-curvature and quantum corrections. It is worth mentioning that in its final de Sitter state, the universe obeys holographic equipartition condition and in consequence the horizon entropy attains a maximum constant value, just like an ordinary macroscopic system. These findings establish a unified thermodynamic picture in which entropy maximization, holographic equipartition, and the suppression of energy fluctuations collectively characterize the asymptotic de Sitter universe as the equilibrium end state of cosmic evolution.
Read the full abstract
Computers and Society
Rights by Architecture: A Human-Compatible Sociotechnical Layer for Digital Protection Across Regulatory Regimes
Digital rights increasingly exist in law but remain difficult to exercise through the information systems that mediate them. Using disciplined conceptual synthesis and problematization, this critical-conceptual IS paper explains the gap through the interaction of legal heterogeneity, conflicting organizational and commercial incentives, fragmented architectures, and asymmetrical control over rights-relevant acts. It then theorizes a human-compatible rights layer: a governed sociotechnical capability for standardized, machine-readable, bidirectional, and jurisdictionally plural communication of requests, consent, refusal, withdrawal, objection, records, and support. Comparing California-style opt-out signals, the EU's mixed lawful-basis regime, and P3P, DNT, GPC, and ADPC, the paper derives seven normative requirements, develops rights by architecture as a bounded emancipatory policy argument, and treats a proposed GDPR provision on automated and machine-readable privacy management as a policy case for moving from banner-based compliance toward rights-supporting digital infrastructure.
Read the full abstract
Dynamical Systems
Exponential Consensus and Flocking in Multi-Agent Systems with Infinite Fading Memory
In this paper, we study the emergent collective dynamics of multi-agent systems driven by infinite distributed fading memory of Volterra type. We establish a unified theoretical framework covering both first-order opinion consensus dynamics and second-order velocity alignment flocking kinematics. By introducing Dafermos past-history transformations, the governing integro-differential systems are reformulated into dynamical systems on extended product Hilbert spaces. For first-order dynamics, we prove that fading memory inherently provides a hidden dissipative mechanism, guaranteeing unconditional global exponential consensus with or without instantaneous communication forces. For second-order dynamics, by developing a unified energy-multiplier framework that seamlessly couples instantaneous communication with memory dissipation, we establish unconditional global exponential flocking for both the pure fading memory system and the mixed interaction system, completely independent of the spatial dispersion or the communication tail.
Read the full abstract
Computers and Society
Addressing Trust in AI Systems through Education: A Didactic Perspective
Machine learning (ML) education faces two persistent and connected obstacles: many educational tools present ML as an opaque black box, which leaves learners with a superficial understanding, and this same opacity prevents users from forming the calibrated trust that appropriate reliance on AI systems requires. We present ICE-T, a didactic framework that integrates three mutually reinforcing facets: intermodal transfer grounded in Bruner's enactive, iconic, and symbolic modes of representation, computational thinking operationalized through the Use-Modify-Create progression, and explanatory thinking supported by a process model. Connecting the framework to the empirical literature on algorithm aversion, AI literacy, and mental model formation, and to systematic reviews of the K-12 ML activity landscape, we argue that the three facets supply the cognitive mechanisms that the trust calibration literature identifies as drivers of appropriate reliance: representational richness, graduated process control, and the capacity to contextualize errors. On this basis, we propose that trust calibration be treated as an explicit educational objective, with ICE-T as a principled and scalable means of achieving it.
Read the full abstract
Computer Vision
WiFlow: Estimating Optical Flow using WiFi Channel State Information
Knowing where and how fast objects are moving within a scene is important across various domains. Usually, cameras are used to capture the data necessary for this task, but adding cameras often raises privacy concerns, and the quality of captured frames is heavily influenced by lighting conditions. In this work, we explore using WiFi channel state information (CSI) instead of camera frames for optical flow estimation. We propose WiFlow, a CSI based flow estimator, a preprocessor evaluation for CSI, and three model architectures that offer different trade-offs between accuracy and complexity. Further, we create the first dataset for training and evaluating CSI-based optical flow estimators, and our experiments provide insights into key design elements for this task. Code and data are available at https://visinf.github.io/wiflow.
