Machine Learning
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Wednesday 9 September
On-policy distillation (OPD) provides dense token-level supervision on student-generated trajectories. Although OPD performs strongly when teacher and student belong to the same model family, we find that its effectiveness degrades in cross-family settings even after tokenizer alignment, with substantially stronger external teachers offering little additional improvement. To understand this disconnect, we decompose the cross-family OPD signal into two components: an offset between a low-capability teacher-family reference and the student, and the within-family log-likelihood shift from that reference to the strong teacher. Standard OPD transfers both components together, allowing the offset to dominate the update direction and obscure the changes associated with teacher capability improvements. We propose CompassOPD, which removes this offset and transfers the within-family shift, while a frozen student reference anchors updates to the student's initial policy. Thus, both teacher-side and student-side changes are measured within their respective model families. Experiments across three student families and multiple teacher families show that CompassOPD consistently outperforms standard cross-family OPD, improving average reasoning accuracy by up to 5.50 points. For an MoE teacher, we further construct the reference directly from the teacher checkpoint by reducing expert activation, eliminating the need for a separate reference checkpoint while retaining a 3.43-point gain over OPD.
Computation and Language
YallaMorph: A Benchmark for Evaluating Arabic Morphological Generation in Large Language Models
Arabic morphology remains challenging for large language models, since fluent generation does not guarantee accurate morphosyntactic control. Existing Arabic evaluations mainly target downstream tasks and do not directly test controlled morphological generation from explicit lexical and feature-based input. We introduce YallaMorph, a large-scale benchmark for Arabic morphological generation covering verbs, nouns, adjectives, their cliticized forms, and invalid configurations. We evaluate multilingual and Arabic-oriented LLMs under diacritized and undiacritized settings over 600K benchmark entries. Results show that Arabic morphological generation remains difficult, especially for cliticized, unseen, and morphologically rare forms.
Read the full abstract
Representation Theory
Hybrid Kazhdan-Lusztig basis and parabolic induction
We provide a categorification for the hybrid Kazhdan-Lusztig bases of the Hecke algebra by constructing and investigating a series of new stratified structures on the principal block of BGG category $\mathcal{O}$.
Distributed Computing
CEDD-optimizer: Enabling Cost-Efficient Dataset Distillation on Geographically Distributed Edge Systems
Centralized learning is a fundamental paradigm in modern AI, in which data are collected from distributed edge devices and aggregated at a central host for model training. However, this pipeline is often bottlenecked by the substantial communication overhead of data collection. Dataset Distillation (DD), with its high compression ratio, is therefore attractive for centralized learning on distributed data. Yet, the cost efficiency of DD in non-uniform edge environments remains largely unexplored. Geographically distributed edge devices face different energy and data-transfer prices, while key DD hyperparameters, such as the target compression ratio and number of distillation steps, substantially affect energy use, transfer overhead, and downstream test accuracy. These hyperparameters must therefore be tuned both locally and globally. We propose the Cost-Efficient Dataset Distillation optimizer (CEDD-optimizer), a hyperparameter-tuning framework for cost-efficient distributed DD. It minimizes total cost under a training-quality constraint by optimizing hyperparameter settings across edge devices while accounting for environmental non-uniformity. The framework comprises two modules: CEDD-calibrator and CEDD-solver. CEDD-calibrator identifies parameters for our energy and training-quality models: the former is determined through offline calibration, whereas the latter is estimated online through a three-step tuning scheme. Based on these models, CEDD-solver solves the cost-minimization problem to steer and improve the distributed DD workflow. Experiments across various image datasets show that our approach achieves up to a 20.8 improvement over the baseline DD method under the same quality constraint.
Read the full abstract
Numerical Analysis
Analytical properties of polygonal Stokes and BGG Hessian complexes with application to Kirchhoff–Love plates
We develop and analyse a new arbitrary-order method for the Kirchhoff–Love plate problem on general polygonal meshes. The construction relies on a discrete Hessian complex obtained through the Bernstein–Gelfand–Gelfand construction, which stacks a Stokes complex on top of a tensorised de Rham complex. A key ingredient of the analysis is a detailed study of the former. We specifically establish primal and adjoint consistency estimates, as well as uniform Poincaré inequalities, and use these results to derive the analytical properties of the resulting discrete Hessian complex. The proposed method is then shown to be coercive and to converge with order $k+1$ with respect to the mesh size, where $k\ge 0$ is the polynomial degree of the complex. Numerical experiments on polygonal meshes confirm the theoretical convergence rates.
