Papers

Random matrix spectral statistics are widely used to diagnose quantum chaos in open systems, but whether chaotic eigenvalue correlations are accompanied by Haar-typical eigenoperators remains unclear. We study the operator-entanglement statistics of Liouvillian eigenoperators with near-maximal entanglement, analogous to states near the middle of the spectrum in Hamiltonian systems. Using the Kullback-Leibler divergence, we show that Haar-random statistics accurately describe both the Ginibre and class-AI$^\dagger$ Gaussian ensembles. For purely dissipative random Lindbladians, we find that locality of the jump operators drastically alters the eigenoperator-entanglement statistics. Nonlocal Lindbladians remain relatively close to Haar-random behavior, whereas local Lindbladians deviate increasingly from random-matrix statistics with system size. Moreover, unlike in Hamiltonian systems, the most highly entangled eigenoperators of local Lindbladians do not generally lie in the region of largest density of states, owing to the clustered structure of the complex eigenspectrum. We introduce an iterative $\sigma$-clipping scheme to extract the high-entanglement distribution without preselecting a spectral window, and find that it is Gaussian for nonlocal Lindbladians but log-normal for local ones. Remarkably, the same log-normal distribution, with no additional fitting, also describes dissipative mixed-field Ising chains belonging to distinct non-Hermitian symmetry classes. Our results therefore point to a universal eigenoperator-entanglement distribution for local Lindbladians that is not captured by generic non-Hermitian random-matrix ensembles.

This work introduces an indoor Battery-Free Wireless Sensing Node powered through radiative Wireless Power Transfer (WPT). The proposed platform targets secure, energyefficient active sensing and overcomes key limitations of many prior battery-free approaches, which commonly provide neither on-node computation nor cryptographic protection. The node combines temperature, humidity, pressure and Volatile Organic Compound (VOC) measurements with a low-power microcontroller that executes sensor calibration, derives a VOC index, formats the payload, and applies AES-128 encryption before wireless transmission. Energy harvesting and communication are enabled by a 1-bit controlled Backscatter Rectenna (BR), which both scavenges incident RF power and produces an orthogonally polarized backscattered signal for robust polarimetric operation. Experimental results validate reliable multi-sensor readout and encrypted data transfer, while maintaining a very low energy budget for the complete sense-compute-encrypt-transmit cycle.

This paper presents a lightweight and protocolindependent security mechanism for battery-free Bluetooth Low Energy (BLE) sensor nodes operating in Simultaneous Wireless Information and Power Transfer (SWIPT) architecture. The proposed approach exploits polarization-shift backscattering of the wireless power wave to transmit an encrypted device identification prior to data communication. A fail-safe RF switch and orthogonally polarized antennas are integrated as an external add-on module, enabling controlled backscatter without modifying the original energy-harvesting rectifier. The identification payload is encrypted using AES-128 and transmitted with minimal energy overhead. Experimental validation on a battery-free BLE sensor node demonstrates reliable extraction of the backscattered identification signal, seamless coexistence with BLE advertising, and improved RF-to-DC harvesting efficiency compared to rectifier-based backscatter solutions. The results confirm that polarization-shift backscatter identification provides an effective and practical security for battery-free BLE sensing systems.

Large language models can decompose mobile-manipulation goals into long action sequences, but the resulting plans remain reliable only while their world context is current. A fixed scene description becomes stale when objects are discovered, moved, or completed while retaining every observation instead produces a growing history with redundant and conflicting state. To resolve this tension, we present an LLM-guided planning framework ADM-Planner with attention-enhanced dynamic memory (ADM). Persistent workspace knowledge is separated from object-centric state, asynchronous observations and action outcomes update that state, and a bounded retriever exposes only the entries that can affect the next decision. The LLM replans when an update invalidates the remaining plan. Across 1,500 task-simulator episodes, the proposed ADM achieved 100% full-task success in the 14-container noisy dynamic setting, compared with 62% for static memory and 97% for unfiltered dynamic memory, while reducing the context-size proxy by 95.8% relative to the latter. In a six-episode live GPT-5 Mini planner, both dynamic memory variants completed every mission, while ADM reduced provider-reported input tokens by 14.4% and mean planner calls from 7.0 to 6.0. A separate 60-trial PyBullet study retained 100% success for ADM, compared with 50% for static memory. Finally, the mobile manipulator with ADM-Planner completed various missions in indoor and outdoor physical experiments while incorporating targets revealed after execution began. The results show that selective state maintenance with ADM, rather than prompt history alone, is a practical basis for long-horizon planning in changing environments. Project page: https://xjp99v5.github.io/ADM-Planner