Read the full abstract
Machine Learning
Scalable Kronecker-Fisher Approximation: Efficient Hessian Analysis for Billion-Parameter Language Models Compression
In this paper, we propose a scalable Kronecker-based approximation that captures cross-layer interactions without storing the entire Fisher matrix, enabling practical Hessian analysis for billion-parameter networks where full computation is infeasible. Our approach reveals consistent vulnerability patterns: value projection layers exhibit the highest sensitivity and strongest cross-layer correlations across multiple model families, while other components exhibit architecture-specific behaviors. Through extensive experiments on quantization, sparsification, inter-layer corruption, and post-corruption fine-tuning, we demonstrate that our approximation strongly correlates with both performance degradation and recovery. Our framework provides a practical, theoretically grounded tool for identifying fragile components in large models, opening new avenues for guided compression and optimization strategies, such as mixed-precision allocation, layer-wise sparsity, and adaptive low-rank decomposition across layers and even individual weight groups.
Read the full abstract
Machine Learning
CACTUS: Mask-Guided Semantic Clean-Label Backdoors in Decentralized Federated Learning
Semantic triggers in federated learning (FL) can be less conspicuous than synthetic patches, but sample-dependent placement may weaken backdoor implantation across aggregation rounds. This challenge is compounded in decentralized FL (DFL), where topology-dependent peer aggregation repeatedly mixes local models. CACTUS converts label-consistent semantic pairs into target-directed representation shifts. Mask-guided, modality-specific operators isolate trigger effects, couple them across samples, and apply the shifts counterfactually to clean non-target embeddings before peer aggregation. Experiments cover speech, text, tabular, and image tasks under nine aggregation rules. With 30% malicious nodes, CACTUS reaches a nine-rule mean attack success rate (ASR) of 51.2% on Speech Commands and the highest nine-rule mean ASR among evaluated attacks on three of four modalities. Sensitivity analyses show that ASR varies with network topology and increases with the malicious-node ratio. These results indicate that CACTUS can propagate backdoors through repeated DFL aggregation.
Read the full abstract
Algebraic Geometry
Finite Directional Coefficient Modules over Euler Fibers of $A$-Hypergeometric Systems
Let A be a d-by-n integer matrix of rank d, let beta belong to the d-dimensional complex vector space, and consider the A-hypergeometric system defined by the ideal H_A(beta), with relation lattice L equal to the integer kernel of A. No homogeneity, pointedness, or positivity assumption is imposed on A. We study directed formal logarithmic solutions simultaneously over all exponent lattices in the Euler fiber over beta, which form an uncountable family when L has positive rank. For a fixed L-generic direction, negative-support strata are described by rational sign polyhedra, and their directional boundedness is characterized by recession cones. On each exponent lattice, normalization reduces the infinitely indexed coefficient equations to a finite system of Euler, transit, and directional vanishing relations, represented by a finitely presented directional coefficient module. We prove that only finitely many exponent lattices contribute nonzero directed solution spaces and that every contributing coefficient module has finite length. Macaulay inverse-system duality then identifies the dimension of the all-lattice directed formal logarithmic solution space with the length of the all-lattice coefficient module. We also give an explicit formal-series realization and a polynomial colon-ideal criterion for all realizable lowest exponents.
