Computation and Language
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Monday 31 August
Trajectory data is getting more vital for training large language models for boosting the agentic abilities. Unlike the verifiable domains such as coding or mathematics, scaling trajectory data for open-ended tasks is much more difficult because these tasks lack singular ground truth and are costly to annotate or verify. In this paper, we propose RetroGen, a self-improving framework of retrospective process supervision. Our key observation is that although expert trajectories are scarce, high-quality final artifacts such as literature reviews, analyst reports and legal judgments, are abundant in pre-training data and can be viewed as compressed traces of the evidence-seeking processes that produced them. RetroGen reconstructs candidate latent trajectories from expert artifacts, verifies them against both the artifact and supporting evidence, and trains models on their own successful reconstruction data, without requiring trajectory data from stronger models. Experiments show that RetroGen improves grounding, faithful synthesis, and long-form evidence-seeking agent tasks.
Optics
Ghost infrared spectroscopy with bright twin beams
Frequency-correlated light offers a route to mid-infrared (MIR) spectroscopy without direct spectral detection in the MIR. Previous MIR ghost spectroscopy has mainly relied on low-gain spontaneous parametric down-conversion (SPDC) and photon-pair coincidence measurements, where the limited photon flux has restricted acquisition times to longer than one minute. Here, we demonstrate ghost infrared spectroscopy using bright twin beams generated by high-gain parametric down-conversion (PDC). High-gain PDC amplifies vacuum fluctuations, producing a different pair of frequency-correlated random spectra in each pump pulse. This pulse-resolved stochastic emission is naturally matched to time-stretch detection, which records the spectrum of the correlated near-infrared telecom signal for every pulse, while the MIR idler is measured by bucket detection. Consequently, each pump pulse yields one paired projection measurement comprising a spectrally resolved reference and the corresponding bucket value. We reconstruct the transmission spectrum of a structured optical filter and the molecular vibrational absorption spectrum of liquid benzene near 3.3 um with millisecond-scale acquisition, in good agreement with Fourier-transform infrared spectroscopy. This reduces the acquisition time by four to five orders of magnitude compared with previous MIR ghost spectroscopy demonstrations. These results transform ghost spectroscopy from coincidence-based photon counting to high-flux analog correlation spectroscopy, establishing a practical architecture for high-speed computational infrared spectroscopy driven by a narrowband semiconductor laser.
Read the full abstract
Astrophysics of Galaxies
Broad-band host measurements of little red dots and similar compact nuclei require source-conditioned polychromatic PSFs
Characterizing compact sources in the early Universe is difficult because light from physically distinct components can overlap within the same diffraction-limited image. A common method uses field stars to model how a single point of light appears and from this estimates the intrinsic structure and contributions of the physical components of the distant object under study. Many important properties inferred for high-redshift objects depend on the accuracy of this method. Using the little red dots (LRDs) analysed by Zhang et al. as a test case, this work shows that applying the method through broad filters to objects whose spectra differ from those of the field stars can systematically misassign light, overestimating an existing extended component or creating an apparent one where none exists. At the measured level, this error is large enough to alter the inferred sizes, luminosities and masses, and therefore the physical interpretation of these systems. The physical and mathematical origin of the error is established and a correction applicable to other compact distant sources is demonstrated.
Read the full abstract
Computer Vision
Federated Multi-Task Learning for Bladder Tumor Segmentation and MIBC Classification Using a Hybrid CNN-Transformer Architecture
Accurate bladder tumor segmentation and assessment of mus- cle invasion from T2-weighted MRI are important for treatment plan- ning, but developing robust models across institutions is challenging be- cause patient data cannot be centrally pooled and imaging characteristics vary across scanners and acquisition protocols. We propose a federated multi-task learning framework for joint bladder tumor segmentation and MIBC/NMIBC classification across four clinical centers. The proposed Swin Hybrid model combines a ResNet-34 branch for local texture and boundary information with a Swin-Tiny Transformer for global anatomi- cal context. A segmentation-guided classification mechanism further uses tumor localization information to support MIBC prediction. We also investigate several augmentation strategies under both centralized and federated training to improve robustness to multi-center variability. Ex- periments on the FedBCa dataset show that the Swin Hybrid provides the best overall balance between segmentation and classification among the evaluated architectures. Under federated training, Geo+Elastic aug- mentation achieved a DSC of 0.8100 and a patient-level AUC of 0.8931, yielding the highest combined score of 0.8474. These results demonstrate that joint segmentation and classification can be effectively performed across multiple institutions using federated training without centralizing patient data.
