Number Theory
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 11 September
We determine all the groups that appear as the torsion group of an elliptic curve over a quintic number field. Apart from the groups that already occur infinitely often, which were determined by Derickx and Sutherland, exactly three groups occur: $\mathbb{Z}/28\mathbb{Z}$, $\mathbb{Z}/30\mathbb{Z}$ and $\mathbb{Z}/2\mathbb{Z}\times\mathbb{Z}/18\mathbb{Z}$. Up to isomorphism of the pair $(K,E)$, the first and the third are each realised by a single elliptic curve and the second by two curves, which are $2$-isogenous over a common quintic field. The curves realising $\mathbb{Z}/28\mathbb{Z}$ and $\mathbb{Z}/30\mathbb{Z}$ were found by van Hoeij, while the group $\mathbb{Z}/2\mathbb{Z}\times\mathbb{Z}/18\mathbb{Z}$ is new, and $5$ is the smallest degree in which a non-cyclic sporadic torsion group occurs. The methods improve on those developed by Derickx and Najman, and are based on Hecke sieves and the arithmetic of cuspidal divisor classes.
Audio and Speech Processing
A Device to Control and Manipulate Occlusion Effects for Own Voice Perception Studies
The occlusion effect (OE) refers to changes of the eardrum sound pressure through ear canal occlusion. It consists of two phenomena: an insertion loss (IL) attenuating air-conducted sounds, and an occlusion gain (OG) amplifying bone-conduction. Perceptual research on this is hindered by high variability of the OE across individuals, complicating repeatable presentation of precise OE conditions. Consequently, a method to control OE conditions reproducibly during perceptual experiments is needed. For such investigations, the system must enable separate control of IL and OG. We present an approach based on modified commercial earmuffs with integrated microphones. The design integrates dedicated impedance measurements and the derivation of digital filters, which are applied to the microphone signals to emulate arbitrary OE curves. The system is evaluated objectively through appropriate measurements. Results show that the headphones' inherent OG is guaranteed to be below 6dB above 115Hz, falling below 0dB above 155Hz. Emulation is shown to work accurately over the entire frequency range of interest. Limitations arise mainly due to a slight residual inherent OG for deep voices and the processing delay of the system. Future work will focus on the perceptual evaluation of the system and its application in perceptual studies.
Read the full abstract
Classical Analysis and ODEs
Hardy–Littlewood Maximal Operator and Two-Layer Muckenhoupt Weights on Infinite Rooted $k$-Ary Trees
Let $k\geq 2$ be an integer, $T$ an infinite rooted $k$-ary tree, and $M$ the Hardy–Littlewood maximal operator on $T$. For any $p\in(0,\infty)$, we characterize the weight $w$ such that $M$ is bounded on $L^p(w)$. To this end, we introduce a two-layer Muckenhoupt weight class $\mathscr A_p$ and prove that, for any $p\in(\frac{1}{2},\infty)$, the boundedness of $M$ on $L^p(w)$, $w\in\mathscr A_p$, and the exponential decay boundedness of spherical averaging operators on $L^p(w)$ are mutually equivalent, and that, when $p\in(0,\frac{1}{2}]$, there exists no weight $w$ such that $M$ is bounded on $L^p(w)$. Moreover, for any $p\in(\frac{1}{2},\infty)$, we establish the quantitative estimate, with the optimal exponent $\frac{1}{p}$ of the weight constant, for the boundedness of $M$ on $L^p(w)$. For any $p\in(1,\infty)$, we also obtain two further equivalent characterizations of the boundedness of $M$ on $L^p(w)$, respectively, in terms of a global Sawyer-type testing condition and an estimate for the weighted product measure of distance incidence sets. As applications, for any $p\in(\frac{1}{2},\infty)$, under the assumption that $M$ is bounded on $L^p(w)$, we establish the boundedness of exponentially decaying kernel operators on $L^p(w)$ and weighted Fefferman–Stein vector-valued inequalities.
Read the full abstract
Computer Vision
Pre-Trained Low-Rank Tensor Decomposition for Multi-Dimensional Image Recovery
Recently, tensor decompositions are prevalent for multi-dimensional image representation, which learn the instance-specific structure of each image from scratch. However, tensor decompositions neglect the common structure across different images, leading to limited semantic modeling capability, high computational cost, and a large number of learnable parameters. To address this challenge, we suggest the first pre-trained low-rank tensor decomposition (PLTD) framework, which organically integrates the pre-trained large vision model into the classical tensor decomposition framework. Beyond the shallow and untrained deep tensor decomposition, the suggested PLTD achieves an unprecedented balance among higher recovery fidelity, fewer learnable parameters, and smaller carbon footprint. Specifically, PLTD factorizes the target tensor into a latent tensor and a learnable transform that maps the latent tensor back to the original data domain. The latent tensor consists of two indispensable and complementary terms, i.e., a fixed pre-trained latent tensor and a learnable low-rank latent tensor. The fixed pre-trained latent tensor is distilled from a pre-trained large vision model (i.e., DINOv3) to capture the common structure of the target tensor, while the learnable low-rank latent tensor characterizes the instance-specific structure of the target tensor. To examine the potential of PLTD, we develop the corresponding multi-dimensional image recovery model and theoretically justify the advantages of this framework. Additionally, we discuss the connections between PLTD and classical tensor decomposition frameworks. Extensive experiments on multi-dimensional image recovery demonstrate that PLTD consistently achieves superior performance compared with state-of-the-art methods.
