Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Wednesday 9 September

Fluid Dynamics

Giesekus Stick-Slip Singularity: Asymptotic Theory in the Log-Conformation Formulation

In this paper, the log-conformation reformulation of a Giesekus fluid near a planar stick-slip singularity is investigated analytically. In what is, to our knowledge, the first use of the log-conformation formulation beyond its numerical purpose, we present results that match the asymptotic stress analysis of Evans [JNNFM 222 (2015) 24-33] for the stick, slip, and core regions under the assumption of a given Newtonian velocity field. Furthermore, the logarithmic reformulation allows us to extend the existing asymptotic theory substantially, showing, e.g., that the determinant of the conformation tensor is asymptotically constant, as well as thoroughly characterizing the transition behavior between the boundary layers and the core region. Combining these results, we obtain a composite asymptotic solution of the log-conformation equation that is valid uniformly throughout a close neighborhood of the singularity.
Read the full abstract

Operator Algebras

Von Neumann algebras as reduced twisted groupoid $C^*$-algebras

We characterize the von Neumann algebras that are isomorphic, as $C^*$-algebras, to reduced twisted $C^*$-algebras of locally compact Hausdorff groupoids equipped with continuous Haar systems of full support. They are exactly the subhomogeneous von Neumann algebras, equivalently finite products of matrix algebras over abelian von Neumann algebras. The groupoid can always be chosen compact, principal, and étale, with trivial twist and counting Haar system. We also prove that, for a groupoid in this class, the full or reduced twisted algebra is unital if and only if the groupoid is étale with compact unit space. This criterion reduces the classification for general Haar systems to the étale case. A further obstruction comes from controlled propagation, defined through faithful representations into the norm closure of uniformly sparse matrices. Every reduced twisted étale groupoid algebra and its Borel completion have controlled propagation, including for non-Hausdorff groupoids with locally compact Hausdorff unit space. For every infinite-dimensional Hilbert space $H$, every $*$-homomorphism from a nonzero quotient of $B(H)$ into an algebra with controlled propagation is zero. In particular, neither $B(H)$ nor the Calkin algebra embeds into any of these reduced or Borel groupoid algebras. We also prove that every von Neumann algebra with controlled propagation is finite. No separability or countability assumptions are required.
Read the full abstract

Quantum Physics

Programmable CMOS DAC Operating in Cryogenic Environments for Controlling Superconducting Qubits

This paper presents the design and test results of a CMOS current-based Digital to Analog Converter (DAC) that operates at cryogenic temperatures and that can be used to precisely control the amount of flux coupled to qubits or that can be used in the readout of superconducting circuits. The current pulse output can be controlled in terms of its amplitude, rise and fall slopes, via digital controls, and pulse width, via external triggers, while driving a superconducting circuit. The design has been implemented in a planar 90 nm CMOS process and test results closely match circuit predictions. The solution's wide degree of digital tunability affords potential application of the same device to many types of quantum circuits, beyond those discussed here. Due to the wide-ranging flexibility of digital CMOS control, we envision that this DAC design will enable the next generation of high-fidelity cryo-CMOS control architectures for superconducting qubits.
Read the full abstract

All authors A. Pelteku, D. Reitz, M. Covington, A. Frederick, H. Adebayo, K.D. Hillaire, N. Mitran, Z. Steffen, R. Schwartz, Z. Stegen, J.X. Przybysz, J. Yager, C. Walter, T. Graves, D. Meyer, J. Clark, G.R. Boyd, A. Pesetski

Physics and Society

Estimating the Time Advantage of Split-Staggered Starts in Record-Breaking Middle-Distance Races

In July 2026, Josh Kerr broke the 26-year-old mile world record, the latest in an unusual concentration of recent middle-distance records that has led to widespread speculation about the role of advancing shoe and other technologies. However, like almost every other recent record-breaking performance, Kerr's race featured a split-staggered start – an often overlooked, relatively novel start configuration that places a subset of athletes in a front group in the outside lanes, reducing the geometric burden associated with accelerating on tight turns while also avoiding the bulk of the extra distance associated with outside positioning in a traditional distance race. Here, these advantages are quantified using empirical lane-performance data, OMEGA measurements of realized race distance, video reconstruction, and a force-based model of curved running. Applying the resulting corrections to record-breaking performances from 2026, estimated advantages range from approximately 0.1 seconds to more than 1 second, depending on the event, race configuration, and model assumptions. Although small in absolute terms, these advantages demonstrate that start configuration represents a meaningful and previously overlooked source of performance advantage in modern middle-distance racing – large enough to make the difference between a world record and an otherwise great performance.
Read the full abstract