Read the full abstract
Machine Learning
Learning Where Outcomes Change:Credit-Addressable Reasoning for Multimodal Geometry
Multimodal geometry reasoning requires VLMs to extract precise visual relations and preserve them through multi-step deduction. Existing free-form traces obscure the decisions that determine the answer, and trajectory-level reinforcement learning distributes a single terminal signal across the entire response. We introduce credit-addressable reasoning, in which the semantic units exposed during inference also define where learning compares alternatives and assigns credit. We instantiate this principle with Code-CoT, which retains the diagram, represents visual relations as line-addressable executable code, and organizes reasoning into typed events, and CE-GRPO, which selects event boundaries using structural priors and type-normalized entropy, samples complete continuations from shared prefixes, and converts outcome differences into localized advantages. Across nine geometry benchmarks, CE-GRPO achieves an average accuracy of 76.04, outperforming Qwen3-VL-8B and trajectory-level GRPO by $8.09$ and 3.43 points, respectively. Its relative advantage increases with the number of intermediate events, demonstrating the value of representation–optimization co-design for long, dependency-heavy multimodal reasoning.
Read the full abstract
Machine Learning
Self-Supervised Pretext Tasks for Infant Cry Analysis: A Controlled Comparison and a Cautionary Result on Donateacry
We compare six self-supervised pretext tasks for infant cry analysis under a fixed budget, meaning the same compact encoder of 1.17M parameters, the same 115 hours of license-verified public pretraining audio, and the same evaluation protocol for every candidate. On cry detection the reconstructive objectives dominate, and a linear probe over a masked-spectrogram encoder reaches 0.988 AUC with subject-wise splits even though the encoder never observed a cry during pretraining. On cry-reason classification over donateacry, the de facto public benchmark for cry reasons, every encoder performs at chance (0.38 to 0.54 macro AUC over 5 classes), and neither domain adaptation on 1.8 hours of real cries nor end-to-end fine-tuning moves the result. Since a frozen HuBERT-base with 80 times more parameters shows the same pattern, the bottleneck must sit in the labels and not in model capacity. We then reproduce the 90%+ accuracies of the donateacry literature on our own system by changing nothing but the evaluation protocol: clip-wise splits raise accuracy to 85.2% (barely above the 83.8% majority-class baseline), and applying augmentation before splitting raises it to 97.9%, matching the reported state of the art, from the same model that measures 0.49 macro AUC under subject-wise splits. Under leakage-free splits, a twentyfold augmentation of the labeled set (vocoder speaker perturbation and noise mixing, 21 hours) leaves cross-subject AUC unchanged: for this task the effective sample size is the number of infants. We release code, seeds and per-clip license manifests.
Read the full abstract
Accelerator Physics
Magnetizing nonlinear plasma wakefields for positron acceleration
It is known that only a narrow plasma wakefield sliver in the electron-beam-driven blowout regime suits positron acceleration. Using 3D simulations, we show that matching the cyclotron frequency $\omega_{c}$ with the plasma frequency $\omega_{p}$ forms a stable electron column on axis, expanding the suitable phase space for positron acceleration sizably. For a plasma density $n_p = 10^{16}\text{ cm}^{-3}$ in a 35 T field, the interval expands 4.3 times, and a witness positron beam gains 100–150 MeV over 6 cm ($1.6$–$2.5$~GeV/m) with a 92% capture rate.
Read the full abstract
Materials Science
Coupling of two individual magnon resonators via a superconducting microwave resonator operating in the strong coupling regime
We demonstrate coherent coupling of ferromagnetic resonance in two individual permalloy stripes via a superconducting coplanar waveguide resonator. The two stripes are placed on top of the resonator with an angle of 50$^{\circ}$ between their respective long axes. This alignment enables us to separately tune their ferromagnetic resonance frequency by exploiting their shape anisotropy by simply rotating the external bias magnetic field in the sample plane. Both individual ferromagnets exhibit strong coupling with the microwave resonator. This method avoids the use of local individual bias coils and facilitates the integration of similar experiments into planar device infrastructures.