Read the full abstract
Information Retrieval
MIMA: Multi-Interest Recommendation via Multi-Positive Exclusive Assignment
Multi-interest recommendation represents each user with multiple interest vectors for fine-grained candidate matching, yet it often suffers from interest collapse, where the learned interests converge to similar representations. We highlight the prevailing single-positive paradigm as one important factor behind this issue. Since each instance provides only one positive item, intents are optimized independently, potentially causing the same best-matching interest to be repeatedly updated toward different positives while leaving the others under-supervised. Moreover, existing methods rarely model how strongly a user activates each interest, leaving scores from different interest channels incomparable at inference. To address these problems, we propose MIMA, a Multi-Interest recommendation framework built on Multi-positive exclusive Assignment. MIMA groups items co-occurring within the same request into a positive set, generates complementary interests with a causal Transformer decoder, and exclusively assigns each positive to supervise a distinct interest via Hungarian matching, so that interest differentiation emerges from the training objective itself rather than auxiliary regularization. A lightweight routing module further estimates user-interest activation probabilities to calibrate scores across interest channels. Experiments on three public datasets and an industrial dataset show that MIMA consistently outperforms state-of-the-art baselines, and an online A/B test yields significant business gains.
Read the full abstract
Cryptography and Security
A Graph-Based Approach for Mapping Kernel-Level Telemetry to MITRE ATT&CK
Mapping observed system behavior to standardized frameworks such as MITRE ATT&CK is essential for threat-informed defense. Existing automated methods mostly depend on Cyber Threat Intelligence reports, which offer only retrospective accounts of attacks. Kernel-level telemetry, i.e., system calls, instead provides first-hand evidence of adversary behavior, but its volume and complexity have limited its use for automated mapping. Its content also makes it unsuitable for cloud-based analysis: command lines carry credentials, events carry user identifiers and network endpoints, and during an incident the telemetry reveals which defenses failed. In this work, we present Trace2ATT&CK, a framework that maps kernel-level telemetry to ATT&CK techniques entirely within the defender's infrastructure. Trace2ATT&CK collects events via eBPF, reconstructs the attacker's activity as a process-centric provenance graph, and condenses it into a compact execution graph. This graph is classified by locally deployed open-weight LLMs, using either pure prompting or retrieval-augmented generation (RAG) grounded in the ATT&CK knowledge base, producing ranked technique candidates with supporting rationales. We evaluate Trace2ATT&CK on 347 Linux Atomic Red Team tests with seven open-weight LLMs. The best local configuration reaches an HR@5 of 66.57% at the technique level. RAG improves over base prompting in every model and setting, by up to $+15.44$ percentage points, and the execution graph substantially outperforms raw telemetry while never exceeding the LLM context window. Finally, a frontier cloud model used as an upper-bound reference quantifies the cost of local inference at 24–28 percentage points. These results show that ATT&CK mapping from kernel-level telemetry is feasible without disclosing sensitive evidence to third-party services.
Read the full abstract
Quantum Physics
State-adapted generalized mean-field projections for Pauli propagation of Heisenberg dynamics
The Heisenberg picture can make quantum many-body simulation efficient when evolved observables admit compact operator representation within low-dimensional structures. Conventional Pauli-string propagation and truncation techniques exploit this structure, but in coherent Hamiltonian dynamics their error control is often heuristic and their stability can be poor. We introduce a state-adapted Krylov framework based on geometric generalized mean-field projections of many-body observables onto low-body operator subspaces. The resulting dynamics approximate expectation values, do not extend spatial support beyond that prescribed by Lieb-Robinson bounds, and compress high-body correlations onto their state-relevant low-body representatives rather than simply discarding them. We derive necessary operator-entanglement obstructions to low-body representation and conditional sufficient bounds on representation and dynamical errors involving nonstabilizerness and controlled high-body tails. Numerical benchmarks show improved, stable finite-$m$ hierarchies and simulations on large three-dimensional lattices, establishing a scalable state-adapted alternative to conventional Heisenberg-picture weight-truncation of Pauli strings.
