Peer Revenue Returns subject to peer review

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Tuesday 1 September

Computation and Language

Replicating the Geometry of Emotion Representations in a Base Open-Weights Model

Sofroniew et al. (2026) report that emotion concepts in Claude Sonnet 4.5 are represented as vectors whose geometry mirrors human affect psychology. We replicate the representational core of that study on the base pretrained model google/gemma-2-27b, inheriting every disclosed parameter, resolving unspecified steps by disclosed rules, and changing only the subject model. From 205,200 newly generated Claude Sonnet 4.5 stories matching the original corpus design, we extract 171 emotion vectors and recover the core results. The leading principal components form an affective circumplex (PC1 carries 26.7% of variance against the original study's ~27%, PC2 13.4% against ~14%), emotions cluster into similar intuitive families, and the geometry holds across a broad late-middle band. The valence axis aligns with human norms (r = 0.72, against 0.81) and is stable across scales and depth. Arousal aligns at r = 0.67 (against 0.66) but only at the full 171-emotion scale and late depth, so we do not classify it as replicated. Extending the original analysis, a 46-layer sweep locates a sharp seam at L22-26, where the geometry consolidates and the vocabulary readout becomes legible. An embedding-layer baseline finds much of the geometry already present in the static token embeddings, with arousal as the exception. On top-activating held-out text, the geometry predicts token-level co-activation at r = 0.907. At least 52% of vectors peak on structurally non-conceptual tokens, a measured floor for max-activation confounds. Even when a document contains the vector's emotion word, the peak lands on that word only 6.1% of the time. Because the subject is a base model and the stimuli are Claude-generated fiction, the recovered structure is a property of the pretrained representation of Claude-rendered emotion. The original's causal and assistant-facing analyses are out of scope. Code and data are released.
Read the full abstract

Signal Processing

An Implant-to-Wearable IR-UWB Transmitter-Receiver Architecture and Layered Protocol for High-Density Brain-Computer Interfaces

High-rate neural telemetry requires an implant-towearable link whose circuit partition, waveform mapping, packet processing, and failure semantics are mutually consistent. This paper specifies a chip-oriented, one-way impulse-radio ultrawideband (IR-UWB) uplink comprising an implanted packet engine and free-running transmitter, a tissue-proxy channel, and a wearable mixed-signal receiver and digital baseband.
Read the full abstract

Computation and Language

Dissecting Training-Free Uncertainty Estimation in Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) are increasingly deployed in safety-critical settings, yet estimating when their answers can be trusted remains underexplored. We systematically compare training-free verbalized, token-level and semantic uncertainty estimation methods on twelve MLLMs and five VQA benchmarks. We find that answer length decides which family wins: token-level entropy on short answers, verbalized methods on sentence-length responses, and semantic methods on long-form generation. Token-level entropy becomes far more informative when estimated at a higher temperature than the low ones used to generate answers, and asking a model to judge its own answer becomes more reliable with model size. Building on this, a length-aware recipe that additionally flags responses on which the model abstains outperforms every individual method on average at about a quarter of the cost of semantic methods, although no method attains the coverage required for safe deployment. These results provide practitioners with principled guidance for selecting estimators by answer length and model scale.
Read the full abstract

All authors Soroush Seifi, Vaggelis Dorovatas, Lin Li, Yarin Gal, Rahaf Aljundi

Machine Learning

Prediction of Nonlinear Oscillations in a Jumping Quarter-Car Model Using Reservoir Computing

Reliable prediction of vehicle dynamics is essential for smart driving applications such as autonomous control and advanced driver-assistance systems. Off-road vehicles used in agricultural and construction settings are particularly prone to nonlinear behavior, including bifurcations and chaotic motion arising from intermittent loss of tire–road contact. Predicting such dynamics is challenging because it requires resolving both smooth nonlinearities and the discontinuous switching associated with contact loss. In this work, we investigate the feasibility of reservoir computing (RC) – specifically an echo state network (ESN) – for data-driven prediction of a jumping quarter-car model. The reservoir is trained on time-series data from a small number of points and evaluated on its ability to reconstruct bifurcation diagrams, phase-space attractors, and time trajectories across periodic and chaotic regimes. The trained reservoir qualitatively reproduces the period-doubling route to chaos, captures the geometric structure of periodic and chaotic attractors. These results demonstrate that reservoir computing is a feasible data-driven predictor of nonlinear dynamics in a practical, non-smooth vehicle system.
Read the full abstract

All authors Masahisa Watanabe, Shiva Dixit, Nirmal Punetha, Swati Chauhan, Manish Dev Shirimali

Computation and Language

Evaluating Personal Information Output from Conversational Interactions in Generative AI Systems

This exploratory pilot study evaluates the scope and perceived accuracy of personal information output from ongoing conversational interactions in generative AI systems using GPT-5.2 Instant and GPT-5.2 Thinking, categorized into three output types: Fact, Inference, and Confidence. Based on the evaluation results obtained from 15 Japanese participants, differences in model design have limited impact on personal information output tendencies. Compared with the Inference type, the Fact type shows a more conservative output pattern. Regarding attribute categories, the findings indicate that Core Personal attributes associated with identification are treated relatively conservatively, whereas Behavioral and Linguistic attributes show higher accuracy across both Fact and Inference outputs. Furthermore, Holistic Profile, Psychological and Cognitive, and Residual attributes are more readily inferred, even when not supported by explicit factual outputs. Notably, the lack of null outputs for these attributes in the Inference type suggests that such inferred profiles may be constructed from indirectly available contextual information. The findings may contribute to future discussions regarding privacy awareness and personal information inference in generative AI systems.
Read the full abstract