Read the full abstract
Cryptography and Security
Sound Debloating of Redundant Checks in Zero-Knowledge Machine-Learning Circuits
Zero-knowledge (ZK) proof systems for neural-network inference compile the model into a system of arithmetic constraints. Many of these constraints are redundant checks: range proofs, sign lookups, and bit decompositions who are globally entailed by the rest of the circuit through chains of reasoning that span distant gadgets. Removing them shrinks the circuit and accelerates proving, but the removal must be carefully justified: an unsoundly debloated circuit becomes forgeable, accepting witnesses the original would have rejected and so allowing a prover to claim, for example, that a neural network produced an output it never actually computed. Such soundness vulnerabilities are not hypothetical: under-constrained circuits in deployed ZK systems have enabled attackers to forge transactions and bypass verification entirely. We present an automated framework that removes redundant checks while provably preserving soundness. For each candidate removal, our tool first checks whether the rest of the circuit, on its own, can still rule out every value the removed check was excluding. Using whole-circuit abstract interpretation, the analysis searches for such alternative justifications and records them in a provenance graph; a check is then removed only when an alternative path through the graph still derives the facts that it is checking. This ensures that the debloated circuit opens no new forging strategy to an adversary. We evaluate circuits spanning MLP, CNN, RNN, and transformer architectures generated by two production frameworks (ezkl and zkml), with up to 25.3 million constraints. Our tool removes up to 48.7% of constraints and reduces prover time by up to 72.8%, without weakening security.
Read the full abstract
Number Theory
Infinite Families of Counterexamples to a Conjecture of Liu and Qian and a Refined Inverse Theorem for Restricted Sumsets in $\mathbb{Z}_p$
Let $p$ be a prime and let $A,B$ be nonempty subsets of the cyclic group $\mathbb{Z}_p$ with $|A|\neq |B|$. The Alon–Nathanson–Ruzsa theorem gives the lower bound $|A\rplus B|\ge \min\{p,\,|A|+|B|-2\},$ where $A\rplus B=\{a+b:a\in A,\ b\in B,\ a\neq b\}$ is the restricted sumset. The inverse problem of characterizing all critical pairs $(A,B)$ attaining equality was posed by Alon, Nathanson, and Ruzsa in 1996 and remains open. Recently, Liu and Qian solved the inverse problem under the assumption that at least one of the sets is an arithmetic progression, and proposed a conjecture for the general case. The main purpose of this paper is to show that this conjecture fails in the boundary case $|A|+|B|=p$. More precisely, we construct infinite families of non-arithmetic critical pairs $(A,B)$ with $|A|+|B|=p$ and $|A\rplus B|=p-2$ for every prime $p\ge 11$. These families show that the boundary case is fundamentally different from the non-boundary case. Motivated by this, we formulate and prove a refined inverse theorem under the natural hypothesis $|A|+|B|\le p-1$, which excludes the boundary.
Read the full abstract
Astrophysics of Galaxies
The internal kinematics and chemistry of 20 Milky Way strings
Context: the recent discovery of filamentary stellar structures in the Milky Way disk raises the question of their formation in the context of the Galaxy evolution. Aims: in this work, we aim to kinematically and chemically characterize several filamentary stellar structures by looking for clues about their origin and formation mechanism. Methods: recent works using machine-learning techniques and $Gaia$ data have identified many previously unknown Galactic strings. We cross-matched such data with the kinematics derived from the most recent Gaia DR3 catalog as well as chemical data provided by the Galactic Archaeology with HERMES (GALAH) survey. Results: we found that most strings contain one or more open clusters and are kinematically hotter than open clusters (with internal velocity dispersions in the 1–7 km/s range) but cooler than field stars. 18 objects appear chemically homogeneous ($\sigma_{[Fe/H]}$ <= 0.1 dex), while two strings have a broad [Fe/H] distribution similar to the field stars but exhibit a different mean metallicity, which may identify them as migrating stellar structures. For seven objects the GALAH observations focus on the embedded open clusters; therefore, the chemical abundances are not representative of the whole structure. An object shows clues of chemical gradients along its extension in Galactic longitude. Conclusions: these objects may be evaporating or disrupted open clusters in which the chemical homogeneity is preserved in an unbound and expanding cloud of stars or co-moving stars belonging to a former and now disrupting star formation hub. In a few cases they appear to be moving groups with no chemical homogeneity, likely formed by the dynamical action of the Galactic bar or spiral arms. More data are needed to confirm such hypotheses.