Read the full abstract
Information Theory
Marker-Delimited Codes for Short-Blocklength, High-Rate Coding over Multi-Read Edit Channels
The read process of DNA-based data storage systems generates multiple noisy copies of the stored DNA sequences, affected by edit errors consisting of substitutions, deletions, and insertions. Motivated by the challenge of ensuring reliable data retrieval in the presence of edit errors, we present a concatenated coding scheme that accounts for practical design constraints in DNA storage. We introduce and apply the marker-delimited code (MDC) as the inner code, which enables fast and reliable computation of symbolwise a posteriori probabilities (APPs). We combine MDC with an outer LDPC code. The LDPC is decoded via belief propagation using the soft information generated by MDC. Our results show that, in comparison with prior work, this construction provides more efficient error correction over multi-read edit channels in the short-blocklength and high-rate regime.
Read the full abstract
Fluid Dynamics
Drag reduction and subcritical turbulence in controlled pipe flows
Pipe flow controlled by streamwise-travelling waves of azimuthal wall velocity is studied using direct numerical simulations at a bulk Reynolds number $Re_b=4900$. A comprehensive analysis of drag reduction shows that the flow response differs fundamentally from that of channel flow. Under suitable forcing, pipe flow relaminarizes, whereas channel flow does not. Depending on the control parameters, the flow exhibits the spatially localized turbulent state characteristic of transitional pipe flow, with turbulent puffs persisting at bulk Reynolds numbers up to three times higher than in the uncontrolled case. The bulk Reynolds number alone does not determine the onset of localization. Moreover, drag reduction, which alters the natural relation between bulk and friction velocities, is insufficient to identify a universal onset criterion. An intermittency indicator based on the spatial variance of the cross-sectional turbulent kinetic energy relates the emergence of localized turbulence to the low wall friction produced by the control, although the correspondence is not one-to-one. Despite their qualitative resemblance to canonical turbulent puffs, the controlled puffs exhibit distinct properties; for example, their fronts may propagate faster than the bulk flow. Overall, this work provides a comprehensive characterization of the subcritical turbulent state and its turbulent puffs in controlled pipe flow, and offers a new perspective on control strategies that aim at flow relaminarization.
Read the full abstract
Computer Vision
SeqAlign3DVG: A Sequence-Aligned Benchmark and Voxel Reasoning Framework for 3D Visual Grounding
Image-based 3D visual grounding is critical for embodied agents, yet existing benchmarks suffer from loose text-observation alignment and neglect temporal ordering. We introduce SeqAlign3DVG, a novel benchmark dedicated to temporally ordered and strictly observation-aligned image-based 3D visual grounding. Unlike prior works using order-agnostic views or global point clouds, SeqAlign3DVG ensures all expressions are human-verified and strictly grounded in the provided RGB observations (single frames or ordered observation sequences). It comprises 9,622 single-view and 14,493 sequence samples featuring rich descriptions, complex relations, and multi-instance ambiguities. To tackle this benchmark, we propose a unified voxel-based pipeline featuring Relevance-Ordered Voxel Memory (ROVM) and Progressive Language-Voxel Fusion (PLVF). ROVM dynamically ranks and aggregates multi-view evidence via a conservative memory to mitigate noisy observations, while PLVF performs coarse-to-fine spatial-linguistic reasoning for precise disambiguation. Our approach achieves state-of-the-art performance under the depth-free protocol, significantly improving localization for targets defined by complex relations and appearance cues.
Read the full abstract
Computer Vision
FlowVVTON: Flow-Guided Mask-Free Video Virtual Try-On
Video virtual try-on aims to transfer a target garment onto a moving person across video frames. Current methods rely on human parsing masks or pose keypoints that frequently fail under large motions and occlusions, causing boundary artifacts and temporal inconsistency. A further limitation is that most approaches rely solely on attention mechanisms for temporal modeling, providing no explicit motion supervision. We propose FlowVVTON, a mask-free framework that eliminates parsing mask dependency entirely. Optical flow is used solely as a training-time supervision signal: a flow-warped latent loss, applied across all layers of the generation model, enforces multi-scale temporal consistency by aligning adjacent-frame features under explicit physical motion constraints. A two-stage training strategy establishes mask-free spatial alignment before introducing flow-guided temporal supervision. Experiments on TikTokDress show that FlowVVTON outperforms baselines by substantial margins, particularly in temporal consistency (5.7$\times$ VFID-R improvement over SwiftTry), while requiring no segmentation masks, pose keypoints, or region annotations at any stage.