Read the full abstract
Cryptography and Security
Evaluating Context Segmentation in Locally Deployable SLMs for Cybersecurity CTF Tasks
The proliferation of highly capable open-weight Small Language Models (SLMs) democratizes access to advanced cybersecurity capabilities, posing an escalating risk as these models can bypass proprietary API guardrails when deployed locally. However, SLMs deployed as autonomous agents often struggle with long-horizon, exploratory tasks like cybersecurity Capture The Flag (CTF) challenges due to context bloat and cognitive degradation from accumulated tool-call outputs. To understand and mitigate this cybersecurity threat, we introduce context segmentation, a two-level agentic framework that divides complex exploitation tasks into manageable, contextually isolated sub-problems. Evaluating on the picoCTF dataset using memory-constrained gemma-4 models, we demonstrate that for the E4B model, our strategy acts as an intelligent search, achieving competitive rewards with superior token efficiency compared to brute-force retries, and successfully solving 18.52% of tasks that standard agentic execution fails to complete. Code is available at https://github.com/9xeb/context-segmentation.
Read the full abstract
Number Theory
Projective Kernels and Replicability in Modular Function Theory
We introduce a projective-kernel framework for the study of replicable functions in modular function theory. The main point is that the same two-point kernel simultaneously encodes the differential projective geometry of a modular function and the Faber–Grunsky data governing its replicability. This makes it possible to translate between Schwarzian invariants, coefficient identities, and modular correspondences within a single structure. The kernel yields a reconstruction theorem showing that the ordinary Schwarzian determines the normalized Grunsky matrix and hence the underlying Laurent expansion. When Norton's replicability relations are imposed, the projective kernel produces strong arithmetic restrictions on the possible cusp data. In the degree-one case this leads to a precise classification: replicability and complete replicability are equivalent to solvability of the projective monodromy, while the icosahedral case is excluded by an explicit Grunsky obstruction. The same principle extends to the arithmetic Hecke triangle groups. The projective-kernel viewpoint also isolates the remaining global difficulty in Norton's Hauptmodul conjecture and provides a natural setting in which replicability, projective monodromy, and modular differential invariants can be studied together.
Read the full abstract
Robotics
Parameter Sensitivity Analysis for Aerial LiDAR-Inertial Odometries in low-altitude flights
LiDAR-based SLAM (Simultaneous Localization and Mapping) and LIO (LiDAR-inertial odometry) algorithms are often used for precise navigation of unmanned aerial vehicles, especially during interactions with the aerial robot's environment. However, the performance of these algorithms is greatly dependent on the scenario, LiDAR, and robot motion characteristics, often requiring an intensive tuning process to achieve the desired performance. To aid these tuning efforts, this paper analyzes the influence on performance of the parameters of an EKF-based LIO algorithm (FAST-LIO2) and the LIO module of a graph-based SLAM algorithm (Cartographer) on aerial LiDAR SLAM datasets recorded using different LiDARs in low-to-moderate-altitude flights in diverse environments. The analysis is conducted on the absolute trajectory error (ATE) resulting from processing the datasets with the LIO algorithms configured with each combination of parameters obtained in an exhaustive grid search. The relationship between individual parameters and the ATE results is assessed using Pearson's correlation, while the influence of each parameter is assessed using random forest permutation importance analyses with random forest models trained to predict the resulting ATE values based on the choice of parameters. The performed analysis obtains for Cartographer and FAST-LIO2: i) the identification of parameters with stronger influence in performance, ii) a simplified tuning procedure, and iii) tuning recommendations. Using the proposed tuning recommendations, both algorithms obtain on the analyzed datasets ATE values within 5 cm to the optimal performance found in the grid search procedure in 94% of the analyzed cases.
Read the full abstract
Materials Science
Ultrastrong Au-Te bonding drives disorder in monolayer ZrTe$_5$ on gold
Monolayer ZrTe$_5$ is predicted to host a large-gap quantum spin Hall phase, motivating efforts to isolate single layers of the material. Gold-assisted exfoliation produces clean monolayers of many chalcogen-terminated van der Waals crystals, but the strong Te-Au bond may compete with the bonding network of the ZrTe$_5$ layer itself. Across tens of samples, low-temperature scanning tunneling microscopy on gold-exfoliated flakes shows a disordered monolayer surface in the overwhelming majority of cases, while thicker flakes preserve the characteristic quasi-one-dimensional chain structure. Ab initio calculations of the ZrTe$_5$/Au(111) interface reproduce this asymmetry and resolve its mechanism: the Te atoms facing the gold chemisorb, rupturing the weak zigzag Te-Te bonds that cross-link the ZrTe$_3$ chains, turning the discrete bond-length spectrum of the crystal into a continuous one, while the chains persist as distorted units. Charge analysis shows the signature of covalent Te-Au bonding with a modest transfer that hole-dopes the monolayer; in the bilayer, the distortion and the doping stay confined to the layer contacting the gold, and the chemisorption and disorder each preclude the predicted quantum spin Hall phase. The same quasi-covalent Te-Au bond may perturb the contact layer of other gold-exfoliated Te-terminated crystals, particularly those with weakly connected intralayer networks.