Read the full abstract
Instrumentation and Detectors
Post-fabrication trimming of a 1024-pixel, single-photon counting, Microwave Kinetic Inductance Detector array with a pixel pitch of 150 micron
Microwave Kinetic Inductance Detectors are superconducting resonators capable of single-photon counting with energy resolving capability at visible and near-infrared wavelengths. Their zero read noise, extremely low dark counts, and compatibility with frequency-division multiplexing make them well suited for large imaging arrays. The detector yield of these arrays is often limited by the fabrication-induced frequency scatter leading to resonator collisions. We find that the tight packing of the pixels can add significant frequency scatter as well. We developed a post-fabrication correction method that is suitable for tightly packed arrays with a pixel pitch of $150\ \mathrm{\mu m}$. We demonstrate this method on a 1024-pixel array multiplexed on a single octave of readout bandwidth and show a reduction in frequency scatter from $1.1\times 10^{-2}$ to $3.5\times 10^{-4}$ and an increase in yield from $76\%$ to $94\%$.
Read the full abstract
Optimization and Control
A Robust Binary Nonlinear Solver for Multi-stage Decisions
This document describes two developments. The first is a Multi-stage Decision Framework that permits the formulation of various sequential decision-making problems in a generic manner. The second concerns a robust stochastic solution procedure designed to handle the binary nonlinear optimization problem stemming from the proposed decision framework. The method adopts Monte-Carlo search with incremental probabilistic learning with reinforcement. The two developments are described herein along with a demonstrative application example. Key words: Multi-stage, decisions, binary, nonlinear systems, probabilistic and reinforcement learning
Read the full abstract
Artificial Intelligence
Kernel-Managed Shared Memory for System-Wide Personalization
AI systems become more useful when they can adapt to the people using them, but in multi-agent systems, useful context learned by one agent often remains unavailable to others. We present kernel-managed shared memory, a system-level abstraction in which specialized agents write structured, tagged memories while the agent-system kernel, not individual agents, governs retrieval, privacy enforcement, and prompt injection. We implement and evaluate this design on AIOS and compare it against three alternatives across three assistant models (GPT-4o, Llama-3.1:8B, Qwen-2.5:7B) and 1,800 total trials. Against an unmanaged external memory backend (Mem0) using identical underlying storage, kernel-managed retrieval and injection improve personalization scores by 2.4-4.0 points on a 5-point scale (e.g., 1.05 to 4.69 profile usage on GPT-4o), with every comparison significant at p < 10^-18. Against standard retrieval-augmented injection, gains are similarly large and consistent across all three models. Against full, unfiltered context concatenation, a soft ceiling on available context rather than on response quality, kernel-managed injection statistically matches performance on two of three models and shows a small, model-specific deficit on the third, while using substantially shorter prompts: end-to-end latency is 15-61% lower across all three models, with corresponding reductions in per-call token usage and inference cost. These results indicate that centralizing memory management in the agent-system kernel, rather than leaving retrieval and privacy enforcement to individual agents, delivers most of the personalization benefit of unconstrained context at a fraction of its cost.
Read the full abstract
Analysis of PDEs
Bramson correction and convergence to the critical wave for Fisher-KPP equations on $\mathbb Z^d$
We consider Fisher-KPP equations with nearest-neighbor diffusion on $\mathbb Z^d$, $d\geq2$, with nonzero finitely supported initial data. We prove a logarithmic delay of the front along each signed coordinate axis and show that the transition region has uniformly bounded width. On every fixed-width half-tube around an axis, the solution converges to translates of the minimal-speed lattice traveling wave, with a bounded phase. We also obtain an upper bound with a logarithmic correction in every direction. The proof uses weighted estimates, bounds for tilted random walks, and a product lower solution. Concavity of the reaction is not assumed.
Read the full abstract
Computation and Language
Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning
Large language models remain fragile against malicious fine-tuning, motivating training-time defenses against harmful persona drift. Preventative Steering injects undesirable-trait persona vectors during fine-tuning and removes them at evaluation time, yet the mechanism behind its lasting protection remains unclear. Analyzing its temporal optimization dynamics, we find that the defense emerges from an early compensatory adaptation phase followed by a steady-state phase where the corrective signal decays; in parameter space, attention output projections emerge as the dominant residual-write route for defensive updates. Through Intervention Delta Preservation (IDP) and IDP Continuation experiments, we further show that preserving or reinjecting the weight offset fails to maintain protection, indicating that preventative steering relies on active adaptation rather than a static defense. Motivated by this finding, we propose Progressive Intensity Scheduling (PIS), which starts with a moderate injection strength and increases it after static-strength alignment begins to decay. Across the evaluated Qwen2.5 and Gemma-3 models, PIS improves safety robustness over static-strength steering while reducing harmful trait expression.