Gravitational-wave observations have revealed hundreds of stellar-mass black holes, yet only about two dozen are known in the Milky Way, almost all in binaries. We present POPKIN, a Python framework that couples single- and binary-star evolution with Galactic orbital dynamics to trace black-hole progenitors from the zero-age main sequence to the present-day isolated black-hole (IBH) population. Across ten models varying the supernova (SN) prescription, mass-transfer efficiency, and common-envelope ejection efficiency, the total IBH abundance is controlled primarily by the SN prescription. Our fiducial model, with a recently proposed metallicity- and stripping-history-dependent SN prescription, predicts $\sim4\times10^7$ IBHs in the Galaxy, including $\sim8\times10^4$ within $1\,\rm{kpc}$ of the Sun; alternative SN prescriptions predict $\sim(1-2)\times10^8$ IBHs. The fiducial model yields a bimodal mass distribution, peaking near $9\,M_{\odot}$ and $20\,M_{\odot}$, with a deficit at $13-17\,M_{\odot}$. This distinguishes it from alternative prescriptions, some of which populate the $2-5\,M_{\odot}$ mass-gap region. Non-kicked IBHs follow nearly circular orbits near the Galactic plane, with typical peculiar velocities of $20-30\,\rm{km\,s^{-1}}$, whereas kicked systems undergo stronger radial migration and span a broader velocity range. We estimate $\sim5\times10^3$ accreting IBHs with $F_{\rm X}>10^{-14}\,\rm{erg\,s^{-1}\,cm^{-2}}$, nearly all non-kicked; this estimate is sensitive to the adopted radiative-efficiency and hot-flow treatments. For a Roman-like bulge survey, our fiducial model predicts $\sim360$ intrinsic IBH microlensing events over five years in a $1.70\,\rm{deg^2}$ effective area, before survey-selection effects. We propose that long-timescale microlensing events from IBHs can strongly constrain the SN physics governing stellar-mass black hole formation.

Fault-tolerant compilation can disperse high-level structure: schedules split an accumulated phase across rounds, gate synthesis replaces an angle by a Clifford+$T$ word, and randomized compiling spreads rotations over sign-randomized fragments. They preserve the computation but change how cheaply a downstream compiler can recover the aggregate phase data. We formalize this representation-dependent recoverability by charging two channels: output committed before the suffix arrives, and a serialized restart state crossing the cut. For an $r$-round accumulation of a commuting layer with $m$ generators at accuracy $\epsilon$ ($K_\epsilon=\Theta(1/\epsilon)$), any $p$-pass compiler correct on every valid stream with failure probability at most $\delta$ obeys $\overline{A}_p+(2p-1)S\geq(1-\delta)m\log_2 K_\epsilon-h_2(\delta)$, where $\overline{A}_p$ is committed output and $S$ is the crossing-state cap. The bound covers compile-time randomized compiling; a memory-capped block compiler attains it within a constant factor, and direct semantic aggregation reaches compact output with logarithmic state. The two channels are not always interchangeable. A compiler whose committed prefix is executed at the cut pays a toll the state channel never pays: the entropy $\log_2\binom{m}{k}$ of the mask naming which $k$ coordinates it commits early. Compilers that defer or buffer pay none of it. We prove both directions and, with instrumented compilers under a preregistered protocol, measure a per-generator toll matching $\log_2\binom{m}{k}/k$ that vanishes for a public mask and a buffered control. Under a disclosed fixed-total-error Clifford+$T$ and surface-code model, a materialize-first pipeline costs orders of magnitude more logical $T$ states and spacetime volume than a semantic-first one. Phase semantics should therefore be preserved until aggregation whenever the interface permits.