Analysis of PDEs

Fractional Hardy–Maz'ya inequality on a half-space

The main purpose of this article is to provide a fractional counterpart of the well-known Maz'ya inequality on the half-space, that is $$ \int_{\mathbb{R}^{d}_{+}}\int_{\mathbb{R}^{d}_{+}}\frac{|u(x)-u(y)|^p}{|x-y|^{d+sp}}dy\,dx\ge\mathcal{D}_{d,s,p}\int_{\mathbb{R}^{d}_{+}}\frac{|u(x)|^p}{x_d^{sp}}dx+C_{d,s,p,\tau}\int_{\mathbb{R}^{d}_{+}}\frac{|u(x)|^p}{x_{d}^{sp-\tau}\left(x_{d-1}^2+x_d^2\right)^{\tau/2}}dx, $$ where $\mathcal{D}_{d,s,p}$ stands for the sharp constant in the fractional Hardy inequality on a half-space $\mathbb{R}^{d}_{+}$. We also obtain a similar result in the setting of Sobolev–Bregman forms.
Read the full abstract

Quantitative Methods

scDEFT: A deep learning framework for drug-effect prediction and counterfactual reasoning

Longitudinal single cell atlases now capture matched pre treatment and post treatment states from responders and non responders, presenting an opportunity to mechanistically explain why two patients on the same drug diverge. We introduce scDEFT (single cell Drug EFfect Transducer), which treats a drug as a conditioning operator on cell representations, enabling prediction and explanation. In scDEFT, feature wise linear modulation produces drug conditioned cell latents, learned under abundant per cell supervision and then frozen. Two independent heads aggregate those latents over shared transcriptional neighborhoods to predict drug induced state change and responder status. A backward stage ranks the latent dimensions by how strongly they separate responders from non responders and maps them to genes under a cell composition control. On a harmonized inflammatory bowel disease atlas of 1.16 million cells, three cohorts and two drug classes, scDEFT predicts state change at 45% of the baseline to reproducibility ceiling headroom and stratifies responders before treatment at AUROC 0.70, where standard predictors remain at chance. These predictions and the drivers behind them support target and co target nomination, patient stratification, and counterfactual prediction of unseen drug cohort effects.
Read the full abstract

Computation and Language

Detectable Only Where It Is Confounded: What Verified Duplication Counts Say About Membership Evidence in Language Models

When a language model finds a sentence unusually cheap to predict, it is tempting to conclude that the sentence was in its training data. Almost every published test of that inference has had to guess which sentences were in the training data, the members, and which were not. This paper removes the guessing. Two model families, OLMo-2 and Pythia, publish their pretraining corpora, and a public index over those corpora returns the exact number of times any sentence appeared in each. Those counts make three questions answerable directly. The answers form a pincer, closing from two sides. At the duplication levels ordinary text actually has, five models from 1B to 13B parameters carry at most a faint trace of their own exposure. We measure that trace with a design that reads the same sentence through two models, which cancels fluency and quality by construction, and it comes to a rank correlation near -0.08, where -1 would be a perfect relation and 0 none. Where the trace does become strong, above roughly a thousand copies, the two corpora agree on which sentences those are, because they are the famous ones, so exposure can no longer be told apart from fame. Two further measurements show how apparent membership signal gets manufactured. A common way to build a non-member is to change one word of a member. The model does prefer the original, but the gap is the same, within noise, whether the original appeared once or a hundred times, so what the model is rewarding is the author's word choice, not memory. Above a thousand copies the gap grows with model size on the twelve sentences we can test there, at the same boundary where the pincer closes. And swapping the controls for sentences that differ from the members in register moves a detector from 0.83 to 0.94 AUC, on a scale where 0.5 is a coin flip and 1.0 is perfect separation. We release the sentence banks, counts, and code.
Read the full abstract

2609.10830 · 9 Sep · Corrects errors in v1: the composition of the mid-book non-members (Sec. 3.3, new App. B), the direction of the Sec. 7 mechanism (the composed controls cost the model more, not less), the Sec. 6 edit gap at 13B, and several overstated sentences; adds Kandpal et al. (2022). No measured result from v1 changes. Code and data: https://github.com/IamArmanNikkhah/detectable-only-where-confounded