Read the full abstract
Computer Vision
Adapting a Foundation Model for Lunar Surface Height Estimation
Digital elevation models (DEMs) can provide accurate height information, making it invaluable for analyzing the lunar surface. As the European Space Agency (ESA) prepares for future lunar missions that aim to land on the Moon, a precise method for height estimation will be essential for hazardous terrain that could endanger the landing approach. Traditional approaches to generate DEMs from imagery, such as shape from shading (SfS) and stereophotogrammetry (SPG) have been proven highly valuable for this task. However, due to advancements in machine learning, especially computer vision, the focus has shifted towards monocular depth estimation via deep learning. The lunar surface is covered by rocks and craters, and classic hazard detection methods rely solely on 2D image data. Our goal is to address this issue by developing a relative lunar surface height estimator that can provide additional information for hazard localization. In this letter, we present a methodology that builds on the well-known zero-shot relative depth estimation model Depth Anything V2 (DAV2). Other works have been using it as a state-of-the-art comparison for their proposed lunar DEM estimation method, but without adaptations to the target domain. Thus, it may underperform. Therefore, we propose a fine-tuning strategy with publicly available SPG-derived DEM data of the lunar surface. Our results demonstrate a significant improvement in performance compared to the zero-shot model, effectively transforming DAV2 into a reliable relative depth estimator of the lunar surface.
Read the full abstract
Trading and Market Microstructure
Price manipulation in nonlinear transient impact models: rigidity before memory and complete positivity after memory
Transient impact models compose a nonlinearity with a memory kernel, and the order of composition determines the criterion for absence of price manipulation. We classify both orders. If an arbitrary instantaneous law $f$ acts on the trading rate before any nonzero integrable Volterra kernel, nonnegative cost on every finite piecewise-constant round trip forces $f$ to be affine, and linear for every nonzero convolution kernel. In particular, the power law $\mathrm{sgn}(x)|x|^\delta$, $\delta>0$, combined with power-law decay $t^{-\gamma}$, $0<\gamma<1$, admits manipulation if and only if $\delta\ne1$: square-root impact is manipulable at every decay exponent, and the region left open by Gatheral's slow-rate two-block bound $\delta+\gamma\ge1$ collapses to the line $\delta=1$. Earlier rigidity theorems require a kernel that is bounded at zero; the argument here is a zero-volume chattering pump read out by two thin baseline trades, and it applies to singular kernels. If instead a monotone readout acts on the impact state after the kernel, safety for all inputs and all readouts is equivalent to complete positivity of the kernel, with a constructive converse; in particular, square-root impact after power-law memory is manipulation-free. A remote compensating block shows that round-trip safety and all-input safety coincide for kernels with uniformly vanishing tails and differ, for permanent memory, by an explicit storage quotient. These mechanisms classify every two-mode Prony kernel, first-order time-inhomogeneous memory, and stable fully actuated matrix memory, and they quantify the friction, the two-block phase, and the switching complexity behind the power-law case. Calibrated exponent pairs all lie in the manipulable set: absent friction, concavity has to enter after the memory, not before it.
Read the full abstract
Logic in Computer Science
Polynomial Invariants for Probabilistic Transition Systems with Unbounded Support
We study the synthesis of polynomial invariants for probabilistic transition systems (PTS) based on martingale theory. We present tractable methods to verify that such polynomials are indeed invariants, in the sense that their expected value upon termination is the same as their value at the start of the computation. We do this by applying the Optional Stopping Theorem (OST) in the form of a specific precondition. This precondition requires the existence of an integrable dominating function for the martingale expression, which implies uniform integrability; we refer to this condition as dui. For linear PTS we simplify the dui property to proving finiteness of the expected value of an expression depending on the update matrix, the degree of the martingale expression, and the stopping time. Specifically, if all random samples have finite moments and we can verify a moment bound on the runtime of a linear loop, then we can automatically synthesise polynomial loop invariants that satisfy the OST. Notably, dui allows for the sampled distributions to have unbounded support, which is a novel contribution to the field.