Read the full abstract
Quantum Physics
The Sample Complexity of Quantum Entanglement Allocation
How many past requests are needed to decide which qubits should share entanglement? We show that the answer depends on the allocation choices created by the queries: a larger memory can require no more data. The memory stores a classical bit and answers requests through a fixed detector that preserves coherence within each measured sector. For independent commuting $X$- and $Z$-type Pauli queries, we characterize the full attainable prediction-contrast region and construct encodings that preserve the bit at every nonzero vertex. With sharp reports, a $d$-qubit path and groups of at most $k$ qubits have minimax excess error after $m$ requests proportional to $k^{-1}\min\{1,\sqrt{d\log(k+1)/m}\}$, uniformly for $2\leq k<d$. Connected biclique regions can grow without increasing sample demand when depth, region count and connections per region stay bounded. Preparation noise introduces a separate calibration requirement. We derive an exact tradeoff with extra fresh detector calls and transfer the learning law to structured transaction co-location. Population-risk experiments test the statistical predictions. We also compare encodings on a native 15-qubit device and learned partitions on public purchase baskets. The full chain wins on the device; frequency grouping outperforms basket search in the largest-capacity retail setting.
Read the full abstract
Algebraic Geometry
Frobenius–Cartier duality via lax equalizers
We prove a duality theorem for lax equalizers of stable $\infty$-categories, and we apply it to Frobenius and Cartier modules on a Noetherian $F$-finite scheme $X$ of characteristic $p$ with a unit dualizing complex. When $X$ has affine diagonal, comparison with the derived categories of coherent modules and passage to homotopy categories recovers Baudin's derived Hom duality. The duality of coherent modules exchanges the standard $t$-structure on one side with the perverse $t$-structure determined by the dualizing complex on the other. In particular, it identifies the abelian category of coherent Cartier modules with a perverse heart of coherent Frobenius modules, and conversely.
Read the full abstract
Computer Vision
TransGaze-Object: Transformer Based Driver Gaze Object Prediction Framework in Real Driving
Driver gaze provides information regarding driver visual attention and situational awareness to the surrounding traffic. Existing driver gaze estimation studies represent gaze in terms of gaze zone or gaze vector/point-of-gaze (PoG). However, object-level gaze information provides a more semantically meaningful representation of visual attention by identifying attended objects, such as vehicles, pedestrians, or traffic signals. In this study, we propose an end-to-end driver gaze object prediction framework, TransGaze-Object, Transformer-based Gaze Object prediction model. The proposed framework first extracts facial features, including face and iris-weighted eye features, along with trafficobject spatial features. A transformer based cross-attention mechanism is then used to compute similarity scores and attention weights for predicting the drivers gaze object. To train this model, we propose a benchmark driver gaze dataset, Urban Driving-Face Scene Gaze (UD-FSG), comprising synchronized driver-face and traffic-scene images, scene objects bounding boxes, and gaze labels in terms of 2D gaze coordinate and gaze object. The TransGaze-Object model achieves an overall accuracy of 60% for gaze-object prediction, compared to 51% accuracy obtained from associating the estimated Point-of-Gaze to traffic objects. The error analysis reveals that TransGaze-Object reduces confusion between traffic objects (predicted) and the background (ground-truth), achieving an error rate of 11.68%, a 49.7% relative reduction compared with 23.21% error obtained from PoG-based gaze-object association. Overall, the results demonstrate the effectiveness of directly predicting gaze objects from driver-face and traffic-scene information, rather than estimating an intermediate Point-of-Gaze and subsequently associating it with traffic objects.