Chemical Physics

Selected Coupled Cluster Guided by Selected Configuration Interaction

Coupled Cluster (CC) theory is one of the most popular correlated wavefunction methods, yet the steep computational cost scaling of CC truncated at higher rank limits its application. Modern Selected Configuration Interaction (SCI) methods successfully exploit the sparsity of ground-state wavefunctions in the determinant space, motivating an analogous approach with the exponential CC ansatz. In this work, we introduce Selected Coupled Cluster (SCC), an arbitrary-order sparse CC method that explicitly solves the CC residual equations within a compact amplitude space guided by an SCI wavefunction. SCC optimizes the CC amplitudes in the sparse amplitude space. Using a pilot Just-in-Time (JIT) compiled implementation, we benchmark SCC against SCI, SCI+PT2, standard rank-truncated dense CC (CCSD, CCSDT), and DMRG across various chemical systems. We demonstrate that SCC exhibits fast, monotonic convergence with the SCI variational threshold, outperforming SCI and surpassing SCI+PT2 in the tight-threshold regime. SCC also provides accurate energies for high-dimensional topologies, where DMRG typically converges slowly with bond dimension. SCC thus presents an efficient drop-in enhancement for modern SCI workflows.
Read the full abstract

Analysis of PDEs

Heat equation with singular drift which has bounded solution discontinuous at an interior point

The goal of this paper is to present a one dimensional heat equation with a singular drift for which solutions of the Cauchy problem with smooth initial data are bounded and are discontinuous at an interior point (common to all solutions).

High Energy Physics (Phenomenology)

Generalized Chiral $U(1)_{B-L}$ with Inelastic Scalar Dark Matter for the LZ 248 keV Event

The high-energy nuclear recoil event with $E_R\simeq248$ keV recently reported by the LZ collaboration, with a background-only significance of $2.6\sigma$, offers an intriguing window into dark matter scattering beyond the conventional elastic picture. We study an inelastic scalar dark matter scenario within a generalized chiral $U(1)_{B-L}$ framework. After symmetry breaking, the dark sector scalar gives rise to two nearly degenerate states with a small mass splitting. The lighter state serves as the dark matter candidate and couples off-diagonally to the $Z'$ boson, leading to endothermic inelastic scattering. The resulting kinematics suppress the contribution from low-velocity dark matter while making the high-velocity tail increasingly relevant for nuclear recoils at higher energies. We identify the parameter space consistent with the observed relic abundance and current experimental constraints, and discuss the inelastic scattering kinematics relevant to the recent LZ observation, together with the complementary collider prospects for the associated $Z'$ boson.
Read the full abstract

Machine Learning

Processing and classifying bird songs using wavelet techniques and supervised learning

This study proposes an integrated framework for the processing and classification of invasive bird species vocalizations within natural soundscapes, characterized by high levels of environmental noise. We address the challenge of signal degradation by employing a Bayesian wavelet shrinkage methodology based on the Epanechnikov kernel prior, which offers a closed form decision rule and high computational efficiency for processing large bioacoustic datasets. The methodology was applied to recordings of three species obtained from the iNaturalist platform: Euphonia violacea, Leiothrix lutea, and Passer domesticus. After signal denoising, we extracted a comprehensive set of features, including Mel-Frequency Cepstral Coefficients (MFCCs) and spectral indices such as entropy and zero-crossing rate. Several supervised learning models: Random Forest, Multinomial Logistic Regression and Support Vector Machine (SVM) were evaluated across different feature dimensionalities. Our results demonstrate that the proposed wavelet based preprocessing significantly enhances classification performance, with the SVM model achieving the highest accuracy (up to 0.9398) under a 10-dimensional MFCC configuration. This research provides a robust statistical tool for automated ecological monitoring and the management of biological invasions.
Read the full abstract

All authors Laura Lucia Dominguez Barrios, Fidel Aniano Causil Barrios, Alex Rodrigo dos Santos Sousa, Mariana Rodrigues Motta

Image and Video Processing

Scale-Aware 3D Deep Learning for Robust Brain Metastasis Detection in Multimodal MRI

Detecting brain metastases in magnetic resonance imaging (MRI) remains challenging because lesions vary widely in size and appearance, with very small metastases occupying only a minute fraction of a three-dimensional input. We investigate whether combining different spatial fields of view (FOVs) improves lesion detection in multimodal MRI and present a scale-aware 3D deep-learning framework. The method uses independently trained $96^3$ and $64^3$ 3D U-Nets whose whole-volume probability maps are combined by weighted late fusion. This design allows us to study the effect of spatial context separately from image resolution and modality choice. On a 97-patient development cohort, cross-FOV fusion improved lesion-level precision and F1 while substantially reducing false positives relative to the individual models. A same-FOV ensemble control showed that these gains were not explained solely by averaging independently trained networks, supporting a contribution from complementary spatial context. An exploratory cross-FOV agreement filter reduced false positives but did not improve overall F1. These results support cross-FOV probability fusion as a simple and computationally practical strategy for improving the precision-false-positive trade-off in 3D brain-metastasis detection.
Read the full abstract