Read the full abstract
Populations and Evolution
An adaptive time-tree transition kernel for Bayesian phylogenetic inference
Bayesian phylogenetic and phylodynamic analyses can be very time-consuming, owing to the combination of complex models that are used to estimate key parameters from increasingly large genomic data sets and their associated metadata. The use of high-performance computer hardware can – to a certain extent – alleviate the computational burden and markedly decrease the time to results. Still, even converging to the posterior can be a lengthy endeavour, with the burn-in aspect of such analyses potentially taking days or even weeks for large data sets. One of the key aspects that hampers performance in Bayesian phylogenetic inference is the efficiency with which tree topology proposals explore tree space. We here propose a novel adaptive tree transition kernel, which we call `subTreeLeap' (STL), which involves modifying the phylogeny by walking along patristic distance paths in the tree according to an adaptable radius parameter. STL is a general proposal, which can be used with contemporaneous or time-calibrated sequence data, being particularly suited to the latter due to respecting temporal precedence constraints. We carefully assess its impact on convergence and statistical mixing of the exploration of posterior tree space, by comparison to replicate “golden runs” obtained from lengthy analyses of empirical data under standard tree transition kernels. We find that STL successfully explores the same posterior tree space as standard kernels, but often does so in a more efficient manner. We discuss limitations as well as future potential improvements to STL that could substantially increase the speed at which Bayesian phylogenetic inferences are obtained.
Read the full abstract
Number Theory
The Hasse norm principle for $A_4$-quartic extensions of global function fields
For a finite extension of global fields $K/k$, the norm map $N_{K/k} : K^\times \to k^\times$ extends to a map on idèle groups. The Hasse norm principle holds if every element of $k^\times$ that is a norm everywhere locally is also a norm globally. In this paper, we study the statistics of the Hasse norm principle in a setting that is out of reach in the number field context, namely that of $A_4$-quartic extensions. We show that failures of the Hasse norm principle are generally rare for $A_4$-quartic extensions of global function fields $\mathbb{F}_q(t)$. We achieve this by introducing a decorated Hurwitz space parametrising the failures of the Hasse norm principle and then using the Chebotarev density theorem to estimate their frequency.
Read the full abstract
Strongly Correlated Electrons
Thermodynamic Electric Toroidal Dipole and Intrinsic Longitudinal Spin Transport
Electric toroidal dipoles (ETDs) characterize ferroaxial order, yet their bulk definition in periodic crystals has remained elusive because conventional multipole operators involve the ill-defined position operator. Here we formulate a thermodynamic ETD by coupling a spatially varying electric field to the relativistic spin-induced electric polarization. The resulting expression is gauge invariant and provides a bulk order parameter for ferroaxial phases. We further establish a direct relation between the chemical-potential derivative of the ETD and the intrinsic longitudinal spin conductivity in insulating systems. To demonstrate the formulation, we construct a minimal ferroaxial extension of the Kane–Mele model. The ETD becomes finite exclusively in the ferroaxial phase and is strongly enhanced near a small band gap, accompanied by a sizable longitudinal spin current. Our results establish a thermodynamic theory of ETDs in crystalline solids and identify the longitudinal spin conductivity as a direct transport manifestation of ferroaxial order.
Read the full abstract
High Energy Astrophysical Phenomena
Transient discs around isolated accreting neutron stars
Mature, isolated neutron stars can accrete from the interstellar medium. Due to turbulence, the accreted gas can have a substantial angular momentum and form a disc around the compact object. In this paper, we perform a population synthesis of isolated neutron stars in the Milky Way that specifically tracks the possibility of disc formation, which typically requires a low spatial velocity of the compact object ($\lesssim 40$ km s$^{-1}$). In general, weak magnetic fields favour disc formation as the magnetosphere occupies a smaller volume. Still, even in the case of fields decaying exponentially over a characteristic timescale of $\sim 1.5$ Gyr, we find that for several realistic models of propeller spin-down, only a small fraction of accretors (down to $\sim 0.02$ %) attain discs. However, disc-accreting isolated neutron stars are relatively numerous among the brightest sources. We estimate that their number can reach a few hundred at X-ray fluxes of $\gtrsim 10^{-14}$ erg s$^{-1}$ cm$^{-2}$. We speculate that disc-accreting isolated neutron stars can manifest as long-period radio transient sources via electron cyclotron maser emission, predicting spectra and critical conditions for quenching.