Signal Processing

Denoising Rather Than Gap Filling: Missing-Data Handling in Sparse Outdoor BLE Positioning

Received signal strength indicator (RSSI) positioning outdoors has one to two orders of magnitude fewer anchors than the indoor systems its methods come from. In the ten-day cattle-tracking deployment reported here, four gateways cover $4{,}302\,\mathrm{m}^2$, or $0.93$ anchors per $1000\,\mathrm{m}^2$. An animal is heard by $0.99$ gateways per second on average, so the observation vector for a given second is almost never complete. How the gaps are filled is therefore a first-order design choice, not a preprocessing detail. Filling at all is worth $8.1\,\mathrm{m}$, $96\%$ of the improvement the hold-length setting can deliver and a $29\%$ error reduction. How long a value is then held is worth the remaining $4\%$, from four seconds to unbounded. Once a value is available, the gain comes from removing noise rather than rebuilding the lost sample: a filtered channel estimate improves on a held raw sample, whereas filling backwards from future samples makes it worse. As that mechanism predicts, smoothing strength has a real interior optimum that is costly to miss in either direction. A smoother allowed to read the future gains only $0.55\,\mathrm{m}$, one fifteenth of what filling is worth, which bounds what any offline method can add. Three method families do not apply here for structural reasons rather than poor performance: two or more of the four channels are live in only $25\%$ of seconds, so the cross-channel structure generative imputation must learn is largely unobserved.
Read the full abstract

All authors Yu Sun, Zhiyi Zhu, Patrick Finnerty, Takenao Ohkawa, Kenji Oyama, Chikara Ohta

Risk Management

An Operator-Based Visual Analytics Pipeline for Synthetic Systemic Risk Dynamics

This work presents an operator-based visual analytics pipeline for exploring synthetic systemic risk dynamics in financial networks. The framework is formulated as a composition of mathematical operators that sequentially transform synthetic financial observations into dynamic scientific visualizations. The pipeline consists of six operators: latent risk mapping, probabilistic score generation, financial network construction, distance-based contagion dynamics, visual encoding, and perspective projection. Together, these operators provide a modular computational structure linking nonlinear risk surfaces, time-dependent probabilistic states, network topology, and shock propagation. A reproducible implementation demonstrates the architecture through controlled synthetic experiments. Nonlinear latent risk representations are converted into probabilistic scores, embedded in a weighted financial network, and propagated via shortest-path contagion. The resulting states are then mapped into dynamic visual representations. The experiments illustrate how visualization can be treated as an explicit stage of the analytical process rather than a post-processing step. The proposed formulation does not aim to introduce new predictive models or contagion mechanisms. Instead, it offers a transparent and modular pipeline that connects generative risk modeling, network dynamics, and scientific visualization within a unified operator-based architecture. This approach supports reproducibility and facilitates the exploration and communication of complex systemic-risk processes in synthetic settings.
Read the full abstract

Statistical Methodology

Bayesian Quantile Deep Echo State Networks for Nonlinear Time Series

Conditional quantiles are central to asymmetric decision losses, tail-risk assessment, and interval forecasts, but Bayesian quantile regression for nonlinear time series can be difficult when the temporal feature vector is high-dimensional. We develop the quantile deep echo state network (Q-DESN), a Bayesian quantile regression model conditional on fixed features generated by a deep echo state network. Conditional on a specified reservoir construction and feature map, posterior uncertainty is assigned to the regression coefficients, likelihood, and shrinkage parameters. Single-level fits use asymmetric Laplace or quantile-fixed generalized asymmetric Laplace working likelihoods with ridge or regularized-horseshoe priors. Posterior computation uses Markov chain Monte Carlo when computationally practical and a model-specific variational Bayes approximation for analyses requiring repeated fitting. For quantile grids, we compare independent level-wise regressions followed by monotone rearrangement with a joint quantile-vector regression that shrinks adjacent quantile-specific coefficient differences. Synthetic studies, a single-origin retrospective Global Flood Awareness System (GloFAS) streamflow case, and a retrospective PriceFM comparison identify settings where this fixed-feature Bayesian regression improves finite-grid quantile scores.
Read the full abstract

General Physics

A Non-Singular Cyclic Universe from Modified Gravity

The Big Bang singularity marks a fundamental breakdown of General Relativity, yet all proposed quantum gravitational remedies invoke microphysical assumptions beyond the observational reach. We show that a minimal, purely classical modification of the gravitational potential — preserving the inverse-square law where tested, but free to depart at unmeasured scales — is sufficient to produce a non-singular, cyclic universe. The resulting modified gravitational formulation leads to a cosmic evolution exhibiting a bounce at a finite minimum size, followed by a regime closely akin to the concordant $\Lambda$CDM model, and an eventual collapse to another bounce. Cosmic Chronometer, Supernovae type Ia, Baryon Acoustic Oscillation, Cosmic Microwave Background and Nucleosynthesis data constrain both the new length scales, implying a cycle period of $\simeq 89\,$Gyr, and this framework — subsuming a broad class of bounce scenarios within a single testable model — opens a direct route for distinguishing cyclic cosmologies from $\Lambda$CDM.
Read the full abstract