Read the full abstract
Machine Learning
PRIME: Mitigating Subgroup Optimization Competition in Shared CTR Top Networks with Plug-in Residual Input-Conditioned Mixture of Expert
Click-through rate (CTR) models vary in feature-interaction design, yet their top networks usually remain a single multilayer perceptron shared by all examples. Heterogeneous user, item, and context subgroups therefore update the same parameters; weakly aligned learning signals make the aggregate gradient a compromise among competing directions. We study the competition on Avazu with 4 models and 4 semantic fields. Across all architectures, semantic subgroups show lower Top-NN gradient cosine similarity than random groups matched by sample size and label ratio, with reductions of 0.23-0.37. This competition motivates input-conditioned experts, but directly replacing an established Dense mapping changes its initial function, sharing pattern, and capacity, obscuring the source of gains. We introduce PRIME (Plug-in Residual Input-conditioned Mixture of Experts), a Dense-anchored mixture of low-rank residual experts. PRIME anchors the original prediction and uses zero-residual initialization to match the Dense baseline exactly at training onset. Input-dependent routing weights low-rank experts for example-specific logit corrections; multi-bag aggregation and EMA load biases stabilize conditional estimation. We evaluate PRIME on held-out Avazu and Criteo test sets across 13 CTR architectures and five paired seeds. Median paired AUC gains are +0.0022 and +0.0066, with LogLoss reductions of 0.0011 and 0.0081, respectively. On FiBiNET and DCNv2, PRIME outperforms APG in all ten seed-level AUC comparisons while using fewer parameters and lower inference latency on both backbones. These results show that function-preserving conditional residuals add input-dependent capacity while preserving the Dense path and its optimization stability. Code is available at https://github.com/YH-learning/PRIME.
Read the full abstract
Differential Geometry
Structure of the Riemann tensor in higher-dimensional Kerr-NUT-(A)dS spaces
We study the algebraic and differential structure of the Riemann curvature tensor of the higher-dimensional Kerr-NUT-(A)dS spaces, motivated by the eventual goal of finding an IDEAL (Intrinsic, Deductive, Explicit and ALgorithmic) characterization of this family of metrics. The special geometry of these spaces is governed by the principal tensor $h_{ab}$, a non-degenerate closed conformal Killing-Yano $2$-form, whose existence singles out the Kerr-NUT-(A)dS family; we therefore focus on the algebraic relationship between the Riemann curvature $R_{abcd}$ and $h_{ab}$. In our investigations we encountered an obstruction, which results in a no-go theorem: no non-trivial $2$-form, including $h_{ab}$ itself, can be constructed covariantly from the undifferentiated Riemann tensor alone. Despite that, we do characterize the family of Riemann-symmetric tensors that could be the curvature of a Kerr-NUT-(A)dS metric by an algebraic and a differential condition, stemming from the integrability of the defining equation of $h_{ab}$ and the second Bianchi identity. Our calculations apply in all higher dimensions, which becomes feasible by a judicious application of representation theoretic techniques related to a semi-direct product group $U(1)^n\rtimes S_n$ that stabilizes $h_{ab}$.
Read the full abstract
High Energy Physics (Theory)
Examples of Gribov copies in the presence of planar boundaries, and an emergent boundary gauge invariance in Yang-Mills theory
First, we explicitly construct zero-modes of the Yang-Mills Faddeev-Popov operator in the presence of two parallel plates, with either perfect electric (PEC) or perfect magnetic (PMC) boundary conditions imposed. This establishes the existence of Gribov copies with finite action, and even finite L_2-norm, in Yang-Mills theory with non-trivial interfaces. We adapt Henyey's construction, although the boundary configuration imposes extra restrictions on the ansatz for the gauge field. Second, we discuss the phenomenon of an emergent boundary gauge invariance for PEC and PMC plates in Yang-Mills theory. This shall be important when applying the Gribov-Zwanziger procedure to parallel plate setups, of relevance for non-perturbatively studying the non-Abelian Casimir effect in the future.