Read the full abstract
Computer Vision
HemaHier: Chain-Conditioned Ordinal Hierarchies for Lineage-Aware Bone-Marrow Cytology
Bone-marrow cytology is inherently structured: each cell belongs to a hematopoietic lineage, and many cell types lie on ordered maturation trajectories. Standard flat classifiers ignore this structure, treating a mild same-lineage confusion the same as a severe cross-lineage mistake and predicting only discrete labels. We propose HemaHier, an ordinal-hierarchical prediction head for a frozen or lightly adapted cytology foundation model. Its central component is a chain-conditioned maturity score that reads a single maturity value under a per-chain query, supervised only on biologically valid healthy chains, while dysplastic and off-chain cell types remain classes but are excluded from maturity supervision. Fine and lineage predictions are coupled through a shared posterior that guarantees hierarchical consistency, and a staged objective first stabilizes recognition, then adds lineage and maturity supervision. On three bone-marrow datasets under a shared ontology, HemaHier achieves competitive recognition while reducing biologically severe errors and adding a within-lineage maturity ordering that flat classifiers lack. Code is available at https://github.com/xmindflow/HemaHier.
Read the full abstract
Computer Vision
Self-supervised Pre-training Helps Retinal Disease Progression Modelling Most When Data Is Scarce
Modelling how a disease progresses over time requires longitudinal imaging cohorts, which are scarce and small, whereas cross-sectional data – one image per participant – is abundant. Self-supervised pre-training on such data offers a way to bridge this gap, but it is unclear which strategy best supports progression modelling, or how that answer depends on the amount of labelled longitudinal data. We study this for age-related macular degeneration (AMD), pre-training encoders on the large cross-sectional NAKO cohort and predicting time to late AMD on the longitudinal AREDS dataset. We compare in-house self-supervised encoders against a general-purpose (DINOv2) and a domain-specific (RETFound) foundation model, across contrastive, masked-autoencoding, and self-distillation objectives, under frozen and fine-tuned protocols, and across labelled training sets from 100 to 32,250 examples. Which model performs best depends on how the encoder is used. When the encoder is frozen and labels are few – the regime typical of longitudinal cohorts – pre-trained representations reach clinically reasonable discrimination from a few hundred labelled samples, while models trained from scratch do not; this advantage fades under fine-tuning. Transfer is governed by the self-supervision objective rather than corpus scale or domain match, so that an encoder pre-trained on a modest cross-sectional cohort matches or exceeds a far larger in-domain foundation model. Together, these results offer a practical recipe for building progression models where longitudinal data is scarce: a frozen self-supervised encoder with a lightweight survival head.
Read the full abstract
Quantum Physics
Fast quantum squeezing of a nanomechanical oscillator with an inverted potential
Nonclassical states of nano- and micro-mechanical motion enable measurements beyond the standard quantum limit and constitute a key resource for quantum sensing and metrology. The most strongly squeezed mechanical states to date have been generated with electromechanical platforms under cryogenic refrigeration and using reservoir engineering. To push mechanical systems deeper into the quantum-squeezed regime requires protocols that increase the rate at which squeezing is generated to more strongly overcome the decoherence rate at which state purity is lost. Here, we squeeze the 800 kHz libration mode of a silica nanoparticle optically levitated in vacuum at room temperature. We expose our mechanical oscillator to an optically generated inverted potential, where the squeezing operation proceeds at an exponentially accelerated rate. We reach a squeezed quadrature variance 11 dB below the vacuum fluctuations within 250 ns. Our protocol establishes a new paradigm for generating quantum squeezing of mechanical motion and offers a platform for quantum-enhanced sensing with massive oscillators.
Read the full abstract
Mesoscale and Nanoscale Physics
Superfluorescence in CdS/CdSe/CdS Spherical Quantum Wells Modulated by Excitation Geometry
Superfluorescence (SF) originates from the spontaneous buildup of macroscopic coherence among initially incoherent emitters. Its formation is highly sensitive to dephasing and collective coupling in the system. Here, we observe room-temperature SF in CdS/CdSe/CdS spherical quantum wells. With increasing excitation fluence, the peak emission intensity shows a nearly quadratic increase, while the emission delay and pulse width decrease. Burnham-Chiao ringing is also observed, revealing the characteristic collective radiation dynamics of SF. Further experiments with stripe excitation show that shortening the excitation length L changes the dominant fast emission from amplified spontaneous emission to SF. More importantly, the threshold for SF under stripe excitation is only about 1/30 of that under spot excitation. This large reduction indicates that the propagating radiation field provided by spatially extended excitation favors the buildup of cooperative coherence. These findings provide direct experimental evidence for understanding ultrafast many-body coherence dynamics and offer a route to actively control collective emission at room temperature.