All authors Saurya Das, Mitja Fridman, Dhiraj Kuniyal, Sourav Sur

Computer Vision

Don't Just Look, Intervene: Perturbation Based Region Labeling for VQA Images

Vision Language Models (VLMs) should rely on visual evidence that directly determines the correct answer, but supervision for grounding visual reasoning is often expensive to obtain manually or tied to dataset-specific annotation primitives. We instead introduce model-causal visual evidence as an annotation target, defined as the set of image regions whose counterfactual intervention changes a model's answer distribution for a given image-question pair. Based on this principle, we introduce Counterfactual Search for Grounding Regions (CSGR). CSGR is a scalable pipeline that proposes candidate regions, perturbs them, measures their effect on answer sensitivity, and aggregates this evidence across multiple judges to approximate answer-critical regions in VQA data. To assess whether CSGR annotations contain a useful supervision signal, we plug them into three existing grounding-aware training routines: attention steering, Visual CoTfinetuning, and latent visual reasoning. These experiments test whether the proposed annotation scheme can provide a useful supervision signal across multiple ways of consuming region labels, rather than introducing a new way of using them. Across competing automatic region-labeling mechanisms, CSGR annotations provide the most consistent gains over Cross Entropy-only finetuning in both in-domain and out-of-domain evaluations, indicating that the proposed labeling scheme captures useful region-level information.
Read the full abstract

General Relativity and Quantum Cosmology

Quark stars in regularized 4D Einstein-Gauss-Bonnet gravity: A perturbative QCD equation of state

We investigate the equilibrium structure and stability of selfbound quark stars in the framework of regularized four-dimensional Einstein–Gauss–Bonnet (4DEGB) gravity, employing the perturbative QCD equation of state of Fraga, Kurkela and Vuorinen [1]. The equation of state, parameterised by a single renormalization-scale parameter $X$, contains no effective bag constant and defines the stellar surface entirely through the vanishing of the quark-matter pressure. We solve the modified Tolman-Oppenheimer–Volkoff equations derived from the scalar–tensor formulation of 4DEGB gravity for $X \in \{3, 4\}$ and Gauss–Bonnet coupling $\alpha \in \{0, 1, 10\}\,\mathrm{km}^2$. For the soft EOS ($X = 3$), the maximum mass increases from $2.0431\,M_\odot$ (GR) to $2.5013\,M_\odot$ at $\alpha = 10\,\mathrm{km}^2$, with only the latter entering the PSR~J0952$-$0607 mass band. For the stiff EOS($X = 4$), the GR baseline already yields $3.0415\,M_\odot$, and all configurations exceed both the GW190814 secondary mass and the PSR~J0952$-$0607 constraint. The compactness $C = M/R$ at maximum mass ranges from $0.2531$ to $0.3123$ across all configurations, remaining well below the Buchdahl bound throughout. Radial profiles of the squared speed of sound confirm that $c_s^2 < 1/3$ holds pointwise throughout the stellar interior for all parameter choices, establishing that the quark matter remains sub-conformal and causal inside the maximum-mass star. These results demonstrate that higher-curvature corrections in 4DEGB gravity systematically enhance the maximum supported mass of perturbative QCD quark stars, with the stiff ($X = 4$) branch already exceeding the most massive compact objects currently observed even at the GR level, while the soft ($X = 3$) branch requires $\alphaGB \sim 10\,\mathrm{km}^2$ to approach those thresholds.
Read the full abstract

All authors Anirudh Pradhan, Safiqul Islam, Safyan Mukhtar, Santosh Kumar Dixit

Computer Vision

Synthetic Leprosy Image Generation Using Mask-Conditioned Latent Diffusion and Transfer Learning from Large Chronic Wound Datasets

Machine learning for neglected tropical diseases is limited by data, not algorithms: public annotated image sets for leprosy (Hansen's disease) number in the hundreds. We ask whether a model trained on abundant chronic wound photography transfers to this low-data regime. We build a three-stage pipeline: a DeepLabV3-ResNet50 segmenter supplies lesion masks for two wound datasets that ship without them; a mask-conditioned latent diffusion model built from Stable Diffusion 1.5 components is trained on 3,280 wound crops; and the model is fine-tuned on leprosy image-mask pairs from CO2Wounds-V2 (764 images of 96 patients). We evaluate on 136 held-out leprosy lesions using LPIPS anchored by a real-versus-real baseline, kernel inception distance (KID), and a paired design that runs both models on the same input. Fine-tuning moved the generated distribution toward real held-out leprosy images: KID fell from 0.167 to 0.089, and LPIPS distance to real images from 0.806 to 0.765, with non-overlapping 95% confidence intervals. Giving the base model the leprosy prompt alone did not reproduce the gain (KID 0.250). For a fixed input, fine-tuning changes the output less than a new random seed does (paired LPIPS 0.544 versus 0.728): it keeps layout and alters tissue color and texture. The generated set shows no mode collapse (internal diversity 0.734 versus 0.667 for real images). A dataset audit shows that a majority of training images carry synthetic flat regions from preprocessing, which the fine-tuned model partly reproduces. Chronic wound photography is a viable donor domain for leprosy lesion synthesis; the remaining barrier is semantic control rather than image quality.
Read the full abstract