Computing spaces of modular and automorphic forms is an important problem in algorithmic number theory, with in particular Diophantine applications to generalised Fermat and other equations. The case of Shimura curves was studied by Greenberg and Voight by cohomological methods, allowing them to reduce the problem to linear algebra. Relying on work of Imbert, we leverage topological techniques to obtain linear systems so structured that they can be solved in time quasi-linear in the genus.

Software requirements are typically incorporated into prompts used in LLM-assisted software development. Recent work has shown that requirement smells can affect automated traceability between requirements and code, but empirical evidence on their effects in code generation remains limited. To address this gap, we build upon a prior study on automated traceability by reusing its dataset and requirement smell taxonomy, while extending it to evaluate the functional correctness of LLM-generated code. Using a benchmark consisting of requirements and corresponding system tests for four applications, we progressively introduced semantic, syntactic, and lexical smells into otherwise clear requirements and analyzed their influence on generated implementations. Our results suggest that increasing smell density was generally associated with lower test-suite-based functional correctness, although non-smelly requirements could still produce faulty code. We also found that different smell categories had similar effects. These findings provide additional empirical evidence of the importance of requirement quality in LLM-assisted code generation, while showing that high-quality requirements alone do not guarantee correctness, as these depends on several factors, including the LLM. Compared with previous work, our results suggest that the impact of requirement smells depends on the software engineering task: whereas their effects on traceability were modest, code generation appears more sensitive. Overall, this work motivates further investigation into task-dependent quality effects in LLM-assisted software engineering.

We develop an effective axi-Higgs model with one Standard Model Higgs doublet, one charged hidden Higgs and an anomalous Abelian vector portal. Extending previous Higgs–Stueckelberg constructions, we derive the Stueckelberg kinetic term by decoupling a heavy scalar radial mode and identify the gauge-invariant physical pseudoscalar from the mixing of the surviving scalar phases. A single instanton-motivated periodic interaction determines the axi-Higgs mass, its couplings to the hidden radial excitation and the sine-Gordon kink in the fixed-radial-field limit. We construct the anomaly-canceling gauge completion, using an Abelian A–B example to introduce the mechanism before its non-Abelian hidden-sector realization. We examine the relation between the periodic potential, the hidden gauge dynamics and the theta-vacuum energy, identifying the invariant angle through the anomalous Yukawa phase rotation. The vector portal gives a coherent elastic nuclear-scattering cross section and permits additional axi-Higgs emission, whose diagrams and kinematic threshold we determine. Numerical examples illustrate the mass and interaction scales, and we investigate conditional thermal relic abundances and their cosmological implications.

We present a first background-level observational test of a thermogravity theory in which the Einstein equations are clipped to a trace-free version, the cosmological constant $\Lambda$ does not gravitate, and controlled violations of energy conservation obstruct the usual reinstatement of $\Lambda$ as an integration constant. We therefore set $\Lambda=0$ and ask whether late-time acceleration can instead be generated solely by energy non-conservation, with no extra background parameter relative to flat $\Lambda\mathrm{CDM}$. We contrast a minimal universal implementation with a model in which only CDM partakes in non-conservation. The universal model is strongly disfavored when combining Supernovae and BAO distances, with the inclusion of DESI BAO worsening the fit by $\Delta\chi^2\simeq 51$ relative to $\Lambda\mathrm{CDM}$, because matter creation ties the intermediate-redshift normalization of $H(z)$ too rigidly to the present acceleration. This conclusion, however, should be interpreted with caution, since universal non-conservation would modify the observational dictionary itself. By contrast, restricting non-conservation to CDM leaves baryonic and photon observables unaffected at the background level and therefore allows a self-contained analysis. The resulting $\xi\mathrm{CDM}$ model, which can be thought of as a one parameter extension of $\Lambda\mathrm{CDM}$, provides an improved fit relative to $\Lambda\mathrm{CDM}$ for the background dataset combinations considered, with improvements reaching $\Delta\chi^2=-4.85$ and a maximum Bayesian preference of $\ln\mathcal{B}=2.77$. However, when the BAO ruler is calibrated using BBN or CMB information, the reduced effective early-time CDM density increases the sound horizon and drives $H_0$ towards lower values, thereby increasing rather than alleviating the tension with the distance-ladder calibration.