Read the full abstract
Robotics
VertexCBF: Improving Neural Control Barrier Functions via Vertex-Restricted Control Search
As the number of autonomous robots continues to grow, safety becomes increasingly important. Control barrier functions (CBFs) provide a theoretically grounded framework for ensuring safety, but existing design methods often face limitations in effectiveness, scalability, or interpretability, and may result in overly conservative safe sets. In this paper, we propose VertexCBF, a framework for learning neural CBFs in a scalable, systematic, and explainable way. We approximate the stationary Hamilton–Jacobi value function using a neural network trained via a combination of physics-informed and sparsely supervised learning. By exploiting control-affine dynamics and a convex polytope control set, under which the Hamiltonian is maximized at the control vertices, we efficiently generate supervision points via GPU-parallel vertex-restricted tree search, while a residual architecture guarantees that the learned CBF is never larger than the specified constraint function. We evaluate the method on 15 systems and compare it against relevant baselines, showing that it reliably recovers large safe sets where the baselines are conservative or fail completely. In addition, we perform a hardware experiment in which a mobile robot safely avoids pedestrians using a neural CBF trained with our method.
Read the full abstract
Computer Vision
Balancing Emotional Alignment and Semantic Consistency in Image Generation via Reinforcement Learning with Valence-Arousal Anchoring
Continuous emotion control in text-to-image generation requires a model to improve affective alignment without changing the objects, layout, or scene described by the prompt. Existing supervised emotion-injection methods often optimize feature-space proxies and may therefore exhibit emotion-semantic drift, in which stronger emotional conditioning is accompanied by unintended content changes. We address this problem with a flow-matching image-generation framework that combines continuous valence-arousal (VA) conditioning, Group Relative Policy Optimization (GRPO), and a neutral semantic anchor. The deterministic probability-flow ODE is converted into a marginal-preserving SDE, yielding non-degenerate transition densities for trajectory sampling and policy-ratio estimation. A frozen CLIP-based VA regressor supplies a terminal reward measuring the distance between the predicted and target VA coordinates, while an image generated from the same prompt under zero VA conditioning provides a feature-space reference for semantic preservation. A reduced denoising schedule is used for online RL sampling, whereas the original schedule is retained at inference. Experiments on 3,300 prompt-emotion combinations show substantially lower valence and arousal errors than the VA-conditioned baseline and an improved CLIPScore relative to EmotiCrafter, with a measurable trade-off in reference-free image quality. The results support anchor-regularized Flow-GRPO as a practical approach to balancing emotional alignment and semantic consistency in continuous-affect image synthesis.
Read the full abstract
Game Theory
Learning-Augmented Strategyproof Facility Location in $\mathbb{R}^d$ with $\ell_p$ Distances
We study learning-augmented mechanism design for locating a single facility in $\mathbb{R}^d$ to minimize the sum of the agents' $\ell_p$ distances to the facility. We analyze the coordinate-wise median with predictions (CMP) mechanism, which adds $cn$ virtual agents at a predicted optimal facility location to the $n$ reported locations and returns their coordinate-wise median. The parameter $c\in[0,1)$ represents confidence in the prediction; CMP is known to be strategyproof when both are fixed independently of the reports. We determine its approximation guarantees under correct predictions (consistency) and arbitrary predictions (robustness) in two settings. First, for $d=2$, we establish exact guarantees for every $p\in[1,+\infty]$. For $1<p<+\infty$, the consistency and robustness are respectively $\Bigl(1+\bigl(\frac{1-c}{1+c}\bigr)^{\frac{p}{p-1}}\Bigr)^{\frac{p-1}{p}}$ and $\Bigl(1+\bigl(\frac{1+c}{1-c}\bigr)^{\frac{p}{p-1}}\Bigr)^{\frac{p-1}{p}}$. We use the median condition in each coordinate to compare the mechanism's cost with the optimal cost, and establish tightness using instances with agents at only three distinct locations. Second, for $1<p<+\infty$, we obtain dimension-independent consistency and robustness upper bounds valid for every $d\ge1$. For each fixed $p$ and $c$, we construct families of instances whose ratios approach the respective upper bounds as $d\to\infty$, proving asymptotic tightness. We also establish exact guarantees for $p=1$ in every dimension and asymptotically tight bounds for $p=+\infty$. Our high-dimensional results recover the prediction-free bounds of Gravin and Jia (STOC 2025) when $c=0$ and their learning-augmented bounds in arbitrary-dimensional Euclidean spaces when $p=2$.