2609.13226 · 1 Sep · 15 pages, 12 figures, 6 tables. Code: https://github.com/Yusufa09/SyntheticImageGeneration. v2: retrained with a held-out test split and an expanded evaluation (KID, paired comparisons, confidence intervals). Fine-tuning now measurably shifts outputs toward real leprosy images; this supersedes the base-vs-fine-tuned comparison and conclusion in v1, which had an evaluation error

Computer Vision

Part Grounding, Not Action Knowledge: Locating the Bottleneck in VLM Affordance Prediction

Benchmarks agree that vision-language models reason poorly about low-level manipulation, but an aggregate accuracy score does not say which step fails. We separate two steps that affordance questions conflate: identifying which part of an object to act on, and knowing what action that part requires. Across 19 articulated objects we asked eight models, spanning three developers, what motion a robot should apply. Under an open prompt, push was produced once in 64 evaluations where it was correct, despite being correct for 8 of 19 objects and appearing in the offered label set every time. Inspecting the outputs showed why: models described a different part than the one being scored, e.g. explaining how to pick up a camera rather than press its button. Naming the target part raises action accuracy by 0.32 to 0.63 for every model, from 0.158-0.474 to 0.684-0.947, and push recall from 0-1/8 to 7-8/8. No model beats a constant answer that ignores the image under the open prompt; once the part is named, all eight do. Asked to describe the same part in free prose with no label set, models produce pressing language for 6 to 8 of 8. These results are hard to reconcile with missing action knowledge, and instead point to part grounding as the dominant bottleneck, a pattern that holds across all three model families and does not diminish with capability. Naming the part supplies the grounding variable, so this bounds what a perfect part detector would offer rather than demonstrating a general model of mechanics. Two supporting results agree: on real photographs only three of eight models localize grasp points better than a constant baseline, and on rendered objects none do. We also document two measurement errors of our own, a threshold that let a constant baseline score 0.929 and a labelling rule wrong on 4 of 19 objects, both caught only by testing our numbers against trivial alternatives.
Read the full abstract

Digital Libraries

AI-Research Agents in the Wild. From GitHub and arXiv to Regularities and Gaps

AI-research agents, or autoresearch systems, combine language models with tools, search, evaluation, and iterative modification of research artifacts. Their public software ecology is hard to compare because repositories, papers, benchmarks, libraries, and companion artifacts are often counted as one population. We connect two agent-drafted, human-adjudicated registries frozen on 10 June 2026: 139 canonical public repositories and 101 papers on AI-assisted research systems, each record carrying an evidence card anchored to its primary sources. Nine promoted design lineages contain 59 memberships among 52 repositories, seven of them assigned to two lineages. Against this evidence we examine six time-stamped candidate design regularities. A prospective audit of 25 newly ingested repositories observed zero of six specified trigger events, supplying bounded support for R1-R4; broader corpus evidence narrowed the original scope of R5, and a failed directional prediction leaves R6 exploratory. The paper-repository graph contains 212 curator-assigned relatedness edges: 64 of 101 papers carry at least one edge, the edges reach 23 of 139 repositories, and the six most-linked repositories carry 120 of the 212 edges (56.6%). A full-text check of seven papers shows the relation measures our curation, not the papers' citation behavior. An identity audit adjudicated against commit histories and public contributor records confirms at least 18 paper authors who also author a canonical registry repository (20 of 63 candidate author slots in 13 of 101 papers); a byline-matched tier reaches 58 identities but is reported separately and more weakly. The study contributes a source-grounded theory under prospective test, a curator-assigned map of paper-repository relatedness, and explicit limits on claims about public visibility and community structure.
Read the full abstract

General Physics

Do Quantum Measurement Advantages Survive to Astrophysical Inference? Seven Benchmarks in Optical Interferometry and Imaging

Quantum techniques matter to optical astronomy when they preserve information in the parameter-bearing measurement channel after realistic loss, calibration, covariance, and inference are propagated to the final astrophysical observable. We compare 7 approaches by the quantity they change: internal optical-path delay, sub-Rayleigh source separation, coronagraphic planet throughput & stellar leakage, coherent field quadratures, nonlocal complex visibility, null depth, active phase sensing. In an Earth-analog astrometry case, 6-dB squeezed internal metrology improves the measured phase quadrature by 1.58 and the complete readout by ~1.20, but fixed-time planet-mass precision by only 1.010 (95 % interval 1.005-1.045); mission duration, useful baseline, cadence, calibration have greater modeled leverage. For a close young binary, signed spatial-mode demultiplexing preserves sub-Rayleigh orbital information that is substantially degraded in the tested direct-imaging comparator. A representative matched-model calculation gives a dynamical-mass RMSE gain 1.77 1.47-2.15). For illustrative equal-prior pre-main-sequence predictions at 0.10 and \(0.12M_\odot\), mass precisions of 15%, 10%, 7% correspond to correct-selection probabilities of ~73%, 82%, 90%. The numerical gain remains provisional because the event-level continuous-image comparator, strict tail convergence, externally measured broadband transfer matrix, and finite-photon calibration are not yet closed. Modal coronagraphy and quantum-enhanced heterodyne reception define conditional receiver opportunities, while the nonlocal, nuller, and prepared-NOON cases identify physical bottlenecks that prevent an astrophysical advantage in the stated configurations. The common result is that quantum technology is most valuable when it changes an information-bearing observable and that channel remains consequential in the final science measurement.
Read the full abstract