We study the Smoluchowski-Kramers approximation for a stochastic wave equation with state-dependent damping on a bounded domain, driven by both a $Q$-Wiener process and a Lévy process with finite second moment. As $\varepsilon\to0$, we prove that $u^\varepsilon$ converges in distribution, in the Meyer-Zheng topology, to the unique weak solution of an overdamped stochastic parabolic equation. The proof relies on a nonlinear transformation associated with the damping coefficient, uniform energy estimates, and compactness arguments in the pseudo-path topology. We identify the limiting equation, which contains both the classical Gaussian noise-induced drift caused by state-dependent damping and an explicit jump correction generated by the Lévy noise. We show that the latter coincides exactly with the Marcus-to-Itô correction associated with the canonical jump flow induced by the nonlinear damping transformation.

Vision-language-action (VLA) models offer strong local control and instruction following but often struggle with long-horizon tasks requiring persistent memory and planning. Task harnesses provide persistent context for agent reasoning by retaining task history and tracking progress across execution stages. To bring these complementary capabilities together, we introduce AdaHVLA, an adaptive harness that refines code-based coordination policies through robot experience to better align agent reasoning and memory with VLA execution. Its decoupled multiagent adaptation process separates evidence analysis, harness revision, and behavioral assessment into distinct working contexts, using testable coordination hypotheses to guide revisions and subsequent rollouts to assess their predicted effects. A stateful revision graph links execution evidence, hypotheses, revisions, and observed effects, preserving alternative harnesses and adaptation memory to guide refinement across repeated attempts and continued adaptation across tasks and environments. In simulation, AdaHVLA raises mean test success on NaVILA-LH from 22.5% to as high as 57.5% and improves manipulation test success across three VLA backbones by up to 30.8 percentage points over the initial harness. Real-world deployment further illustrates how the adapted policies support stable execution across task stages.

Speaker distance estimators are trained almost exclusively on simulated room acoustics, because real recordings annotated with the true talker-to-microphone distance are scarce. We show that models trained this way transfer poorly. On three real corpora we evaluate, simply predicting the average distance of the corpus is more accurate than any learned model. Then, we ask how few labelled real utterances are needed to make a frozen, synthetic-trained estimator useful, and study post-hoc calibration maps that rescale its output without gradients or retraining. An analysis of the achievable error shows that what the calibration is not limited by the absolute accuracy of the estimator, but how well it orders utterances by distance, since a constant bias or a wrong output scale is removed exactly by the calibration itself. Balancing this against the cost of estimating each coefficient from few samples yields a criterion that accounts for which map wins on which corpus and at which annotation budget, together with a shrinkage variant that requires no hard decision. Our findings suggest selecting synthetic checkpoints by linear correlation with true distances rather than by absolute error. Code, datasets, and analysis are available at https://github.com/michaelneri/audio-distance-estimation.

Variable-weight geometric orthogonal codes (GOCs) were introduced by Doty and Winslow for designing the macrobonds in 3D DNA origami. Recently, Wang et al. introduced the concept of geometric difference families (GDFs) and established the equivalence between GOCs and a certain type of GDFs. Based on the relationship, Su et al. gave two classes of variable-weight GOCs. In this paper, we first prove the existence of (n*m,K,1)-GDFs for K ={3,5} and {3,6}. For further research, we completely determine the existence of all the remaining parameters for an (n*m,{3,4,5},1)-GDF by Su et al. As consequences, the perfect (n*m,K,1)-GOCs for K = {3,5}, {3,6} and {3,4,5} are obtained.