Artificial Intelligence

Studying Without a Syllabus: Task-Agnostic Environment Preprocessing

Before an LLM agent tackles tasks in a new environment, it can inspect available corpora and tools and construct reusable resources such as indices, scripts, or procedural guidance. Most automated adaptation methods, however, rely on task examples, trajectories, or evaluation feedback to decide what to build. Existing task-agnostic approaches avoid this supervision but commit in advance to a preparation strategy for a particular type of environment. We study a more open-ended setting: can an agent study an unfamiliar environment without a syllabus, i.e. before test time and without knowledge of the downstream task distribution, and choose how to prepare it? We formalize task-agnostic environment preprocessing, in which a studying system explores an environment under a budget and produces artifacts for a frozen solver. We compare unaided and archive-equipped meta-agents with fixed synthetic-practice and corpus-processing methods across six heterogeneous benchmarks. A meta-agent variant achieves the highest Avg@3 reward on five benchmarks, while fixed corpus processing remains best on the largest corpus benchmark. Larger study budgets do not reliably improve downstream reward. Nevertheless, studied artifacts reduce the test-time sampling needed to reach a given score, demonstrating how reusable preparation can shift computation from repeated test-time attempts to a pre-task study phase.
Read the full abstract

All authors Vinay Samuel, Varun Ursekar, Vijay S. Kalmath, Apaar Shanker, Veronica Chatrath, Yuan Xue

High Energy Physics (Theory)

Interpolating between single and double trace $T\bar T$ by moving LST into the bulk

We describe a large class of theories that continuously interpolate between single and double trace $T\bar T$ deformations of AdS$_3$/CFT$_2$. It is obtained by starting with the spacetime $M_3$, which interpolates between AdS$_3$ in the IR and a linear dilaton spacetime in the UV, and imposing a UV cutoff on the radial direction. When the cutoff is deep in the AdS$_3$ region of $M_3$, this leads to double trace $T\bar T$. When it goes to infinity, we recover the single trace deformed theory. We present a large class of possible boundary conditions on the dilaton, and compute the resulting black hole spectrum in a few examples. In all these constructions, to leading order in the low energy expansion, the resulting spectrum is that of a single $+$ double trace deformed CFT, but at higher orders the different theories have different spectra. In one of the examples that we analyze, the full theory is single $+$ double trace $T\bar T$ deformed CFT. Another gives rise to an interesting spectrum, that consists of an infinite set of energy bands.
Read the full abstract

High Energy Physics (Phenomenology)

A Colour-Casimir Adjacency Matrix Approach to Fully-Heavy Tetraquarks

We present a phenomenological framework for fully-heavy tetraquark spectroscopy based on spectral graph theory. The four valence partons are the vertices of \(K_4\), with edge weights fixed by the colour-Casimir factors of the two colour-singlet diquark–antidiquark channels, \(\bar3\otimes3\) and \(6\otimes\bar6\); a physical state is a coherent mixture of the two, controlled by one mixing angle. Unlike an earlier version, which passed the colour-weighted adjacency matrix through an absolute-degree graph Laplacian, here the spectrum is obtained by directly diagonalising the adjacency matrix; we show analytically that the Laplacian step inverts the physical correspondence between colour attraction and mass ordering. Fixing the model's three parameters (mixing angle, energy scale, effective charm mass) against \(X(6900)\), \(X(7100)\), and the more tentative \(X(7200)\) gives \(\alpha=0.740\), \(\gamma=23.2\,\mathrm{MeV}\), \(m_c=1.77\,\mathrm{GeV}\); with three parameters fixed by three inputs this has zero residual degrees of freedom and is not itself a statistical test. The framework's value rests instead on two genuinely predictive, zero-additional-parameter applications: the all-bottom ground state falls at \(18.7\)–\(18.9\,\mathrm{GeV}\) for standard \(m_b\), with no retuning; and for \(T_{cc}^+\) the same mechanism under-binds by \(\approx93\,\mathrm{MeV}\), a quantitative diagnostic of missing long-range (molecular) dynamics. We discuss the sensitivity to the tentative status of \(X(7200)\), the tension between this static colour-configuration picture and the diquark radial-excitation interpretation favoured by recent CMS data, and what further data would make the model falsifiable in a stronger sense.
Read the full abstract

Mesoscale and Nanoscale Physics

Dissipation-induced nonlocal phase-coherent transport in a topological superconductor

Nonlocal phase-coherent transport via Majorana bound states, so-called teleportation, requires Coulomb blockade to suppress local Andreev reflection. We show that dissipation can replace the need for Coulomb blockade. We study a Kitaev chain coupled to a Markovian bath and identify a regime where the magnitude of nonlocal conductance contribution reach the local one. In this regime, the lowest-energy subspace is maximally mixed, reminiscent of a directly-grounded nontrivial topological superconductor. Phase coherence is confirmed by Aharonov-Bohm oscillations. Our results establish dissipation as a tool for controlling quantum teleportation in Majorana systems.
Read the full abstract