Read the full abstract
Nuclear Theory
Effects of Strong Magnetic Fields on the Equation of State and Mass-Radius Structure of Hyperonic Neutron-Star Matter with Anomalous Magnetic Moments
We investigate the effects of strong magnetic fields on the equation of state (EoS) and stellar structure of cold, charge-neutral, $\beta$-equilibrated hyperonic neutron-star matter within a relativistic mean-field (RMF) framework. The matter sector contains the full baryon octet and leptons, while charged particles are Landau quantized and all baryons are coupled to the magnetic field through their anomalous magnetic moments (AMM). The calculation is performed with the RMF FSU2H hyperonic parameterization and compared for zero field, constant magnetic fields, and density-dependent magnetic-field profiles. We find that strong magnetic fields modify the hyperonic composition through the competing effects of Landau quantization and AMM-induced spin splitting. Landau quantization softens the magnetized hyperonic equation of state. The inclusion of AMM provides an additional magnetic stiffening mechanism in hyperonic matter. The results for the inclusion of the AMM coupling and not are still consistent with observations of $2.00$-$2.02\,M_{\odot}$ neutron stars and small radii. The present work therefore provides a benchmark for assessing the influence of AMM on the composition, magnetization, equation of state, and mass-radius structure of magnetized hyperonic neutron-star matter.
Read the full abstract
Machine Learning
Towards One-for-All Robustness Across a Continuum of Threat Levels
Adversarially robust models often overfit to a specific attack budget, necessitating multiple specialized models for diverse and dynamic adversarial environments, a strategy that becomes fundamentally intractable as the threat space grows. This raises an open challenge: can we achieve strong robustness across a continuum of threat levels within a single model? We propose the Threat Conditional Network (TCN), grounded in a representation factorization framework that decomposes representation learning into a threat-invariant shared backbone and a lightweight threat-conditional adaptor. TCN conditions a single model on the perturbation level via Fourier-based embeddings and channel-wise affine modulation, and is trained against a distribution over perturbation budgets, enabling flexible and seamless adaptation across an infinite continuum of threat levels during inference. Extensive experiments on CIFAR-10, CIFAR-100, and Tiny-ImageNet show that TCN matches or surpasses a full ensemble of budget-specialized models with a single set of parameters, generalizes to unseen perturbation budgets, and transfers robustly under mismatched threat conditions, with only 4.6% parameter overhead. These contributions chart a promising path toward adaptive and generalizable robustness in dynamic and diverse threat environments.
Read the full abstract
Metric Geometry
Dual Geometry of Spherical Designs: Polarity, Self-Polar Rigidity, and Quadrature Structure
Let $X=\{x_1,\ldots,x_N\}\subset\Sph^{d-1}$ be a spherical $t$-design, $t\ge2$, and let $P_X\subset\R^d$ be its Minkowski polytope. If $h_i=h_{P_X}(x_i)$, then \[ P_X^\circ=\conv\{x_i/h_i:1\le i\le N\}, \] so the design is the radial projection of the polar vertex set. This yields exact degree-dependent moment identities and quantitative control of the unweighted polar moments from the Hausdorff sphericity of $P_X$. Our main results concern self-polarity. If \[ P_X^\circ=cUP_X,\qquad U\in O(d), \] we obtain the structured slack factorization \[ A=c\,hh^T-X^TU^TX,\qquad \rank A=d+1, \] whose zero pattern records the facet–vertex incidences. For node-transitive designs, the common incidence level equals the inradius-to-circumradius ratio $r/R$. Combining this with the one-dimensional moment problem underlying the Fazekas–Levenshtein covering bound, we prove \[ \frac rR\ge\eta_{t,d}, \] with equality forcing the twisted inner-product rows to realize the corresponding Gaussian or Gauss–Radau quadrature rule; in particular, the quantities $N\lambda_k$ become integer incidence multiplicities. We further prove a quantitative near-equality theorem: if \[ \delta=\frac rR-\eta_{t,d} \] is small, then each row is $O_{d,t}(\sqrt\delta)$-close in $W_1$ to the extremal quadrature measure, yielding near-incidence rigidity and an arithmetic stability gap. As applications, we obtain three-dimensional rigidity and identify the regular simplex and the $24$-cell as the regular self-polar examples.