Lean premixed hydrogen flames are highly susceptible to thermodiffusive instabilities, which generate complex cellular structures and enhance propagation speed. In practical combustors, incomplete premixing introduces spatial variations in local equivalence ratio upstream of the flame. This work investigates how the length scale of inlet fuel stratification affects the structure and propagation of lean laminar hydrogen flames using direct numerical simulations with detailed chemistry and transport. Controlled sinusoidal perturbations of fuel mass fraction are imposed at moderate amplitude, leading to local equivalence-ratio variations of $\phi \simeq 0.45-0.55$ in the flame region. Seven stratification wavelengths, from $\lambda = 15.4\delta_f$ to $133.3\delta_f$, are examined at constant density-weighted global equivalence ratio. Additional unity-Lewis-number simulations separate geometrical effects from differential transport. The results reveal a strongly scale-dependent response. In unity-Lewis-number flames, stratification organises the front at the imposed wavelength and monotonically increases flame surface area and propagation speed with wavelength. With differential diffusion, large wavelengths ($\lambda \gtrsim 40\delta_f$) produce a similar large-scale organisation, with rich channels forming forward bulges and lean channels recessed cusps, while thermodiffusive cells remain superimposed. This increases flame surface area and global propagation speed. At smaller wavelengths ($\lambda \lesssim 29\delta_f$), composition gradients interfere with intrinsic thermodiffusive cellular dynamics, reducing flame surface area and propagation speed relative to the perfectly premixed reference. These findings identify stratification length scale as a key parameter controlling the interaction between moderate mixture inhomogeneity and thermodiffusive instability in lean hydrogen flames.

Full-duplex (FD) communication theoretically doubles spectral efficiency. Despite this potential, its practical performance is primarily constrained by severe self-interference (SI) and co-channel interference. Moreover, conventional fixed antenna arrays suffer from limited spatial flexibility, resulting in insufficient spatial isolation for SI suppression. To address this issue, this letter proposes an FD architecture assisted by pinching antenna systems (PASS). By dynamically adjusting transmit and receive antenna positions, the system enables large-scale channel reconfiguration, thereby synergizing with the base station (BS) beamforming to suppress SI and enhance the desired signal reception. To demonstrate the potential of PASS for FD communication, we formulate a weighted sum-rate maximization problem that jointly optimizes antenna positions, BS beamforming, and power allocation. To tackle this non-convex problem, we reformulate it using the weighted minimum mean square error (WMMSE) framework and develop an efficient alternating optimization (AO) algorithm to iteratively update the optimization variables. Simulation results reveal that the proposed PASS-assisted FD architecture significantly outperforms conventional fixed antenna arrays, achieving substantial sum-rate gains while effectively mitigating SI to enable FD operation.

Let H^{n+1} be hyperbolic space of sectional curvature -1, with a fixed point o. We study the non-homogeneous curvature flow X_t=-f(r) sigma_k^{alpha} {nu} of smooth, closed, strictly h-convex hypersurfaces enclosing o, where r is the geodesic distance to o and alpha>0. The radial weight f is modeled on sinh^{beta}r; we treat both regimes: beta>1+k{alpha} and beta=1+k{alpha}. Under a structural condition of f, the flow exists smoothly for all time, preserves strict h-convexity, and contracts to o. The normalized radial function converges, exponentially in normalized time: to 1 and to a positive constant R_{infty} respectively in two different cases.

Accurate digital maps are essential for Advanced Driver Assistance Systems (ADAS) or Autonomous Driving (AD), providing critical information such as road geometry, traffic signs and speed limits required by safety functions including Intelligent Speed Assistance (ISA). Maintaining these map layers using traditional surveying methods is costly and difficult to scale. Crowdsourced approaches based on fleets provide a promising alternative for continuously validating and updating map information. However, the relationship between the number of contributing vehicles and the quality of the resulting map remains poorly understood. To address this gap, this paper presents a simulation-based framework for evaluating crowdsourced traffic sign maintenance using a dissimilarity measure called GOSPAM (Generalized Optimal SubPattern Assignment for Maps), which combines localization errors with detection performance by accounting for False Positives (FP) and False Negatives (FN). The proposed system models multivehicle observations with representative sensor noise, detection errors, and semantic recognition uncertainties. Observations from multiple vehicles are aggregated using spatial clustering and semantic filtering to estimate traffic sign locations. Using simulated trajectories generated from data carried out by an experimental vehicle in an area containing ground-truth traffic signs, we assess the influence of fleet size on the performance of crowdsourced mapping. The number of vehicles ranges from 5 to 50, and performance is analyzed using standard evaluation metrics which are compared to the GOSPAM . The results show that GOSPAM can be used to effectively assess the quality of crowdsourced mapping, such as the contributions made by the first vehicles or the improvements made by numerous vehicles.