Fluid Dynamics

Long interfacial internal waves with currents: current profiles with quadratic depth-dependence

Internal waves often propagate in the presence of currents. Internal waves and currents are widespread in the world's oceans. The equatorial internal waves and the equatorial undercurrent (EUC) are perhaps the most representative illustration of this phenomenon. The current profile in this study is a quadratic function of depth, which is in agreement with the known results of the EUC. With this assumption the integrable nonlinear Gardner's equation is derived as a model for the wave motion of the thermocline, separating the two layers of slightly different densities. In addition, we derive the generalised Gardner's equation in the case of variable bottom. We demonstrate that the coefficients of the equation are very sensitive with respect to the parameters of the EUC.
Read the full abstract

Analysis of PDEs

Sharp Lifespan Estimates for a Semilinear Wave Equation with Nonlinear Damping

We study the maximal existence time $T^{*}(\varrho)$ of the solution of the semilinear wave equation $u_{tt}-\Delta u=u|u|^{p-1}-u_{t}|u_{t}|^{q-1}$ in a bounded domain, with Dirichlet boundary condition and initial data $(\varrho f,\varrho g)$, where $1<q<p$ and the amplitude $\varrho$ is large. For nontrivial $f$ and sufficiently large $\varrho$, the concavity method gives $T^{*}(\varrho)\leq C\varrho^{-(p-1)/2}$ for $q\leq2p/(p+1)$ and $T^{*}(\varrho)\leq C\varrho^{-(p-q)/q}$ for $q>2p/(p+1)$, whereas the energy method gives a lower bound of order $\varrho^{1-p}$ only. We prove lower bounds with the same exponents as the upper ones, so that $T^{*}(\varrho)\asymp\varrho^{-\vartheta(p,q)}$ with $\vartheta(p,q)=\min\{(p-1)/2,\,(p-q)/q\}$. The proof rests on a hyperbolic rescaling which converts the large amplitude into a dilation of the domain and a coefficient $\varrho^{(q(p+1)-2p)/2}$ in front of the damping term, on a local existence theory in uniformly local energy norms whose existence time does not depend on that coefficient, and on a quantitative use of the dissipation when the coefficient is large. The threshold $2p/(p+1)$ is the value of $q$ at which the damping term is invariant under the rescaling. No attempt is made to optimize the constants: sharpness is meant throughout at the level of the exponent.
Read the full abstract

Solar and Stellar Astrophysics

Frozen-composition effects on rotational failure of neutron-star crusts

The elastic response of a neutron-star crust to rotational loading depends on whether weak interactions can restore chemical equilibrium during the deformation. We quantify this effect using unified equations of state and compare chemically equilibrated and frozen-composition responses within the same Newtonian hydro-elastic framework. Frozen composition increases the effective bulk modulus and redistributes the strain throughout the inner crust rather than producing a simple global rescaling. For BSk24, the breaking frequency decreases from $0.4997\Omega_K$ to $0.3912\Omega_K$, a reduction of 21.7 per cent, while the location of first failure moves to higher density. Comparable reductions are obtained for BSk21, whereas SLy4 gives a smaller effect of 16.7 per cent. For BSk24 the relative reduction remains close to 21–22 per cent over the mass range $1.2$–$2.0,M_\odot$. A density-resolved perturbation analysis shows that the result is controlled by the radial overlap between the microphysical frozen-composition stiffening and the mechanical susceptibility of the crust, with most of the linear sensitivity in the BSk models arising from $0.01<n_B<0.03,{\rm fm}^{-3}$. These results show that chemical non-equilibrium can substantially modify both the rotational failure threshold and the layer in which crustal failure begins.
Read the full abstract

Multiagent Systems

Tapes Together Strong: The Co-evolution of Computation and Cooperation

How does cooperation evolve in complex agentic systems? Prior work in evolutionary game theory studies why individuals are incentivized to cooperate by isolating social interactions from the physical costs of behavior, while artificial life models traditionally study emergent self-replication without formalizing the dilemma between acquiring resources and preserving the shared energy needed to reproduce. In contrast, we introduce Autopoietic Game Theory, a computational model where social interactions, replication mechanisms, and their associated computational costs are endogenous and simultaneously co-evolving. We study these dynamics using a computational substrate of randomly initialized programs in Z80 machine code, showing empirically, and motivating with a simplified theoretical model, that embedding a social dilemma directly into the physics of computation can favor the emergence of self-replicating, cooperative strategies. When resources are scarce, our analysis shows that defection can become self-limiting even in well-mixed populations: parasitic stealing destroys shared energy, slows execution, and can prevent reliable replication. Empirically, evolved programs suppress stealing across several Z80 environments, while spatial assortment further supports structural complexity and task performance. We further show that the framework can incorporate exogenous pressures, such as math tasks structured as sequential social dilemmas, when rewards are tied to computation budgets. These results suggest that coupling an agent's capacity for computation to its available energy transforms cooperation into a dominant scaffolding for building sustainable, self-organizing systems.
Read the full abstract