Read the full abstract
General Relativity and Quantum Cosmology
Nonlinear Degrees of Freedom of Born-Infeld New Massive Gravity
Born-Infeld New Massive Gravity (BINMG) is an all-order curvature extension of New Massive Gravity in three dimensions, whose linearized spectrum around its unique maximally symmetric vacuum contains the two polarizations of a massive spin-two field. Whether an additional scalar degree of freedom appears in the full nonlinear theory has remained an open question. We perform a nonlinear canonical analysis of BINMG using its auxiliary-metric formulation. On the regular branch connected to the maximally symmetric vacuum, the Dirac algorithm yields three first-class diffeomorphism constraints and a second-class pair of scalar constraints. The resulting count gives exactly two local degrees of freedom per spatial point. Thus, nonlinear BINMG propagates no additional Boulware-Deser-type scalar mode on this branch.
Read the full abstract
Computer Vision
A Dual Cross-Attention Framework for Colposcopic CIN Grading and Swede Score Prediction Using a New Multi-Center Dataset
Cervical cancer is a major global health challenge, with disease burden falling disproportionately on low- and middle-income countries (LMICs) due to a shortage of trained specialists and the subjective nature of colposcopy-based screening. To address this challenge, we propose a novel deep learning framework for the automated grading of Cervical Intraepithelial Neoplasia (CIN) and the prediction of clinical Swede scores. We also introduce the BUET Multi-Center Colposcopy Dataset, a novel, multi-center cohort designed and annotated for Swede score prediction and CIN grading. Our proposed dual-stream cross-attention architecture mimics the visual reasoning of an expert colposcopist by explicitly fusing paired multimodal cervigrams to evaluate comparative tissue responses. Furthermore, we introduce a custom composite loss function to address severe class imbalances and scoring inconsistencies across the five Swede score components. The proposed framework achieved 71.85% accuracy and an 86.23% AUC-ROC for three-class CIN grading, outperforming existing methods. For Swede score component prediction, the architecture achieved AUC-ROC values ranging from 75.7% to 88.4%, with the composite loss function yielding consistent F1-score improvements. Finally, the total predicted Swede Score, which ranges between 0 and 10, shows a Mean Absolute Error (MAE) of 1.489. The results show that the proposed method can pave the way towards developing AI-assisted colposcopy screening tools to support risk-based triage in resource-limited healthcare settings. The dataset and source code are publicly available(url: https://github.com/mHealthBuet/BUET-colposcopy)
Read the full abstract
Computer Vision
CoralscapesV2: Panoptic and Fine-Grained Visual Scene Understanding in Coral Reefs
In order to design conservation and restoration strategies to counter the global decline of coral reefs, ecological monitoring of reefs needs to be scaled up dramatically. Computer vision methods are increasingly used to tackle the vast amount of data: as the paradigm of data collection in reefs shifts from highly standardized and constrained survey images to unconstrained imagery on scalable platforms, it is necessary to design machine learning methods that help to get a fine-grained understanding of reefs from general-purpose reef imagery. This paper provides CoralscapesV2, an extension of the Coralscapes dataset for general-purpose visual scene understanding in reefs. CoralscapesV2 increases the dataset size, scope, label completeness and quality for semantic segmentation, and extends the number of classes from 39 to 95 fine-grained visual categories. Furthermore, CoralscapesV2 provides 65k exhaustive fish instance mask annotations, meticulously annotated to completeness by using the video, revealing that annotation of fish based on only static images is insufficient. CoralscapesV2 is the first dataset for panoptic segmentation in coral reefs, capturing a wide range of scenarios in the wild, posing a challenging benchmark for contemporary semantic segmentation and instance segmentation models. CoralscapesV2 is an important step towards general-purpose panoptic segmentation in coral reefs, which has substantial implications for scaling up coral reef monitoring, as it can be employed in a wide range of applications from benthic cover mapping from robot or handheld videos to designing methods for automated quantification and understanding of fish behavior and fish-reef interactions.