Read the full abstract
Cosmology and Nongalactic Astrophysics
Observational constraints on a damped harmonic oscillator model of dark energy
We constrain a damped harmonic oscillator (DHO) dark-energy equation of state using the full cosmic microwave background (CMB) likelihoods in combination with DESI BAO and three distinct Type Ia supernova compilations: Pantheon+, DES-Dovekie, and Union3. The equation of state obeys a second-order damped oscillator equation in number of $e$-folds, so that its frequency $f$, damping rate $b$, and equilibrium value $w_{\rm m}$ fully specify the late-time dynamics. The model exhibits oscillatory behavior only at low redshifts, around the equilibrium value $w=-1$, with distinct characteristics for the different supernova compilations: an underdamped solution for DES-Dovekie and Union3, and an overdamped solution for Pantheon+. At higher redshifts, the model closely mimics $\Lambda$CDM and deviates significantly only at $z<0.6$, with the magnitude of the deviation depending on the supernova compilation. We further identify a region of the $(f,b)$ parameter space, corresponding to rapid variation of the equation of state at low redshift, in which the perturbation equations become numerically stiff and cannot be integrated with a canonical dark-energy sound speed i.e., $c_s^2 =1$. We show that reducing the rest-frame sound speed removes this obstruction while leaving the observables unchanged at the $10^{-3}$ level, and therefore treat it as a numerical prescription rather than a physical modification of the model. The model yields $H_0 = 67.53^{+1.22}_{-1.18}$ km/s/Mpc for Pantheon+, $H_0 = 69.08^{+1.23}_{-1.16}$ km/s/Mpc for DES-Dovekie, and $H_0 = 70.67^{+1.89}_{-1.87}$ km/s/Mpc for Union3. The present-day equation-of-state parameter is constrained to $w_0 = -0.521^{+0.891}_{-0.414}$, $-3.01^{+1.14}_{-1.21}$, and $-3.16^{+1.11}_{-1.16}$ for Pantheon+, DES-Dovekie, and Union3, respectively.
Read the full abstract
Robotics
Assembling Two Parts in One Hand
A hallmark of human dexterity is the cooperative use of fingers, where different fingers take on distinct yet coordinated roles to accomplish fine manipu- lation, such as capping a pen with the hand that holds it. We study this finger-level coordination through in-hand assembly: mating two rigid objects within a single dexterous hand, with no second arm and no fixture. We present a reinforcement learning formulation to solve this problem in a unified framework, which is driven by a goal relative pose between the two parts. Finger coordination is shaped by a function-based auxiliary reward and regularized toward a single human reference pose, while domain randomization and a fusion of historical proprioception and object observation confer robustness to occlusion-induced estimation noise. The same recipe solves three different assembly tasks (Bottle, Syringe, and Marker). Trained purely in simulation, the policies transfer zero-shot to hardware with a single camera, demonstrating robustness to state-estimation errors caused by oc- clusion. Our experiments also reveal that in-hand assembly places demands on hand morphology and can serve as a benchmark for modern robotic hand systems. Videos and code are available at https://ltbgbird.github.io/in-hand-assembly-page/.
Read the full abstract
Signal Processing
Low-Complexity Maximum Likelihood Detection for Type-Based Over-the-Air Computation
Type-based multiple access (TBMA) is a digital over-the-air computation (OAC) scheme that exploits symbol collisions over the wireless multiple-access channel to construct a histogram of the transmitted data at the receiver. This paper addresses the optimal detection of the transmitted histogram in TBMA over an additive white Gaussian noise (AWGN) channel. Since the number of possible received histograms grows combinatorially with both the number of transmitters and the data alphabet size, maximum-likelihood (ML) detection renders computationally infeasible. We prove that the ML detection problem can be solved by a greedy algorithm based on successive interference cancellation (SIC). We derive the analytical mean squared error (MSE) of the ML detector and show that it decreases exponentially with the SNR. Numerical results validate the theoretical analysis and demonstrate that TBMA achieves superior performance compared with conventional analog OAC, particularly for nonlinear functions.
Read the full abstract
Artificial Intelligence
Agent-Based ML-LLM Fusion with Self-Optimizing Prompts for Plateau Weather Alerts
To address insufficient contextualization, weak generalization, and poor scenario adaptation in tourism meteorological services, we propose SmartWeatherAgent–a unified three-stage architecture integrating intent recognition, hazard prediction, and reasoning-enhanced generation. The system fuses rule-based methods with large language models to parse queries at multiple granularities and employs a LightGBM model enriched with highland-specific features (e.g., wind speed abruptness rate), achieving an F1-Macro score of 0.605 with 1.60 ms latency on high-wind, precipitation, and low-temperature events. A 12-round micro-step prompt self-optimization loop boosts the composite warning quality score S_final from 4.2 (B01) to 8.9 (B12, +112%). Key improvements include a sharp rise in B08 from data source citation (6.5 -> 8.5), sustained high performance in B10 via physical mechanism explanation, and a peak scientific rigor score of 9.2 in B12 through explicit uncertainty statements. The system autonomously generates structured warnings that integrate causal mechanisms, spatiotemporal evolution, quantitative evidence, regulatory references, and confidence statements–enhancing professional depth, logical rigor, and scientific soundness, and advancing meteorological services toward proactive perception, explainable decision-making, and intelligent agency.