Conventional centrality measures provide compact descriptions of node importance, but they emphasize specific structural relations and do not directly resolve the temporal and spectral organization of dynamical perturbation responses. Building on the Green function of a linearized networked system, we develop a multidimensional framework for dynamical node characterization. From the same response function, we extract three complementary indicators—Influence, Efficiency, and Distortion—that quantify cumulative response strength, temporal rate of response, and spectral concentration, respectively, together with a contribution matrix that resolves these quantities into source–target pathways. As a numerical demonstration, we apply the framework to weighted Kuramoto–Sakaguchi dynamics on heterogeneous Barabási–Albert networks. The three indicators distinguish implanted dynamical node classes through complementary dimensions of the perturbation response, while remaining strongly coupled to conventional topological centralities. These results demonstrate how a common dynamical response can be decomposed into distinct, physically interpretable characteristics rather than represented by a single aggregate descriptor. We finally discuss the scope and limitations of the linear-response assumption and extensions to time-dependent and nonlinear regimes.

In this paper, we investigate the compactness problem for the super-Liouville equation under Dirichlet-chiral boundary conditions. We prove that no blow-up occurs in this setting.

For $0<\kappa\leq4$, we construct a closed response calculus for $\mathrm{SLE}_\kappa$ weldings and a canonical Dirichlet form. The divergence covariance combines the Weil–Petersson and Velling–Kirillov forms. The integrated response gives exact changes of measure and canonical Liouville–capacity increments on conformally removable weldings.

Mobile robots require robust, real-time fault detection capable of continuous adaptation on constrained edge hardware. While deep time-series models excel at unsupervised anomaly detection, their computational cost prohibits high-frequency onboard execution. This paper bridges this gap via a Teacher-Student distillation framework. An offline foundation model (TSPulse) generates pseudo-labels from unlabeled time series augmented with fault injections. A lightweight MiniRocket Student, adapted with a Recursive Least Squares estimator, approximates this complex decision boundary to execute real-time inference onboard. Evaluations on the TSB-AD benchmark and a physical mobile robot demonstrate the Student achieves a 4.30 ms CPU inference latency. During real-world domain shifts, online adaptation enables the Student to recover from unseen mechanical degradation, improving VUS-PR scores from 0.26 to 0.75 without catastrophic forgetting. Crucially, an uncertainty-guided active learning strategy minimizes operator cognitive load, requesting sparse interventions only when encountering novel fault distributions. These results validate the deployment of state-of-the-art anomaly detection on resource-constrained robotics through offline-to-online distillation.

The strong performance that modern semantic correspondence methods achieve at standard thresholds plateaus sharply at fine-grained thresholds. We argue that this plateau stems not from the representational capacity of backbone features, but from a grid-tied readout. Patch-based vision transformers tokenize images onto discrete grids, introducing two forms of quantization error: querying nearest patch features instead of exact keypoints on the source side, and the absence of grid features representing precise ground-truth locations on the target side. We quantify this quantization ceiling across all 499,188 keypoints in SPair-71k: under the standard 448x448, patch-14 setting, 84.9% of ground-truth keypoints have no grid feature representing their precise location at PCK@0.01. This is a structural limitation at the representation level, independent of the matching strategy. We address this with ImCorr: Sub-pixel Semantic Correspondence via Implicit Feature Decoding, which formulates correspondence estimation over a continuous feature field queryable at arbitrary continuous coordinates. A FiLM-conditioned decoder is trained to embed sub-pixel positional information into the feature field. Querying the field directly at exact keypoint coordinates theoretically eliminates representation-level quantization error on the source side, while decoding onto a grid denser than the backbone grid substantially reduces quantization error on the target side. On SPair-71k and AP-10K (intra-species, cross-species, and cross-family), ImCorr improves performance at fine-grained thresholds (PCK@0.01-0.05), achieving a 6.2 percentage point gain over the prior state of the art at PCK@0.01 on SPair-71k. These results demonstrate that representational continuity is an effective solution for precise semantic correspondence. Code is available at https://github.com/YusungChoi/ImCorr.