Read the full abstract
Computer Vision
SCDM: Spatial-Contextual Disentanglement Mamba via Differential Inference for Efficient Image Classification
State Space Models (SSMs), particularly VMamba, have emerged as efficient alternatives for modeling long-range dependencies in medical image analysis. However, distinguishing subtle pathological features from visually similar anatomical backgrounds remains a significant challenge. Existing SSM architectures often learn entangled representations, lacking explicit mechanisms to separate disease-specific signals from normal anatomy. To address this limitation, we propose Spatial-Contextual Differential Mamba (SCDM), an asymmetric dual-branch architecture designed for selective representational disentanglement. SCDM introduces a Positive Branch for extracting discriminative features and a Negative Branch that actively models and suppresses normal anatomical context. This separation is achieved through a similarity-driven repulsion gate and a differential inference rule, which promote competitive feature learning without requiring additional branch labels or increasing model capacity. Evaluated on the RSNA Pneumonia dataset, SCDM achieves competitive classification performance (AUC of 0.858) while requiring significantly fewer parameters (29.4M) and FLOPs (1.44G) compared to standard VMamba and vision transformer baselines. Furthermore, activation analyses demonstrate that our differential mechanism yields highly precise localization, effectively isolating lesions by inhibiting irrelevant anatomical distractors.
Read the full abstract
Fluid Dynamics
Newtonian fluid invasion into a Yield-Stress Fluid-Saturated Medium
We investigate pressure imposed two-phase flows in porous media where a Newtonian fluid displaces an immiscible Bingham yield-stress fluid. Using a pore-network model, we identify invasion patterns governed by three dimensionless parameters: the pressure-based capillary number $Ca_P$, the viscosity ratio $M$ , and a number $R_\tau$, the ratio of yield stress to capillary forces. Beyond the classical Lenormand regimes (viscous fingering, capillary fingering, stable displacement), we discover two new regimes: a tree-like pattern at low viscosity ratios and a progressively widening column-like pattern at high viscosity ratios. We characterize all regime transitions and establish a breakthrough condition for the invading fluid.
Read the full abstract
Optimization and Control
Continuous-Time Constrained Linear Quadratic Regulator for Switched Linear Systems
Switched systems, characterized by a family of subsystems governed by a switching rule, widely apply to complex real-world scenarios. However, their inherent switching dynamics pose significant challenges in phase analysis and control design. To address these challenges, we propose a framework for solving the Continuous-Time Constrained Linear Quadratic Regulator (CT-CLQR) problem for switched linear systems. Our approach partitions the time horizon into a finite number of intervals, each associated with a specific system mode. The duration of these intervals is parameterized by the switching instants, enabling a reformulation of the problem. We indirectly optimize the switching sequence by fixing the switching sequence and optimizing the interval durations. We derive analytical expressions for the cost function and its gradient, which are critical for efficient optimization. Unlike state-of-the-art methods that impose equality constraints on state evolution, our approach inherently considers the state evolution in the cost function. This not only simplifies the problem formulation but also reduces computational overhead by precomputing shared terms offline, enhancing efficiency during online operations. The proposed method significantly advances existing techniques, offering improved computational efficiency and flexibility. We demonstrate the effectiveness of our approach through comprehensive numerical examples, showcasing its potential for practical applications.
Read the full abstract
Artificial Intelligence
Scaling Clinical Judgment to Evaluate Medical AI
Blinded physician evaluation has been considered by many to be the gold standard for assessing clinical reasoning in large language models (LLMs). This is difficult to scale; thus, prior studies typically rely on small physician panels, often from a single institution or specialty, which both limits the scientific questions investigated and makes it unclear whether findings would be reproduced with a different set of evaluators. To more rigorously and scalably study clinical reasoning in AI models, here we introduce PrecepTron, an LLM fine-tuned for physician-level evaluation of open-ended responses. PrecepTron was trained using low-rank adaptation (LoRA) of a 32-billion-parameter model on a small number of physician examples. We also release GRAND-ROUNDS, a new large-scale physician-annotated benchmark of 9,217 scored responses from 160 clinicians across seven studies. We show that frontier LLMs in typical "LLM-as-a-judge" approaches often disagree with physicians and with each other, but fine-tuning PrecepTron on a small number of cases enables physician-level consistent scoring across tasks. We use PrecepTron to reproduce headline findings from five influential studies assessing LLMs for clinical care in JAMA, Science, and Nature Medicine without new human grading. Using PrecepTron, we then pose new questions about how LLMs reason in medicine that would have been infeasible with human grading alone, including measuring the diagnostic accuracy of frontier LLMs when clinical cases are provided piecemeal, even token by token. Together, PrecepTron and GRAND-ROUNDS provide a foundation for reproducible, large-scale study of how LLMs reason in medicine. All code, data, and labels are made freely available for researchers.