Read the full abstract
High Energy Physics (Theory)
Quantum Tunneling from Perturbation Theory Revisited
In the late 1990s, Suzuki and Yasuta proposed a compact formula that extracts the decay rate per unit volume of a false vacuum in the $D$-dimensional $O(N)$ $\lambda(\phi^2)^2$ theory with an unbounded potential from conventional perturbative coefficients of the vacuum energy density, i.e.\ vacuum bubble diagrams. The idea was to identify the imaginary part arising from the Borel integral along the discontinuity of the Borel transform with that of the vacuum energy density. While the formula works quite well for~$D=1$, i.e.\ quantum mechanics, its validity remained unclear for~$D\geq2$ because only the first three perturbative coefficients for~$D=2$ were available and the result showed no sign of convergence. In the present paper, we reexamine this approach using the first seven nontrivial perturbative coefficients (up to nine loops) for~$D=2$ and~$N=1$ obtained by Serone, Spada, and~Villadoro. Introducing two tunable parameters in the finite-order truncated Borel transform following these authors, we find that the imaginary part converges as the perturbative order increases, with the last few orders agreeing to within a few percent. In the intermediate range of the coupling constant, $3\lesssim\widetilde{g}\lesssim8$, this approach yields an imaginary part rather close to the leading-order semiclassical approximation with the one-loop determinant; it is $10$–$20\%$ larger than the semiclassical result with the two-loop correction computed by Malatesta, Parisi, and~Rizzo. Our result is consistent, within errors, with the continuum extrapolation of the Padé-resummed linked-cluster expansion in a quantum $\phi^4$ chain.
Read the full abstract
Cosmology and Nongalactic Astrophysics
Safe Phantom Divide Crossing from Unscreened Non-Minimal Coupling to Gravity
Recent hints of dynamical dark energy, including a possible crossing of the phantom divide, have motivated the search for viable theoretical explanations. Scalar fields non-minimally coupled to gravity with a coupling of the form $F(\varphi)=1+\alpha\varphi^2$ have emerged as promising candidates, providing an improved fit to the data compared to $\Lambda$CDM and even the $w_0w_a$CDM parametrization. However, in the absence of a screening mechanism, the regions of parameter space examined exhaustively so far (with $\alpha<0$) violate stringent local constraints on the strength and time variation of gravity and enhance the matter growth rate at low redshifts, thereby compromising the viability of these solutions. In this work, we investigate a largely unexplored region of parameter space (with $\alpha>0$) that realizes the crossing while remaining consistent with local constraints. This setup, which is subject to some fine-tuning, respects the BBN constraints and leads only to deviations of the effective gravitational coupling from Newton's constant below $0.3%$ in the past. We review the background and linear perturbation dynamics of the model, deriving analytical results that provide physical insight into its behavior, and update the constraints on the model using the latest cosmological data, including the DES-Dovekie supernova sample. We perform for the first time a full-fledged Monte Carlo analysis with $\alpha>0$, studying the impact of the variation of $G$ on the supernovae absolute luminosity, and compare the results with those obtained for $\alpha<0$, using in both cases a linear potential. In the context of the model with non-minimal coupling and $\alpha<0$, the standard model is excluded at the $2.50\sigma$ CL, whereas for $\alpha>0$, the exclusion significance decreases to $\sim 2\sigma$ CL, indicating a moderate preference for these models over $\Lambda$CDM. [abridged]
Read the full abstract
Systems and Control
Context operations to architecture modelling output from large language models and evaluation criteria for their use in systems engineering design
The development of generative artificial intelligence resources enables opportunities of speeding up systems and engineering design work. This contribution introduces a framework of formal operations for assembling context in LLM-based engineering design. This framework involves the assembly of modular context units, including policy prompts, reference units with persistence, and user questions with prompt vectoring. This approach enables the systematic structuring of interactions with generative models. A formal method for evaluating modelling-as-code LLM outputs is also presented, which enables the evaluation of compliance to intent from LLM answers and thereby asses the support from LLMs for systems architecture modelling.