Read the full abstract
Portfolio Management
End-to-End Neural Shrinkage of Indefinite Pairwise Correlation Matrices for Small-Cap-Inclusive Portfolios
Small-cap-inclusive equity universes contain recently listed and intermittently traded securities, so enforcing a common look-back discards a substantial fraction of the available information. Pairwise-complete estimation preserves the longest overlap for each asset pair, but the resulting correlation matrix can be indefinite because its entries are computed on different samples. This prevents direct use in Markowitz optimization and falls outside the assumptions of standard random-matrix shrinkage. We adapt a rotation-invariant neural covariance estimator to this setting. The model computes mask-aware marginal moments and a pairwise correlation matrix proxy, processes its signed spectrum, and uses a bidirectional gated recurrent unit conditioned on factor-aligned effective sample lengths derived from the overlap matrix and eigenvector loadings. It maps all eigenvalues, including negative ones, to a positive inverse spectrum. The reconstructed covariance is positive definite and is trained end-to-end to minimize five-day realized global-minimum-variance risk. We evaluate 26 expanding-window models from 2000 to 2025 on up to 1,500 U.S. equities in a closing-auction simulator with point-in-time selection, commissions, financing, corporate actions, and market impact. Across the 26-year out-of-sample period, the neural estimator reduces annualized five-day volatility by approximately 20% and increases the Sharpe ratio by approximately 40% relative to the next-best covariance estimator. These improvements are consistent across realized risk, risk-adjusted performance, and drawdown control, remain after the modeled execution frictions, and are supported by a 99.9% Model Confidence Set that retains only the neural estimator.
Read the full abstract
Analysis of PDEs
Derivation of a linearly elastic elliptic membrane shell theory in the framework of magnetoelasticity
Starting from a three-dimensional model of linearized magnetoelasticity, we investigate the asymptotic behavior of a thin shell as its thickness approaches zero. Our analysis focuses on a family of linearly elastic elliptic membrane shells. Using $\Gamma$-convergence, we rigorously derive a two-dimensional linear magnetoelastic shell model and establish its relationship to the original three-dimensional model in terms of the convergence of minimizers.
Read the full abstract
Hardware Architecture
Non-uniform Memory Partitioning For Low-Power Spiking Neural Networks
Spiking Neural Networks (SNNs) naturally excel in processing temporally rich and sparse data. However, because of their time-stepped processing, memory access, specifically to synaptic weights stored in SRAM (static random-access memory), tends to dominate total power consumption. To address this issue, without incurring a large area overhead, we propose to leverage the greatly varying average firing rate of neurons in the network to efficiently allocate synaptic weights to an on-chip memory consisting of multiple non-uniformly sized memory banks. By assigning weights of frequently firing neurons to shallow, low-access cost memory and less actively accessed weights to deeper, high-density memories, the average power consumption of the synaptic weight memory is decreased without incurring a large area overhead. To benchmark our proposed architecture and find optimal configurations of memory arrangements, we perform an automatic exploration based on application requirements and hardware constraints. For memory designs synthesized in 28-nm CMOS technology, we show that our architecture can achieve a synaptic weight memory access power reduction of up to 61% compared to a conventional design, with a 2.1$\times$ lower area overhead, as compared to a traditional uniformly partitioned memory bank that achieves a comparable reduction.
Read the full abstract
Analysis of PDEs
On smooth self similar solution to compressible Euler equation with physical vacuum
We consider the barotropic Euler equations in dimension $d > 1$ with decaying density at spatial infinity. The phase portrait of the nonlinear ODE governing the equation for spherically symmetric self-similar solutions has been introduced in the pioneering work of Guderley. The existence of smooth decaying solutions through the sonic line has been obtained in the work (F. Merle et al., 2022) in a suitable range of parameters. We work in this paper in a different range of parameters and show the existence of holosphere like solutions which are smooth through the sonic line, but display vanishing density at a sphere.
Read the full abstract
Machine Learning
Confounding Masquerading as Improvement: A Systematic Evaluation of Offline Reinforcement Learning for Stroke Antithrombotic Treatment in a 129,000-Patient Registry
Recent offline reinforcement learning (RL) studies report policies that outperform physician decisions on clinical outcomes. We conduct a systematic, partially crossed evaluation of five offline RL algorithm families and 14 reward designs in 44,894 post-2018 acute ischemic stroke patients from a nationwide registry (N = 129,033). Standard Fitted Q-Evaluation (FQE) yields an apparent policy-improvement estimate of +0.0069; adding an Early Neurological Deterioration penalty increases it to +0.0101. We identify reward-embedded confounding, in which a proxy terminal reward encodes baseline severity and prognosis as well as treatment efficacy. A 2 x 2 factorial analysis finds that terminal reward confounding accounts for 218.6% of the observed signal change, so its removal overshoots the null. After DML-inspired GBM reward residualization, the FQE estimate attenuates to +0.0033 (p = 0.132), and full deconfounding yields +0.0025 (p = 0.291). FQE-based diagnostics, T-learner analyses, and direct recurrence analyses converge away from a clinically meaningful aggregate improvement. A 1-year mRS factorial analysis replicates the attenuation. We provide an empirically motivated six-step evaluation checklist. NIHSS-stratified heterogeneity is hypothesis-generating for prospective trial design; hospital-level disagreement does not persist after full reward deconfounding.