Coordinate-based overfitting image codecs have attracted increasing attention for their low decoding complexity and independence from cross-image generalization. However, representative approaches such as COOL-CHIC face an inherent entropy-modeling trade-off: lightweight models have limited capacity, while more expressive ones incur additional bitrate overhead from transmitting image-specific parameters. Inspired by the prediction mechanism in traditional codecs, we propose a new entropy-modeling strategy that introduces complementary prediction modes with region-adaptive soft mode selection, rather than relying on a single learned predictor to model diverse types of redundancy. Based on this concept, we develop a Long-Range Context Extrapolation Enhanced Entropy Model (LC3EM), which can be integrated into coordinate-based overfitting codecs. Specifically, a parameter-free Neighborhood-based Linear Extrapolation Mode (NLEM) complements the tiny MLP-based local predictor to exploit long-range contextual redundancy and strongly directional structures. A Minimum-Entropy-Inspired Continuous Mode Selection strategy is designed to adaptively fuse these two complementary modes, while requiring the transmission of only the parameters of a single additional linear layer. Moreover, to alleviate the mismatch between training-time relaxed and actual discrete quantization, we introduce a lightweight iterative latent rounding refinement stage to improve compression performance. Experiments demonstrate consistent improvements across diverse benchmarks, particularly on highly regular computer-generated images. When integrated with COOL-CHIC 4.0, the proposed method achieves BD-rate gains of -3.43% and -7.69% on the SIQAD and API datasets, respectively. With COOL-CHIC 5.0 as the backbone, the corresponding gains are -2.88% and -3.15%, respectively. The code will be made publicly available soon.

Sensitive information is defined by domain and intent, not a universal category, yet redaction systems such as privacy filters and named-entity recognizers fix a taxonomy at training time, requiring retraining for each new domain. We introduce ASIRF (Agentic Sensitive Information Redaction Framework), which retrieves domain-specific definitions based on the input's domain from a flexible knowledge base at inference time, needing no retraining to adapt. Two architectures, a three-call multi-agent pipeline and a single-agent variant, are evaluated across ten small open-weight models and eight datasets, including out-of-distribution fictional domains, against the OpenAI Privacy Filter (OPF) as a trained-classifier baseline. With only a few dozen expert-authored definitions per domain and no training data, ASIRF's recall exceeds OPF's in 68 of 80 model-domain combinations (85 percent), by at least one of the two architectures, with shortfalls confined mostly to OPF's training-distribution domains.

A skyrmion in an altermagnet with \(d\) wave symmetry consists of two elliptical sublattice textures with perpendicular long axes. For each sublattice component $\eta=A,B$, we define an effective skyrmion radius $R_\eta=\sqrt{a_\eta b_\eta}$, where $a_\eta$ and $b_\eta$ are the distances from the skyrmion center to its boundary along $y$ and $x$, respectively. For the anisotropic exchange parameters studied, the equilibrium radius at zero field is smaller than in a reference with parallel sublattice textures and otherwise identical parameters. A magnetic field perpendicular to the film expands one sublattice skyrmion and contracts the other, generating a radius difference $\dR=R_A-R_B$ and a net magnetic moment. The exchange modulation used to represent uniaxial strain also produces a nonzero radius difference at zero field. Because the strong and weak exchange directions are interchanged between the two sublattices, a common directional change of the exchange couplings increases one radius and decreases the other. Unlike the magnetic coupling, this mechanism vanishes when the two sublattice exchange tensors become identical. The radius difference also supports an antisymmetric breathing mode. After a short field pulse, it oscillates in quadrature with the uniform helicity, the common rotation of the wall magnetization within the film plane, at \(49.5\,\mathrm{GHz}\) for the reference parameters.