All authors Kunal Jha, Francesco Cicala, Blaise Agüera y Arcas, Blake Aaron Richards, Natasha Jaques, Max Kleiman-Weiner, Eyvind Niklasson

Instrumentation and Methods for Astrophysics

Magnetising the quokka code with fast, second-order accurate, error-correcting schemes for magnetohydrodynamics on GPUs

We implement a second-order accurate constrained transport (CT) magnetohydrodynamics (MHD) module with first-order flux correction in the open-source, adaptive mesh refinement (AMR), GPU-accelerated code quokka, supporting both ideal and resistive (constant Ohmic) regimes. For computing and averaging the electromotive forces (EMFs) at cell edges, the basis of any CT-MHD method, we experiment with a wide range of recent, state-of-the-art schemes, together with different reconstruction schemes. Alongside these, we develop a new EMF compute scheme that requires fewer reconstruction steps than existing approaches. We evaluate these scheme combinations on the basis of accuracy, stability, and GPU throughput, and demonstrate that our new scheme achieves accuracy comparable to the best-in-class existing methods, together with greater stability in reconnection-dominated flows and roughly 25–60% higher GPU throughput. Using this scheme, quokka achieves excellent results across a wide range of MHD flow regimes, reaching >50 million cell updates per GPU per second, with >70% parallel efficiency out to >500 GPUs. Finally, we confirm that our CT implementation preserves divergence-free magnetic fields (to machine precision) under AMR.
Read the full abstract

All authors Neco Kriel, Elizabeth Cole-Kodikara, Benjamin Wibking, Mark R. Krumholz, Chong-Chong He, Anish Sarkar, Pak Shing Li

Computer Vision

BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation

Police body-worn camera (BWC) footage has emerged as a critical aspect of law enforcement that ensures legal transparency, officer accountability, and the protection of civil rights. However, effectively processing this data remains a significant challenge due to its multimodal video format. BWC videos, in many cases, comprise chaotic scenes with low visual quality, rapid movement/interactions, and high-noise audio that make visual understanding a challenge for even SOTA multimodal models. Current Vision-Language Models (VLMs) frequently overlook critical forensic details, such as the presence of valuable evidence or the latent nuances of suspect-officer interactions, which are vital for fair legal outcomes and civilian/officer safety. To address these limitations, we propose an Adaptive Visual Question Answering (VQA) framework engineered for high-stakes law enforcement. Our framework employs a structured reasoning approach to extract fine-grained visual evidence that traditional captioning systems fail to capture. We experiment with multiple question generation models, including foundation models and fine-tuned open-weight models, to observe performance variation among question generation model implementations. Our results demonstrate that this VQA-driven architecture provides a more reliable, objective, and detailed record of enforcement events, ultimately serving as a powerful tool to protect both law enforcement officers and the public through AI-assisted forensic clarity.
Read the full abstract

All authors Karish Gupta, Matthew Alex, Alex Li, Yang Wu, Yun-Wei Chu, Kashif Munir, Xiaotian Zhou, Zhengping Ji, Xiaozhong Liu

High Energy Physics (Theory)

Spectral Suppression of Asymptotic States in Supersymmetric QFT on de Sitter Backgrounds

We investigate the fate of asymptotic particle states in supersymmetric quantum field theories on de Sitter backgrounds, distinguishing algebraic field content from localized LSZ-like excitations. We study how long-wavelength gravitational fluctuations may suppress pole residues and redistribute spectral weight from isolated contributions toward continuum sectors. The analysis is formulated within a Kallen-Lehmann spectral framework and is intended as a structural infrared mechanism rather than a non-perturbative proof of de Sitter quantum gravity. The paper develops three connected levels. First, we analyze residue suppression in a quasi-de Sitter setting and its possible transmission through Yukawa and gravitational interactions. Second, we formulate the spectral decomposition as a map between the full field-theoretic spectrum and the observable particle sector. Third, we examine a conditional cosmological realization of redistributed spectral weight, including possible effects on expansion, structure growth, and lensing. This cosmological layer is a downstream phenomenological test and is not used as evidence that pole loss has occurred. In the current MCMC analysis we find S8 approximately 0.803 and, when the smooth fraction is varied, fS8 = 0.0025 +/- 0.0017, consistent with the benchmark value 0.00262. These results test the internal consistency of the proposed realization but do not establish the dark-sector assignment or the underlying infrared mechanism.
Read the full abstract