Read the full abstract
Cryptography and Security
ECLIPSE: Self-Evolving Stealthy Prompt Injection Attack against Long-Horizon Agentic Systems
Recently, large language model (LLM) agents, such as Codex, Claude Code, and OpenClaw, have become capable of planning and executing long-horizon tasks through repeated tool calls. This capability also creates new opportunities for prompt injection. Existing attacks either place the malicious objective in one explicit instruction, making it easy to detect, or distribute the intent across multiple execution stages, making successful completion unreliable. In this work, we propose ECLIPSE, a self-evolving and stealthy prompt-injection framework for long-horizon agentic systems. ECLIPSE combines direct user-prompt injection with indirect tool-side injection through two components. On the one hand, Stealthy Attack Trajectory Synthesis uses a sandbox to generate and iteratively verify candidate tool chains, then renders a verified chain as a natural one-shot prompt to serve as the direct instruction. Then, Tool-Chain Steering transfers this plan to the target environment through Static Workflow Encoding (SWE), which embeds state-transition cues in target-tool descriptions, and Dynamic Trajectory Correction (DTC), which supplies corrective signals when execution deviates from the planned chain. To enable systematic evaluation, we further introduce LASE-Bench, a long-horizon agent-safety benchmark with 120 malicious tasks and 198 unique tools; 96.7% of its tasks make at least five tool calls. The experimental results show that ECLIPSE is highly effective: it achieves up to 96.7% attack success without defense and 69.2% under the common safety filter, exceeding the strongest baseline by 27.5% in the defended setting. Evaluations against representative defenses further show that existing safeguards do not reliably defend it, which raises the need for more effective defenses.
Read the full abstract
Algebraic Geometry
Abundance of affine algebraic structures on stabilized cotangent bundles of surfaces
We show that for any $g\geq2$, there exist uncountably many non-isomorphic smooth affine threefolds that are Stein deformation equivalent to $T^\ast\Sigma_g\times\mathbb{C}$, the product of the cotangent bundle of a genus $g$ oriented surface with the complex plane, answering a question of Ivan Smith. In fact, we prove that our family of smooth affine threefolds is parametrized by a $(6g-7)$-dimensional complex orbifold.
Read the full abstract
Neural and Evolutionary Computing
Event-Driven Language Models with Sparse Neural Activity for Neuromorphic Hardware
Inference with transformer-based large language models (LLMs) is often limited by the memory-bound KV cache and quadratic attention cost. State-space models (SSMs) mitigate this through linear attention and fixed-size recurrent states, but their large dense linear projections remain computationally expensive even after quantization. We introduce a method that induces sparse neural activity in heavily quantized linear-attention models with minimal performance loss. Activations below a per-projection trainable threshold ($\pm \Delta$) are nullified while preserving crucial outliers, achieving comparable performance to dense models with up to 4$\times$ fewer effective arithmetic operations. Targeting a multi-core, multi-chip neuromorphic platform, where event-driven execution converts unstructured sparsity into throughput at both the compute and communication levels, a capability GPU architectures fundamentally lack, we project up to 37$\times$ higher throughput and 16$\times$ lower power versus edge GPU inference of a comparable transformer-based model, and up to 5.4$\times$ improvements over the non-sparsified baseline. These results position sparse, quantized linear-attention models as a natural fit for deploying LLMs on event-driven multi-core platforms.