General Physics

The quantum harmonic oscillator from binary sequences

The quantum harmonic oscillator (QHO) is normally built on a structure that contains a Hilbert space, ladder operators, and a Born rule. Here we derive it from counting. We adopt three information-theoretic postulates: information is carried by binary sequences of length $n$; only symbol counts, not the sequences themselves, are observable; and the transition measure must account for exactly the $2^n$-configuration capacity of the register. The first two postulates force the counting problem into the Hamming/Delsarte association scheme and endow it with an exact symmetry under relabeling of the two symbols. Given an exchange-symmetric bilinear transition measure we prove a rigidity theorem: relabeling symmetry and normalization together force the alternating sign $(-1)^t$ which drives interference. The variable that carries the sign, i.e. the overlap between input and output sequences, or equivalently their Hamming distance $d_{ab}$, is a hidden quantum number of the transition as a whole, defined only by the two endpoints jointly and unknowable from either alone. Interference thus enters as alternating-sign bookkeeping over a nonlocal, relational variable $d_{ab}$, rather than as a separate dynamical ingredient. This forced weighting makes the transition probability proportional to the square of a Krawtchouk polynomial, whose difference equation has an exactly equally spaced spectrum and which converges to the QHO as $n\to\infty$. At fixed excitation, deviations from the QHO are $\mathcal{O}(1/n)$, and are testable in laboratory realizations of the oscillator, since any real system has finite information capacity. In this framework, interference, quadratic (Born-rule-like) probabilities, and the particle-hole symmetric oscillator spectrum are consequences of counting; only the bilinear form of the measure is assumed (though well motivated) rather than derived.
Read the full abstract

General Physics

Study of dynamical dark energy using DESI DR2 BAO, CMB, and Type Ia Supernovae data

Dark Energy Spectroscopic Instrument (DESI) Collaboration reported an evidence for dynamical dark energy model using baryon acoustic oscillation (BAO) measurements from DESI Data Release 2 (DR2), cosmic microwave background (CMB), and three Type Ia Supernovae (SNe) data (Pantheon+, Union3, and DES-SN5YR) recently. The significances are from $2.8\sim4.2\sigma$ depending on which SNe sample is used, and the highest significance is obtained based on the DES-SN5YR data. But more recently, Dark Energy Survey (DES) Collaboration recalibrated the DES-SN5YR data, and a significance of only $3.2\sigma$ is obtained based on the calibrated data, which is called DES-Dovekie. In this work, we perform a comprehensive analysis of five representative dynamical dark energy parametrization models, including Chevallier-Polarski-Linder (CPL), Barboza-Alcaniz (BA), exponential (EXP), logarithmic (LOG), and Jassal-Bagla-Padmanabhan (JBP), using DESI DR2 BAO, Planck CMB, ACT DR6, and SNe data from Pantheon+, Union3, and DES-Dovekie. Similarly, a significance of around $3\sigma$ is obtained based on the DES-Dovekie data. And the significances are $2.1\sim3.7\sigma$ for different SNe samples and different parametrization models, indicating an evolutionary dark energy dynamics. Furthermore, a phantom-like behavior in the past and has transitioned into quintessence-like behavior today for the dynamical dark energy is preferred.
Read the full abstract

Neurons and Cognition

Pulling Illusion in Individuals with Neurological Disorders

The pulling illusion induced by asymmetric vibration stimuli has attracted attention for its potential applications in rehabilitation and sensory assessment. However, the underlying mechanism of the pulling illusion remains unclear. This study addressed the central question of whether peripheral vibrotactile sensitivity alone is sufficient for the illusion to emerge or whether processing beyond basic vibration detection is also required. Neurological disorders can involve impairments at different levels of the nervous system, providing an opportunity to examine this question. Accordingly, we evaluated directional discrimination performance for the pulling illusion and fingertip vibration detection thresholds in 25 participants with diverse neurological disorders affecting different levels of the nervous system, from peripheral to central. Clustering analysis identified contrasting profiles, with high directional discrimination performance despite elevated vibration detection thresholds and chance-level performance despite relatively low-to-intermediate thresholds. In the generalized linear mixed model, motor-related signs, including hemiplegia and tremor, showed a robust negative association with directional discrimination performance, whereas vibration detection threshold was not robustly associated with performance. Furthermore, in participants with hemiplegia, directional discrimination performance was around chance level on the affected side and close to 100% on the unaffected side, despite stimulus amplitudes well above the measured vibration detection thresholds on both sides. Collectively, these findings suggest that the pulling illusion depends on perceptual processing beyond basic vibration detection, through which asymmetric vibration is experienced as directional pulling.
Read the full abstract

All authors Takeshi Tanabe, Satoshi Yamamoto, Toru Yamada, Daisuke Ishii, Yutaka Kohno

General Relativity and Quantum Cosmology

One Gravity, Many G's: Generalized Horizon Entropy and Gravitational Susceptibility

Effective and thermodynamic derivations of gravity theory in the context of generalized entropy can give rise to multiple, and potentially varying, gravitational couplings, raising the question of which quantity should be identified with the physically measured Newton constant. Motivated by this ambiguity, we formulate the notion of gravitational susceptibility of the horizon. In spacetimes containing multiple horizons, if the entropy is a function on a multi-horizon state space, $S=S(A_1,A_2,\cdots,A_n)$, then the derivatives of the entropy with respect to individual horizon areas define the horizon gravitational response coefficients. Different horizons may possess distinct gravitational responses without requiring multiple fundamental gravitational couplings, analogous to direction-dependent response coefficients in macroscopic media. This framework naturally suggests the possibility that horizon entropies (such as the cosmological apparent horizon and a black hole horizon) can influence one another, which could explain the black hole-cosmology coupled "black hole mass growth" previously discussed in the literature.
Read the full abstract