Optimization and Control

Method for Constrained Computational Guidance and Control with Application to Hypersonic Entry

A method for constrained computation guidance and control is described. The method employs guidance updates by periodically re-solving a constrained optimal control problem. Feasibility of the optimal control problem is improved by preventing constraint violations in the perturbed system. Constraint violation is prevented by augmenting the optimal control problem with a penalty term intended to increase the margin between a constraint and its limit when the perturbed system is within certain proximity of the constraint limit. The additional penalty term employs a smooth maximum approximation to increase the margin between the constraint and its limit. The method developed in this paper is demonstrated on the maximum cross-range reusable launch vehicle (RLV) entry problem, for which the reference solution contains a long-duration constrained arc. The results demonstrate that the method developed in this paper is viable for path-constrained optimal guidance and control.
Read the full abstract

Distributed Computing

ExaServe: Large-Scale LLM Serving on Exascale HPC Systems

Cloud-native LLM serving frameworks have made deployment routine in data centers, yet deploying them on leadership-class supercomputers remains an engineering challenge requiring scheduler integration, MPI launch, accelerator selection, node-local weight staging, and platform-specific patches. We present ExaServe, a pip-installable framework that transforms a declarative YAML specification into a reproducible large-scale LLM serving deployment. Using ExaServe, we deploy LLM serving on ALCF Aurora from 1 to 256 nodes (3072 vLLM replicas). Non-streaming inference scales nearly linearly to 256 nodes, reaching 27.1k requests/s (3.8M tokens/s). Token streaming scales differently: a centralized proxy plateaus at about 4.7k requests/s despite the model servers remaining within the service-level objective. We also identify an O(N^2) Ray Serve control-plane bottleneck that increases cluster bring-up to roughly 30 minutes at 256 nodes. ExaServe provides a practical, reproducible deployment path while exposing key barriers to future exascale LLM serving.
Read the full abstract

All authors Wenyi Wang, Shu Shi, Yadu Babuji, Ian Foster, Kyle Chard

Computer Vision

Overpainting: Localized Context-aware Diffusion Image Editing

We present "overpainting", an image editing operation which offers both control over the location of the edit and awareness of the previous content in that location. The overpainted area is given by a trimap, where white-annotated pixels must be edited, gray-annotated pixels may be edited, and black-annotated pixels must not be edited. This enables both precise and loose control, depending on user intent. We implement overpainting by adapting a pretrained image editing diffusion model using a combination of joint attention and low-rank adaption across input images with attention-dropout to balance the information flow between noise, source and mask images. We present a novel, automated, training data generation pipeline that (1) generates a set of candidate image pairs leveraging existing language-based editing models, (2) carefully curates those pairs, and (3) extracts a trimap from each usable pair. We demonstrate the versatility of our overpainting model on a wide range of editing tasks.
Read the full abstract

All authors Sam Sartor, Iliyan Georgiev, Michael Fischer, Valentin Deschaintre, Pieter Peers

Computation and Language

Larger Context Window, Fewer Overcorrections: Optimizing Prompts and Batching for Minimal-Edit Grammatical Error Correction

Minimal-edit Grammatical Error Correction (GEC) is a challenging task for zero- and few-shot prompted Large Language Models (LLMs), which systematically overcorrect and degrade $F_{0.5}$ by rewriting well-formed spans. While fine-tuning provides an effective solution, it imposes substantial infrastructure demands. We introduce a prompt-based approach that closes the gap to fine-tuned models through three advances in GEC prompting methodology. First, we introduce taxonomy-based instructions to enforce minimal-edit constraints with a comprehensive list of grammatical error rules, equipping the LLM with a bounded, metric-aligned scope of correctable edits, which benefits the strongest models while remaining model-dependent overall. Second, we show that batching multiple uncorrected sentences into a single input context acts as a targeted regularizer against overcorrection, systematically reducing the edit rate across diverse LLM families; we hypothesize this arises from attention dilution effect induced by the bounded capacity of self-attention scores. Finally, LLM-assisted Prompt Optimization refines these instructions. Powered by Gemini 3.1-Pro, our prompt achieves $F_{0.5}=78.32$ on the BEA-2019 test set - establishing a new prompt-based SOTA while shrinking the gap to the fine-tuned single-model SOTA (Staruch et al., 2025) to a mere $0.38$ points. Code, prompts, and outputs are publicly available.
Read the full abstract

Earth and Planetary Astrophysics

Fast and Furious: Long-term orbit integrations with collocation integrator Lobbie