Read the full abstract
Mesoscale and Nanoscale Physics
Evanescent-mode Casimir-Josephson force and gate-controlled resonances in ballistic graphene Josephson junctions
We develop a microscopic scattering theory of the phase-dependent equilibrium mechanical response of a ballistic superconductor-graphene-superconductor Josephson junction. The calculation is based on an energy-dependent Dirac scattering matrix embedded in a normalized Matsubara determinant, so the reported force correction contains the complete Bogoliubov-de Gennes spectrum of the stated ideal model. At charge neutrality the full determinant approaches a closed-form evanescent-mode result proportional to $\Delta W/L^2$, reaching $2\ln 2\,\Delta W/(\pi L^2)$ at phase difference $\pi$ in the zero-temperature short-junction limit. Gate doping produces propagating channels and Fabry-Pérot structure, causing large oscillations and sign reversals of the phase-dependent force correction $\delta F=F(\phi)-F(0)$. We distinguish the interband-to-intraband Andreev crossover, controlled by $|\mu|/\Delta$, from the evanescent-to-propagating crossover, controlled by $|\mu|L/(\hbar v_F)$. Exact real-energy subgap poles obtained from the same energy-dependent scattering matrix are used to diagnose specular/interband and retro/intraband character; these labels are not treated as separately measurable thermodynamic forces in the mixed regime. We also quantify the difference between the complete determinant result and the frozen-scattering short-junction approximation without identifying that difference with a pure continuum force. The resulting gate- and phase-dependent mechanical signal provides a Dirac-material extension of earlier superconductivity-induced mechanical-force proposals.
Read the full abstract
Computation and Language
Graph Evidence Is Not Enough: Diagnosing Native Decoder Use in Graph-Augmented LLMs
Graph-augmented large language models often assume that graph evidence produced by external computation and placed in the input can be used by the native decoder. We test this assumption with HopQA, a deliberately bounded diagnostic that asks for the shortest-hop distance between two query nodes. Because the answer is a small integer and the target is purely topological, failure cannot be dismissed as open-ended generation or ambiguous evaluation. Yet existing graph-augmented baselines still fail on this setting, showing that providing graph evidence is not the same as making it usable. We introduce an intervention triangle with three matched conditions: readable graph evidence, shuffled graph evidence, and no-graph input. This separates evidence inclusion, structural readability, and decoder-usable topology. Guided by this diagnosis, we present S$^2$GE as an instance showing that diagnosis-driven interface design can improve native decoder usability. S$^2$GE uses query-aware sampling, endpoint and proximity-based ordering, and structure-preserving alignment. Across DBLP, Biomedical, GoodReads, and PubMed, S$^2$GE achieves strict exact-match scores of $36.5\%$, $57.8\%$, $76.6\%$, and $52.0\%$, improving over the strongest native-generation baseline by $53.5$ points on average. The interventions further reveal harmful-shuffle, shuffle-robust, and no-graph-saturated regimes.
Read the full abstract
High Energy Physics (Phenomenology)
Two-loop QCD corrections to $CP$-even Higgs boson decays into $V\gamma$ in the Type-I Two-Higgs-Doublet Model
We investigate the QCD corrections to the loop-induced decays of the heavy $CP$-even Higgs boson, $H \to V\gamma$ ($V = Z, \gamma$), within the framework of the Type-I Two-Higgs-Doublet Model (THDM). Owing to the similarity of the underlying loop topologies, the two decay channels exhibit closely correlated behaviour. After imposing relevant phenomenological constraints, we evaluate the decay widths for 5000 viable parameter points at next-to-leading order (NLO) in QCD under the alignment limit. The $H\to\gamma\gamma$ width exceeds that of $H\to Z\gamma$ by roughly a factor of four. Both are dominated by the top-quark loop: the $W$ contribution vanishes in the alignment limit, and the charged-Higgs loop is suppressed throughout most of the viable parameter region. The corrections exhibit a pronounced dependence on the heavy Higgs boson mass $m_H$, attaining a maximum in the vicinity of the $t\bar{t}$ production threshold. To elucidate the individual parameter dependences, dedicated benchmark scenarios are analysed, revealing that the decay widths diminish monotonically with increasing $\tan\beta$ while growing with $m_H$. The QCD corrections for both decay modes typically span approximately $-20\%$ to $+20\%$, with larger corrections in the non-alignment scenario: in the $H\to\gamma\gamma$ channel they reach up to $+38\%$ at large $\tan\beta$, and in the $H\to Z\gamma$ channel they cross zero to become positive. The $CP$ selection rule forces the $t\bar{t}$ pair into a $P$-wave configuration at threshold, which strongly suppresses the Coulomb corrections and secures the perturbative stability of the fixed-order prediction. A comparison with ATLAS limits shows that current data do not yet constrain the parameter space, whereas the $\mathcal{O}(15\%)$ NLO corrections may become relevant for interpreting exclusion bounds at the HL-LHC.