Read the full abstract
Mesoscale and Nanoscale Physics
Reconfigurable field-free spin Hall nano-oscillators enabled by crystallographic anisotropy in epitaxial Co/Pt
Spin Hall nano-oscillators (SHNOs) are nanoscale microwave sources for wireless communication, neuromorphic computing and oscillator-based Ising machines, but conventional devices require a global magnetic bias. Here we replace this bias through crystallographic anisotropy in epitaxial Co/Pt. Growth of hcp Co with its c-axis in the film plane produces an anisotropy field of about 0.36 T and enables field-free auto-oscillations above 10 GHz in nanoconstriction SHNOs. The active current polarity is selected by the remanent magnetization, providing nonvolatile reconfiguration of the oscillation state. Micro-focused Brillouin light scattering confirms that the nonlinear response is confined to the nanoconstriction region. Lithographic control of the angle between the current and anisotropy axes tunes the excitation threshold and drives two spectral branches from separated modes to a dominant single branch, consistent with mutual synchronization. These results establish epitaxial crystallographic anisotropy as a route to reconfigurable field-free spintronic oscillators and oscillator networks.
Read the full abstract
High Energy Physics (Phenomenology)
Decoding Heavy Top-Philic Resonances at the HL-LHC: From Parton Kinematics to Deep Learning Signatures
The stabilization of the electroweak scale strongly motivates the search for new heavy resonances and top-partner states. We investigate the discovery potential of exotic resonances mediating the production of vector-like quarks at the High-Luminosity LHC ($\sqrt{s} = 13.6$ TeV), focusing on the $pp \to X \to t \bar{t}_p$ cascade decay. Using a model-independent Effective Field Theory approach, we classify the intermediate mediator $X$ by its spin (0 and 1) and color (singlet and octet) representations. Parton-level kinematics demonstrate that the normalized differential cross-section with respect to the transverse momentum provides a robust observable for spin discrimination. To address the experimental challenges of highly boosted, semi-resolved hadronic decays, we introduce a cut-flow strategy based on an inclusive four-jet invariant mass reconstruction, $M(4j)$. Because this standard approach remains sensitive to systematic uncertainties, we implement a Generalized BSM Tagger based on Deep Neural Networks (DNNs). By exploiting non-linear multi-jet correlations, the DNN achieves robust background rejection factors ranging from $\mathcal{O}(600)$ to $\mathcal{O}(1000)$ while preserving the underlying partonic signatures. Projecting to an integrated luminosity of $3000\text{ fb}^{-1}$, this combined strategy yields a projected statistical significance of $3.74\sigma$ for the dominant color-octet vector channel, establishing strong evidence potential and offering a robust phenomenological baseline for future HL-LHC searches.
Read the full abstract
Analysis of PDEs
Controlled harmonic map heat flow and blow-up from the disk to the sphere
We study the one-corotational harmonic map heat flow from the disk to the sphere with a Dirichlet boundary control. We introduce a control-adapted gluing strategy to prevent blow-up. More precisely, we construct explicit upper profiles and well-prepared boundary traces such that every nonnegative datum below one of these profiles and having boundary value above $\pi$ blows up under the corresponding constant trace, whereas the corresponding solution is global under every control below the associated well-prepared boundary trace. Controls with uniformly bounded time derivative also prevent infinite-time concentration. We also show that boundary control can drive any nonnegative datum to blow-up before any prescribed time, whereas bounded controls cannot prevent suitably fast concentration.
Read the full abstract
High Energy Astrophysical Phenomena
Hadronic origin of gamma rays and neutrinos from blazars: Multi-messenger implications and observational constraints
We investigate whether cosmogenic gamma rays and neutrinos from ultra-high-energy cosmic rays (UHECRs) produced in hadronically active blazars can significantly contribute to the observed spectral energy distribution (SED), and examine the potential of the Cherenkov Telescope Array Observatory (CTAO) to detect and discriminate this component. For four nearest neutrino bright blazars modeled by Rodrigues (2024), specifically AP Librae, TXS 1700+685, PKS 2326$-$502, and PKS 2345$-$16, we constrain the UHECR luminosity by requiring that the total emission does not overshoot available multi-wavelength data, assuming acceleration in the optically thin large scale jet, propagate UHECRs through realistic Galactic and extragalactic magnetic fields, and forecast CTAO performance. We find that cosmogenic gamma rays contribute up to about 7.5% of the total gamma-ray emission at the highest energies ($10^{20}$ eV), and thus remain subdominant in the SED. In contrast, the associated neutrino emission, at approximately $10^{44}$ to $10^{46}$ erg s$^{-1}$, offers a robust propagation signature of hadronic acceleration. Our simulations of 50 hour CTAO exposures show that the observatory can accurately reconstruct the intrinsic gamma ray spectrum, enabling strong constraints on the hadronic origin and propagation of UHECRs. These results demonstrate the power of combining CTAO gamma-ray measurements with neutrino observations to probe blazar hadronic processes.