General Relativity and Quantum Cosmology

On scalar and electromagnetic perturbations of the root–Kerr object

We study scalar and electromagnetic perturbations of the root–Kerr object. Its source-free exterior carries the electromagnetic field obtained in the vanishing-mass limit of the Kerr–Newman black hole, while its physical source is a rotating disk with an essential distributional rim contribution. For a massless charged scalar, the minimally coupled Klein–Gordon equation separates into a confluent-Heun (CHE) radial equation with a direct coupling between the scalar and background charges. We give the associated Nekrasov–Shatashvili (NS) dictionary and show that the disk is an ordinary radial point. For electromagnetic pertubations, we derive the equations for the radiative Newman–Penrose Maxwell scalars directly. They have the same confluent-Heun structure, but a Maxwell perturbation has no bulk coupling to the background potential. Charge-dependent photon scattering data therefore cannot be fixed by the exterior equation alone. We formulate the disk/rim boundary-data problem, identify the response data required to convert exact CHE/NS connection coefficients into a physical scattering matrix, and show how all-spin helicity-flip Compton information constrains the response on the radiative subspace. The analysis separates exact exterior propagation and amplitude matching from the additional source dynamics required for a microscopic response.
Read the full abstract

Statistical Methodology

Metatheoretical multiverse analysis: Improving the reliability of research with theories-as-data

Here we present a method for measuring, analyzing and reducing theoretical uncertainty. We call it metatheoretical multiverse analysis (MMA). The method is important because the reliability and replicability of scientific observation and testing of a phenomenon are a function of theoretical uncertainty thus reducing it is a method for improving theory. The MMA method requires encoding theory as propositional logic between variables (nodes) and their relationships with one another (edges). Propositional logic models are treated as causal path models in our method, which enable mathematical properties of testing and identification underlying their causal propositions. By encoding theories-as-data including their unknown components, it is possible to generate a multiverse of alternatively plausible theoretical models which can then be meta-analyzed using techniques borrowed from (statistical) multiverse analysis. We introduce metrics for studying theoretical multiverses that enable measuring and identifying the causes of metatheoretical uncertainty. This guides researchers to where it is best to invest theoretical development to reduce uncertainty. We use three simulations and a dedicated software package to demonstrate this method.
Read the full abstract

Image and Video Processing

Reliability-Aware Hybrid-K Ensemble Selection for Cervical Cytology Classification: Integrating Discrimination, Calibration, and Selective Prediction

High classification accuracy alone is insufficient for clinical image analysis, where calibrated confidence and reliable uncertainty estimates are essential. This study proposes a reliability-aware Hybrid-K ensemble selection framework for multiclass cervical cytology classification using the SIPaKMeD dataset. Nine deep learning architectures were evaluated using a fixed stratified five-fold partition and three training seeds. After post-hoc temperature scaling, models were assessed using macro-F1, accuracy, AUROC, expected calibration error (ECE), worst-class ECE (WC-ECE), area under the risk-coverage curve (AURC), Brier score, and negative log-likelihood (NLL). Models were ranked using an equal-weight composite score, and Hybrid-K ensembles were formed from the top-ranked models using soft voting. Robustness was examined using 5,000 Dirichlet-sampled metric-weight vectors, leave-one-metric-out analysis, and corrected paired testing across 15 fold-by-seed evaluations. The final Hybrid-2 ensemble, comprising Swin-Tiny and TinyViT-5M, reduced AURC by 43%, NLL by 17%, and WC-ECE by 36% relative to the best individual model. It was selected in 96.8% of random weighting scenarios, remained unchanged across all leave-one-metric-out analyses, and improved the full composite score. However, per-metric gains were not statistically significant after Holm-Bonferroni correction (all adjusted p >= 0.168). Because post-hoc calibration did not use a fully independent calibration set, calibration-dependent results should be interpreted as exploratory internal estimates. Overall, the framework identified a compact ensemble robust to alternative metric weightings and improved reliability point estimates under internal validation on a single dataset.
Read the full abstract

General Mathematics

Tensor constructions for Euler magic matrices and proper examples of orders 9, 27, 81 and 243

An Euler magic matrix is an integer matrix $M$ satisfying $MM^{\mathsf{T}}=\gamma I$ together with two diagonal square-sum conditions; it is proper when its entry squares are pairwise distinct. Müller proved that Euler magic matrices exist in every order other than $3$, and that no Euler magic matrix of order $3$ exists at all, while proper examples are considerably more restrictive. We describe a tensor construction whose factors are only required to satisfy the orthogonality equation $AA^{\mathsf{T}}=\gamma I$: for a linear reindexing $L$ of the row index group $\mathbb{F}_3^k$ obeying an explicit support condition, the reindexed Kronecker product of $k$ such $3\times3$ factors satisfies the full Euler magic conditions in order $3^k$. Choosing factors whose entry squares have pairwise distinct products, we obtain proper Euler magic matrices of orders $9$, $27$, $81$ and $243$. The construction therefore produces proper examples in powers of three even though order three admits no Euler magic matrix, and the passage from the factors to the product is exactly where the Euler conditions are created rather than inherited. We give an explicit support condition on $L$ and show that it characterises the linear reindexings forcing the two Euler diagonal identities for every tuple of semi-magic factor arrays; over $\mathbb{F}_p$ with two factors, such a reindexing exists only when $p\le3$. The four existence results are formalised in Lean 4, as are the two instances of the construction used to obtain them; the order-$243$ certificate is taken from the archived development and was not rebuilt in preparing this paper, although its witness was reproduced here by exact integer arithmetic. Further proper examples of orders $729$ and $2187$ are verified by exact integer computation only.
Read the full abstract