Read the full abstract
High Energy Physics (Phenomenology)
Supernova cooling from neutrinophilic dark matter
Core-collapse supernova serve as a powerful laboratory for testing physics beyond the Standard Model (BSM), particularly regarding new, light states interacting feebly with SM particles. In this work, we investigate for the first time the production of dark matter (DM) via the neutrino-neutrino scattering processes inside a core-collapse supernova, which contributes to the excessive cooling. By incorporating state-of-the-art supernova simulation data , we derive stringent and robust limits on sub-GeV dark matter with effective couplings to neutrinos. We find that the existing and projected constraints from indirect detection are quite weak. Our supernova cooling bounds on DM-neutrino reference cross section can improve indirect detection limits more than ten orders of magnitude for DM masses below $\mathcal{O}(100)$ MeV, and it can also provide strong complementarity with other cosmological constraints. Our results highlight the exceptional sensitivity of core-collapse supernova to feebly interacting particles and motivate future supernova neutrino observations as a powerful probe of light dark sectors.
Read the full abstract
Probability
Moments Comparison Inequalities for Critical 2d Stochastic Heat Flow, Polymers and GMC
We employ a Gaussian convex inequality developed in [AC15], [Che26], [Gue22], and [FUMPC26] to establish (one sided) moment comparison inequalities for Gaussian multiplicative chaos, directed polymers in Gaussian environment and, in particular, the Critical 2d Stochastic Heat Flow (SHF), including upper bounds on negative moments, as well as similar lower bounds for $p$-th moments with $p\in (0,1)$. For the Critical 2d SHF, averaged over small balls, we also obtain the matching (up to multiplicative constants) complementary bounds. We establish the complementary bounds using a bootstrap argument inspired by [DS10] and a geometric decomposition introduced in [GT26]. Specifically, we prove that for locally averaged SHF, for every $p\in\mathbb R$, the $p$-th moment is, up to multiplicative constants, bounded in both sides by the $\frac{p(p-1)}{2}$-th power of its second moment.
Read the full abstract
Optics
Full complex wavefunction reconstruction with direct-interference ptychography
We demonstrate the separable reconstruction of the sample wavefunction amplitude and phase using direct-interference ptychography in scanning transmission electron microscopy. We reconstruct the sample wavefunction in simulated and experimental data under the Weak Object Approximation (WOA) with an extension of single-sideband (SSB) ptychography using information in the double- and triple-overlap regions. The contrast transfer function for the amplitude information is equivalent to the ideal contrast transfer function of Center of Mass imaging (COM). Our results support reporting the imaginary part of SSB reconstructions.
Read the full abstract
Computer Vision
Online Video Agent Harness for Long Video Understanding
Long video understanding often behaves like a visual needle-in-a-haystack problem: query-relevant evidence is sparsely distributed across long temporal spans, while packing dense frames into a single VLM context incurs context rot and high cost. Existing video agents often rely on query-agnostic offline preprocessing or ad hoc tool sets, which can miss query-specific details and waste computation. In this work, we present VideoXAgent, a purely online video-agent harness for long video understanding that starts from the given video file and user query, plans and decomposes the task, invokes specialized expert tools on demand, and aggregates multimodal evidence to produce a final answer while resolving conflicts among observations. To support this on-demand invocation, we design a suite of heterogeneous expert tools guided by a data-driven taxonomy of atomic capabilities, spanning scripts, VLMs, and domain models (e.g., detection, OCR, ASR, face recognition). The harness further enforces objective evidence prompting and budget-aware control to curb hallucination and non-termination. Across Video-MME-Long, LongVideoBench-Long, LVBench, and MINERVA, VideoXAgent is competitive with frontier LMMs and video agents under a smaller context footprint—about 50k tokens of agent context per sample, even on hour-long videos. In particular, on complex video-reasoning benchmarks such as MINERVA, it matches this level while using only about 15% of the context of a 1,024-frame dense-packing baseline. Notably, the harness remains effective with a visually weak or even text-only orchestrator, suggesting that strong long-video understanding can emerge from progressive agentic evidence seeking rather than from packing the full video into a single context. Project page: https://go-agent-x.github.io/video_agent_harness/
Read the full abstract
Mathematical Physics
Static equilibrium of the charged $N$-body problem
We study non-collision static equilibria in the charged \(N\)-body problem. Unlike the classical Newtonian case, such equilibria may exist because the effective interaction coefficients $\delta_{ij}=m_im_j-e_ie_j$ can be positive, negative, or zero. We give a complete classification of three-body equilibria and prove that all of them are collinear and linearly unstable. For the four-body problem, we derive sign restrictions for convex and concave non-collinear equilibria, exclude non-trivial concyclic quadrilateral equilibria, and obtain a geometric necessary condition expressed through an auxiliary triangle and a resultant equation. We also study collinear equilibria by reducing the force-balance equations to a homogeneous polynomial system. This yields a resultant necessary condition and an inverse realization theorem showing that every prescribed collinear configuration of distinct points can be realized by suitable positive masses and real charges. Finally, we analyze regular polygon configurations and centered regular polygon configurations, showing in particular that equal masses cannot form a non-trivial regular \(N\)-gon equilibrium, while the centered case reduces to a single scalar condition.