Efficient and accurate numerical integration methods are fundamental to studying long-term solar system dynamics, where numerical errors can accumulate over billions of orbital periods. In this work, we investigate the performance of our modification of Lobbie, a Lobatto collocation integrator for long-term simulations of the Solar System. We compare Lobbie against the widely used WHFast symplectic integrator and the high-order adaptive collocation integrator IAS15 implementations in the Rebound package. The performance of the integrator is evaluated using five benchmark problems: a circular two-body orbit integrated over 100 Myrs, two-body orbits with eccentricities up to 1-10^{-6} to test the adaptive step size control during repeated close encounters, the full eight-planet Solar System integrated over 100 Myrs, the four giant planets integrated over 10 Gyrs, and the eight-planet Solar System together with 50 asteroids on Aten-like orbits integrated over 1Myr to test the integrator's consistency in a chaotic, close-encounter-rich environment. Across all tests, Lobbie demonstrates excellent long-term stability and competitive computational performance. For full Solar System integrations, Lobbie achieves relative energy conservation at the level of 10^{-14} while outperforming IAS15 in computational efficiency and approaching the speed of WHFast at comparable accuracy. In the Solar System plus asteroids test, Lobbie also keeps different realizations of Jupiter's orbit consistent with each other to within about 1.5 km after 1 Myr of integration, two orders of magnitude better than IAS15, being twice faster. Our results demonstrate that Lobbie provides a powerful universal alternative to traditional symplectic mappings or Gauss-Radau collocation methods for long-term dynamical simulations in celestial mechanics.
Read the full abstract

Quantum Physics

Tight Time-Space Lower Bounds for Collision Finding and Element Distinctness under Label Symmetry

How much memory is needed to retain the quantum speedup for collision finding? For a uniformly random function $f:[N]\to [N]$, the BHT algorithm finds a collision using $O(N^{1/3})$ queries and a quantumly accessible classical table containing $O(N^{1/3})$ input-output pairs, whereas a logarithmic-space Grover search uses $O(\sqrt N)$ queries. Determining the optimal query-space tradeoff between these extremes remains a major open problem. We resolve this equation within the class of label-symmetric algorithms, which treat the function $f$'s output labels as interchangeable. We prove that such algorithm that makes $T$ queries, uses $S$ qubits, and finds a collision in a uniformly random function $f:[M]\to [N]$ with constant probability satisfies $$T=\Omega(N^{1/3}) \qquad\text{and}\qquad T^2S=\Omega(N\log N).$$ For the setting where $M=N$, these bounds are matched by a space-efficient implementation of the BHT algorithm. As a consequence of our tradeoff, any label-symmetric algorithm for the search version of Element Distinctness on $f: [n] \to [n^2]$ must satisfy $$T=\Omega(n^{2/3}) \qquad\text{and}\qquad T^2S=\Omega(n^2\log n),$$ matching Ambainis's quantum walk. Thus, both tradeoffs are optimal within the class of label-symmetric algorithms. To prove these results, we develop a space-sensitive version of the compressed oracle technique. The compressed oracle records the information learned by the algorithm in an evolving superposition of databases. Using label symmetry and representation theory, we show that an algorithm using $S$ qubits can effectively retain information about only $O(S/\log N)$ collision-free database entries. Substituting this estimate into the compressed oracle technique yields the stated tradeoffs.
Read the full abstract

Instrumentation and Methods for Astrophysics

Instrument-Level Validation of Upper Sideband Syntonization Concept

Upper Sideband Syntonization (USBS) is a two-way frequency-transfer concept proposed by the European Space Agency for distributed space radio interferometric systems. Previous investigations validated the concept primarily through frequency-stability and phase-transfer measurements. This work demonstrates the first end-to-end instrument-level validation of the USBS concept through coherent W-band cross-correlation measurements and image reconstruction. A P=2 USBS breadboard was integrated with a W-band radio interferometer developed for this test. The synchronized local oscillators (LOs) were evaluated through long-term cross-correlation measurements and coherent imaging of a point-like noise source. The results were compared with those obtained using a conventional common LO as well as two independent free-running LOs. The USBS-derived LOs maintained coherence over 500 s of cumulative effective integration time acquired over approximately five days of elapsed time and preserved inter-session phase stability over measurements spanning 40 days. The reconstructed image is nearly indistinguishable from that obtained with a common LO, whereas independent free-running LOs lead to severe coherence degradation, image distortion, and approximately 38 dB loss of peak response. The results demonstrate that USBS preserves the interferometric coherence required for W-band imaging and present the first end-to-end instrument-level validation of the USBS syntonization concept for future distributed space radio interferometric missions
Read the full abstract