Multiagent Systems

DART: A DAG-Based Reputation and Incentive Framework via Blockchain-Enabled Governance for Trustworthy LLM Multi-Agent Collaboration

Large language model (LLM)-based multi-agent systems (MAS) predominantly rely on centralized orchestration and lack formal verification mechanisms for agent reliability, participation, and system-level behavioral alignment. These shortcomings leave open environments severely vulnerable to uncooperative or malicious agents. This work proposes DART, a Directed Acyclic Graph (DAG)-based reputation and incentive regulation framework for trustworthy multi-agent collaboration, combining centralized operational orchestration with blockchain-enabled decentralized governance and accountability. DART unifies DAG workflow orchestration, capability and reputation-aware task allocation, dynamic behavior updates, multi-factor incentives, and smart contract accountability paired with IPFS storage. Under this paradigm, agent selection dynamically balances task alignment, historical reputation, and workload, while post-execution behavioral evidence continuously calibrates agent trust and the probability of future participation. Evaluated across four axes, DART achieves 93.6% Pass@1 on GSM8K and builds a full-stack application in 142 s using two agents, outperforming centralized baselines. Across five independent 150-round longitudinal trials, Full DART achieves a mean task success rate of 93.33 +/- 2.26%, output quality of 0.9357 +/- 0.0117, retry rate of 0.2307 +/- 0.0816, and allocation delay of 1.1153 +/- 0.0408 s, consistently outperforming its ablated configurations DART isolates persistent and intermittent malicious agents, obtaining a 99.3% output containment rate and restoring system success to 99.8%. These results demonstrate the potential of coupling reputation, incentives, DAG-based coordination, and verifiable blockchain-enabled governance to support adaptive and accountable multi-agent collaboration.
Read the full abstract

Instrumentation and Methods for Astrophysics

Photon-Ring Retarded- Time Tomography: The Mahakal Phenomenon

Near-critical null geodesics generate image-order channels with distinct mappings, delays, and weights, so one observer-time data set samples multiple source epochs. We organize the inverse problem by historical reach, dimension, and held-out recovery. Using continuum-noise-consistent, matrix-free Kerr/Schwarzschild operators for orders n = 0, 1, 2, we audit 12 spin-inclination geometries. At normalized SNR0 = 100, resolved orders extend anchor-connected history beyond direct imaging in all cells by 40 M, 24 M, and one 4 M grid step at inclinations 20 degrees, 50 degrees, and 75 degrees. Operational rank rises from 153 to 202 and old-age innovation from 8.29 M to 40.72 M; complete order summation retains only 9.66 M. Compact temporal support gives an exact result: temporal basis functions outside the direct retarded-time footprint generate identically zero direct columns, while higher orders can lift those old-epoch blind blocks. Direct old-structure rank is zero, while resolved orders support 38, 93, and 185 operational directions across three classes. Yet source enrichment leaves reach nearly fixed while rank fraction and singular-value stability collapse; this is the Mahakal phenomenon. Held-out tests show bounded benefit, not a movie. Resolved orders extend stable baseline-inclusive emissivity-level span from 48 M to 80 M, but that endpoint is level dominated. On 60 histories, physical morphology error falls by 13.3 percent with TSVD and 16.4 percent with ridge. Reliable two-feature recovery fails and the stable morphology interval is zero. Complete order summation yields no material morphology gain. The positive results are therefore best-case, known-geometry image-domain benchmarks under ideal order separation; SNR0 is an operator normalization, not an observation.
Read the full abstract

Artificial Intelligence

Beyond "AI Helps Humans": Decision-Targeted Evaluation Design for Human-Agent Teams in the Agentic Era

Wherever a coding agent works under engineer supervision, or a clinical model assists a radiologist, the deployment question is whether to keep the human-AI workflow or replace it with the human alone or the agent alone. The human-AI workflow is worth keeping only if it beats both of those alternatives. Yet once it is deployed, neither alternative outcome is observed: recovering one means replaying the task under that alternative, and every replay costs expert time or compute. Under a fixed replay budget, the design question is therefore which tasks should be more likely to receive a human-only replay, and which an agent-only replay. Existing methods do not directly target this decision. Agent benchmarks do not choose which missing baseline to measure, variance-based sampling ignores which of the two comparisons is closer to failing, and Bayesian information methods focus on learning model parameters instead of making the deployment decision. We propose TEAM-Design, a rule that gives every task two replay probabilities, one per baseline. It raises a probability where the missing baseline outcome is hard to predict from what is already known about the task and where that comparison is harder to establish, and lowers it where replay is expensive. We prove that the rule solves this budgeted design problem, and that drawing the replays at random from recorded probabilities still controls the chance of wrongly declaring that the workflow beats both. We reanalyze 6 clinical settings, where no human-AI workflow beats both alternatives, and a coding benchmark, where one does, then evaluate TEAM-Design on synthetic designs and on a semi-synthetic design built from a real chest X-ray reader study. TEAM-Design works best when one of the two comparisons is clearly harder to settle than the other, and can do worse than variance-based allocation when the two are similarly difficult.
Read the full abstract