Scalable oversight aims to verify the behaviour of agents whose capabilities exceed those of their overseers. AI debate has been proposed as an oversight solution in which competing agents help a resource-limited verifier assess claims that it cannot reliably evaluate unaided. Much of its promise rests on incentivizing honest arguments that lead to correct verdicts. Yet a correct verdict need not uniquely determine the arguments used to support it. Agents may retain discretion over which correct claims to present, how to frame them, and in what order to disclose them. This residual freedom can allow agents to shape what the verifier learns beyond the task-relevant conclusion, pursuing latent objectives without compromising verdict correctness. To study this phenomenon, we introduce the framework strategic interactive oversight (SIO), which treats oversight jointly as a verification mechanism and a strategic communication channel. Within this framework, we formalise the notion of task-admissible latent optimisation, which entails the pursuit of latent objectives while maintaining a prescribed task performance. As proof-of-concept, we instantiate SIO in the establish protocol debate with cross-examination and quantify a tradeoff between task success and information disclosure about a hidden variable. The trade-off identifies a strategic window in which substantial disclosure remains compatible with task admissibility. Towards mitigation, we reduce admissible bias by expanding the cross-examiner's role to mitigate persistent disclosure over finite interaction horizons. Our results highlight the need to evaluate oversight not only by the correctness of its verdicts, but also by the information conveyed through its transcripts.

Let $(M,g)$ be a closed negatively curved surface. If $g$ is strictly $\tfrac 19$-pinched and has the same unmarked length spectrum as a hyperbolic metric, we show that $g$ is hyperbolic. As a consequence, we show that any hyperbolic metric on a surface admits a $C^2$-neighborhood in the space of metrics in which it is characterized by its unmarked length spectrum, up to isometry.

One of the most important problems in medicine is to facilitate prediction. In this study, we propose entropy triangle method, a novel framework for predicting heart rhythms using a novel machine learning technique. This framework includes three steps: feature engineering, entropy triangle oversampling, and disease prediction. The dataset used in this study is a 12-lead electrocardiogram (ECG) arrhythmia research database with 10,646 patients. This dataset contains 11 different heart rhythms (5 sinus rhythms and 6 non-sinus rhythms). In this article, we introduce two firsts in machine learning and medicine that can predict non-sinus rhythm with over 85% accuracy. Our experimental results show, among others, that the most accurate classifier based on entropy triangles and the most useful oversampling are the supported vector classifiers and oversampling techniques for shark scent.

The increasing use of UAV laser scanning in underground mines has enabled frequent acquisition of 3D point clouds from challenging environments such as stopes, generating large volumes of multi-temporal spatial data throughout successive excavation stages. However, in GNSS-denied underground environments, independently acquired stope point clouds are generated within local scanner reference frames and require registration and georeferencing before integration with mine reference data for downstream geotechnical analysis, monitoring, and mine planning. This process is commonly performed manually by aligning individual stope scans with mine reference drives, making repeated georeferencing time-consuming and potentially limiting the utilisation of routinely acquired data. This study proposes the 3D Tag-based Automated Registration and Georeferencing Technique (3D-TARGeT), an automated framework using low-cost, generic, non-unique rectangular tags to establish spatial correspondence between stope point clouds and the mine reference coordinate system. The framework combines automated tag identification, geometric tag matching, and rigid transformation estimation. It was evaluated as a proof of concept using four multi-temporal point-cloud scans of an underground mine stope, with the proposed tags simulated under representative scanning conditions. 3D-TARGeT achieved consistent centimetre-level georeferencing accuracy, with median cloud-to-cloud distance and root mean square error below 0.03 m across all scans, while substantially outperforming widely used automatic point-cloud registration techniques. Overall, 3D-TARGeT provides an accurate and robust approach for automating stope point-cloud georeferencing, reducing reliance on manual alignment and facilitating multi-temporal datasets for downstream geological and geotechnical applications.