Read the full abstract
Spectral Theory
Resolvent bounds and eigenvalue estimates of generalized Schrödinger operators with complex potentials on compact manifolds
We extend Cuenin's compact-manifold spectral bounds for Schrödinger operators with complex potentials to a general pseudodifferential setting. More precisely, we study operators \(P+V\), where \(P\) is a positive self-adjoint elliptic classical pseudodifferential operator of positive order and \(V\) is complex-valued. The main analytic input is a resolvent principle showing that spectral cluster estimates for \(P\) imply \(L^p\)-\(L^{p'}\) resolvent estimates along suitable complex curves. Combined with Sogge's spectral cluster bounds, this yields exterior-region resolvent estimates extending those of Krupchyk and Uhlmann; we also prove direct resolvent bounds in the interior region. On Zoll manifolds, we discuss the sharpness of the resulting spectral bounds.
Read the full abstract
Hardware Architecture
SAGE: Semantic-Aware Geographic Error Recovery for AI Data Movement
AI interconnects typically protect and replay packets uniformly, yet numerical bit faults differ sharply in consequence: a low-order mantissa flip may resemble quantization noise, while a high-significance exponent flip can produce a catastrophic outlier or non-finite value. We present SAGE, a semantic-aware geographic error-recovery architecture that decouples whether a detected fault merits replay from where replay restarts. For BF16-like data, a workload-calibrated contract separates catastrophic Class-H faults from bounded Class-M and precision Class-L damage. It first applies a Class-H silent-delivery constraint, then ranks admissible policies by quality-normalized terminal latency, $\Psi_{\rm del}$. Independently, a source-local region table adapts checkpoint intervals to fault geography, shortening recovery segments in noisy regions. Detected Class-H failures may trigger protected negative acknowledgments and full-flit replay; Class-M and Class-L outcomes do not trigger default network replay. We implement SAGE's endpoint and replay protocol in gem5 Garnet and synthesize its fully pipelined checker in ASAP7. At a stable synthetic operating point, a ten-seed contention-faithful direct-Garnet campaign shows that SAGE reduces $\Psi_{\rm del}$ by 30.1% relative to fixed 34-hop recovery, combining 28.0% lower mean latency with improved delivered semantic quality. Under higher-BER synthetic stress at the same offered load, SAGE maintains bounded queues while the fixed baseline accumulates backlog. Application-derived DeiT-S communication traces also show lower mean and p99.5 latency at the evaluated nonzero BERs. Within the qualified operating envelope, CRC32 decoder trials yield a simultaneous 95% per-original Class-H silent-delivery upper bound of $3.18\times10^{-7}$.
Read the full abstract
Computer Vision
SA-Profile: Automated Sulcus Angle Profiling from Super-Resolution MRI
Trochlear dysplasia (TD) is an abnormality of the femoral trochlea associated with anterior knee pain and patellar instability. The sulcus angle (SA) is used to assess trochlear morphology, but it is typically measured on a single axial MR slice with no clear guidance on which to select, making it sensitive to slice selection and landmark placement. We propose an automatic framework for continuous SA profiling from super-resolved MR volumes. Clinically acquired axial, coronal, and sagittal MR scans are combined using implicit neural representations to reconstruct a high-resolution volume. SA measurements are computed across the trochlear region using two landmark detection U-Net models. The approach was evaluated on the public fastMRI dataset and a small in-house cohort of patients with TD. Compared with conventional manual single-slice SA measurements, the proposed automated method yielded a mean absolute error of 11.6$^\circ$ while providing continuous characterization of trochlear morphology. Population-level analysis demonstrated distinct mean SA profiles between the public cohort and the in-house TD cohort, highlighting the potential of profile-based assessment to characterize TD. By reducing reliance on a single manually selected axial slice, the proposed framework extends conventional SA assessment to a continuous profile-based description of trochlear morphology without additional imaging, while remaining conceptually linked to current clinical assessment. Further validation is required. The code is available: https://github.com/wehrlimi/SA_Profile.