General Mathematics

Bernoulli polynomials series via analytic summability of functions

Analytic summability of real and complex functions was introduced in 2016. In the topic, the Bernoulli numbers and polynomials were used for defining analytic summand of a given function with a power series on an open domain $D$. In this paper, we study derivatives and integrals of the analytic summand functions and show that the derivatives are the same Bernoulli polynomials series $\sum_{n=0}^{\infty}c_nB_n(z)$ up to a unit forward shift. Therefore, the topic of the analytic summability is a platform for studying Bernoulli polynomials series. In the way, we obtain many new results for the Bernoulli polynomials series such as some related series convergent tests and upper bounds for the Bernoulli polynomials and the mentioned series. For instance, we observe that $\sum_{n=0}^{\infty}c_nB_n(z)$ is absolutely convergent on $\mathbb{C}$, if the numerical series $\sum_{n=0}^{\infty}\frac{n!}{\pi^n}c_n$ is absolutely convergent. Also, we present some applications and various examples of the topic such as the inequality $$|\sum_{n=1}^{\infty}\frac{\pi^n}{n!n^p}B_n(z)|\leq 2e^{\pi|z|}\zeta(p)$$ held for every fixed real number $p>1$ and all $z\in \mathbb{C}$.
Read the full abstract

Computer Vision

DIVA: Exploiting Cross-Step Conditional Propagation for Visual Jailbreaks in Discrete Diffusion Vision-Language Models

Large vision-language models (VLMs) are increasingly deployed in safety-critical settings, yet existing visual jailbreak research has focused almost exclusively on autoregressive architectures, leaving an important emerging family unstudied: multimodal discrete diffusion vision-language models (dVLMs). We identify a vulnerability specific to diffusion generation: because the visual embedding conditions every reverse denoising step rather than acting as a one-time prefix, adversarial visual semantics are repeatedly propagated and amplified across the generation trajectory, a phenomenon we term cross-step conditional propagation. We provide empirical evidence through stage-sensitivity analysis, prompt-level switch rates, and pairwise denoising-bin disagreement metrics, confirmed by bootstrap resampling. We propose DIVA (Discrete-diffusion Vision-language model Attack), a white-box visual jailbreak framework using cross-modal intent obfuscation and diffusion-aware multi-timestep adversarial optimization. Across three dVLMs, DIVA reaches 58.8%, 67.7%, and 69.1% HADES ASR under the Beaver reward-model metric, outperforming visual jailbreak baselines designed for autoregressive models. Code: https://github.com/loststars2002/DIVA
Read the full abstract

All authors Guorui Song, Runqing Tang, Jingye Zhang, Luyuan Zhang, Feice Huang, Cong Ray, Guocun Wang, Dake Zhong, Choo Sin Wai, Bingquan Dai, Chuming Wang, Tongxu Lin, Wanyu Guo, Haoqian Wang

Popular Physics

"It's getting away from us!" - Black Hole Horizons and Relative Speed

Black holes hold considerable fascination for the general public and students alike, and are commonly included in general-science courses on astronomy or modern physics. But teaching about the basics of black holes poses a considerable challenge: Any rigorous description requires concepts and techniques from general relativity, Einstein's theory of geometry and gravitation. And any half-way rigorous introduction to that theory, including the required mathematical tools, is significantly beyond the level of general-science courses. Inevitably, accounts of relativistic physics at the introductory undergraduate level make use of analogies, approximations and simplified models to teach about topics like black holes, gravitational waves, gravitational lensing, or cosmology. The purpose of this article is to given an account of one particular set of analogies for teaching about black holes, all of which are based on modelling the motions of observers in the vicinity of the black hole and rely on the concept of (relative) speed to describe properties of the black hole. While most elements of what I am about to describe can be found in the existing literature, I am not aware of any text that attempts to pull them together into a unified picture at a suitable level of presentation for undergraduate-level teaching; that is the goal of the present text.
Read the full abstract

Computer Vision

Infrastructure-based Monocular 3D Vehicle Localization Framework with Experimental Validation

This paper presents a one-stage learning framework that maps monocular roadside-camera images directly to vehicle states in a ground-fixed coordinate frame. Unlike conventional approaches that first detect vehicles in the image plane and subsequently apply geometric post-processing, the proposed method leverages features from a pretrained object detector to jointly estimate each vehicle's ground-plane position, dimensions, and yaw angle. The framework therefore uses visual features not only for vehicle detection but also for direct spatial and orientation estimation. To support model training and evaluation, we develop a data-collection and label-generation pipeline based on synchronized video from a roadside camera and an unmanned aerial vehicle (UAV). Acting as a temporary top-view sensing platform, the UAV provides vehicle trajectories, dimensions, and orientations, which are transformed into the ground-fixed coordinate frame and temporally aligned with the roadside-camera images to generate ground-truth labels. The framework is evaluated using data collected during multiple experiments at the Mcity Test Facility. Results show that the proposed method can recover vehicle trajectories and orientations from monocular roadside imagery without a separate geometric post-processing stage, demonstrating its potential as a scalable approach to infrastructure-based perception at urban intersections.
Read the full abstract

All authors Akos T. Kopeczi-Bocz, Tian Mi, Gabor Orosz, Denes Takacs