Read the full abstract
Hardware Architecture
Scalable AXI4 Transaction Monitoring for Mixed-Criticality SoCs: From Phase-Level Precision to ID-Level Efficiency
Mixed-criticality Systems-on-Chip (SoCs) with on-chip interconnects based on the AXI4 open standard protocol lack a protocol-level timeout mechanism, exposing systems to deadlocks and missed real-time deadlines when subordinate devices or managers fail or stall due to hardware faults, radiation-induced upsets, or software errors. This work presents a configurable hardware intellectual property (IP), non-intrusive in fault-free operation, that detects AXI4 protocol violations and timing faults at runtime and restores interconnect liveness through a cut-and- drain isolation mechanism. To address the fundamental trade-off between monitoring granularity and area cost, we introduce three designs at decreasing monitoring granularity: Phase-Level Track-ing (PLT), which provides cycle-accurate fault localization across individual protocol phases; Channel-Level Tracking (CLT), which coalesces per-phase monitors into channel-level supervision; and ID-Level Tracking (ILT), which achieves sub-linear area scaling by monitoring only per-ID transaction boundaries. Synthesized in GlobalFoundries 12 nm technology, CLT reduces area by 36.7% relative to PLT while preserving worst-case detection bounds at a minimal detection latency overhead, whereas ILT achieves an 89.2% area reduction suitable for tightly constrained deployments at the cost of a 3.7x higher median detection latency with coarser fault localization. Fault injection campaigns on a RISC-V SoC across 1.2 million scenarios confirm that no fault manifesting as an AXI4 protocol or liveness violation escaped detection, with observed detection latencies consistently bounded by theoretical worst-case predictions.
Read the full abstract
High Energy Physics (Phenomenology)
Meson gravitational D-form factors and symmetry breaking in low-energy QCD
We investigate meson gravitational D-form factors in the three-flavor linear sigma model and their connection to symmetry breaking in low-energy QCD. Within the scalar meson dominance picture, we examine the scalar meson exchange contributions and their relation to meson masses through scalar-meson couplings. In the three-flavor symmetric limit, we first derive analytical expressions for the D-form factors of pseudoscalar and scalar mesons. In the chiral limit, the forward-limit value of the octet pseudoscalar D-form factor is fixed to $-1$, whereas the $U(1)_A$ anomaly generates additional contributions to the singlet pseudoscalar and scalar-meson D-form factors. For realistic flavor breaking, the interaction introduced to reproduce the scalar-meson mass hierarchy below $1$ GeV significantly affects the scalar-meson D-form factors. We further compare the linear sigma model results with those obtained from a chiral perturbation theory Lagrangian including a dilatonic scalar field and show that their differences are governed by the canonical mass dimensions of the interaction terms contributing to meson mass generation. These results indicate that meson D-form factors provide sensitive probes of symmetry-breaking structures in low-energy hadron physics.
Read the full abstract
Robotics
A Hybrid PEM-GP Framework for Uncertainty-Aware System Identification of Quadcopters
Accurate dynamic models play a central role in achieving reliable control of quadcopters. Classical system identification methods remain widely used, mainly because of their interpretability. However, they often fail to capture important nonlinear effects, especially in small-scale aerial platforms where such effects become more pronounced. Data-driven approaches offer a different perspective. They can represent complex nonlinear dynamics more effectively, but this comes at the cost of reduced interpretability and the absence of well-calibrated uncertainty estimates. In this work, we propose a framework that combines physics-based modeling with data-driven learning, while explicitly accounting for uncertainty. A physics-based model is first identified using the Prediction Error Method (PEM), which captures the main structure of the system. The remaining dynamics are then modeled using a Gaussian Process (GP), allowing the residual behavior to be learned directly from data. This separation makes it possible to distinguish between known physical effects and unmodeled dynamics. The proposed framework is validated on a Duckiedrone-like experimental setup. The results show that the PEM-GP model achieves prediction accuracy comparable to that of a Long Short-Term Memory (LSTM) network, while additionally providing calibrated uncertainty estimates. This combination improves model reliability and supports uncertainty-aware decision-making.
Read the full abstract
Fluid Dynamics
A Physics-Informed, PIV-Based Method for Surface Pressure Reconstruction
A physics-informed method for PIV-based surface pressure reconstruction that combines least-squares optimization with both local and global physical constraints is proposed. The standard deviation of the surface–normal pressure gradient is introduced as a physically meaningful weighting function in the optimization process. Moreover, a physical global constraint is also introduced to reconstruct surface pressure outside the suction region. The results show very good agreement with data obtained from pressure taps. Near the leading edge, the surface pressure reconstruction error has been reduced by one-third.