Mathematical Physics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 28 August
Spin-boson models are paradigmatic examples of open quantum systems and serve as the theoretical paradigm for current quantum computers based on single-ion trap technology. Despite their ubiquity, a complete spectral diagonalization of these models remains an open problem, except in highly singular regimes. This paper establishes a rigorous framework for the approximate diagonalization of generalized spin-boson systems. Our approach is inspired by the Brockett-Wegner double-commutator flow, which is a non-linear differential equation governing the evolution of (here unbounded) operators. Unlike relatively recent applications of this flow to quadratic Hamiltonians in quantum field theory, it does not close in the spin-boson context. We overcome this fundamental obstruction by performing a detailed analysis of the resulting non-closed algebraic structure, allowing us to explicitly bound the higher-order error term with respect to the spin-boson coupling strength. Consequently, this work provides the first mathematically rigorous justification for several heuristic diagonalization techniques widely employed in the theoretical physics literature for small-coupling regimes, in the simplest non-trivial cases. More broadly, our framework renders a flow-based algorithm feasible for systematic higher-order diagonalization and self-energy renormalization. This strategy is conceptually akin to multi-scale analysis, or, much more recently, to the iterative, local Lie-Schwinger block-diagonalization method by Fröhlich and Pizzo.
Computer Vision
Explainable Diabetic Retinopathy Classification Using Vision Foundation Models
Diabetic retinopathy (DR) is a leading cause of preventable blindness, and automated screening must be both accurate and trustworthy. We study an explainable DR classification framework built on vision foundation models and three transfer learning strategies. Three backbones, DINOv2, CLIP, and Vision Transformer (ViT), were adapted by full fine-tuning, linear probing, and Low-Rank Adaptation (LoRA). Models were trained and internally evaluated on the ODIR dataset and externally evaluated on APTOS to assess generalization. DINOv2-LoRA achieved the highest internal AUROC (0.758), whereas DINOv2 and ViT full fine-tuning achieved the highest external AUROC (0.920). Probability calibration was assessed with reliability analysis after isotonic regression. For explainability, Grad-CAM and HiResCAM maps were compared with expert-annotated lesion masks from the IDRiD dataset using Dice, Intersection over Union (IoU), and Pointing Game metrics. The results show that foundation models, particularly DINOv2, provide strong predictive performance, and that LoRA is a parameter-efficient alternative to full fine-tuning. Spatial overlap between explanation maps and annotated lesions was limited (mean Dice of about 0.13 for the combined lesion mask), which highlights the value of quantitatively testing whether model attention corresponds to clinically relevant retinal lesions.
Read the full abstract
Computer Vision
NumBench: Diagnosing Counting Failures in Text-to-Image Models
Text-to-image (T2I) models often generate the wrong number of objects, yet existing benchmarks are too small or weakly controlled to explain why. We introduce NumBench, a benchmark of 640{,}000 prompts spanning 1{,}600 categories and counts from 1 to 100. Its factorial design varies object composition, spatial guidance, and appearance conditions while balancing counts and category exposure. We also develop a process model in which requested instances compete for a finite set of resolvable image regions. The model predicts a near-quadratic collision deficit at low occupancy and shows how coordinated placement reduces it. For scalable evaluation, we propose the Confidence-Weighted Numeric Precision Score (\cwnps), which aggregates three calibrated detectors and discounts uncertain proposals. Across five commercial systems, two open models, and two specialized counting methods, performance declines sharply with requested count; all evaluated methods are weak above 50 objects. Count range has the largest measured effect, followed by layout and composition. Grid guidance is strongest among guided layouts, consistent with the coordination prediction, although the analysis does not establish collision as the sole cause. A 14{,}400-image human study supports automated evaluation through count 50, while results on 243 natural-language prompts show transfer beyond NumBench templates.
Read the full abstract
Computer Vision
Cut-ViT: Task-Specific Model Pruning via Gram Anchoring Subspace Consistency
Pruning visual foundation models has attracted considerable attention. However, existing methods focus on rigid point-to-point token alignment on a single dataset for pruning, suffering from two limitations: i) robustness degradation, and ii) task-specificity deficiency. To address these limitations, we propose a task-specific pruning pipeline, named Cut-ViT. Specifically, we first construct gram anchoring matrices from both spatial and semantic perspectives, and perform the subspace decomposition to extract the corresponding subspace bases. Basis-agnostic and residual constraints are then adopted to align the gram subspaces between the native and pruned DINOv3 models along spatial and channel dimensions, enabling subnetworks to inherit robust feature representations of native DINOv3. Furthermore, we design spectral entropy adaptation, which quantifies the information density of feature manifolds along spatial and channel dimensions, thereby adapting the pruning objective to specific downstream tasks. Experiments show that Cut-ViT requires approximately one minute on a single A100 GPU to obtain subnetworks at various sparsity levels, using only 20.9% of the time and 45.5% of the GPU memory compared with previous methods, while achieving SOTA performance on six tasks across nine datasets.
Read the full abstract
Algebraic Geometry
Arithmetic purity of strong approximation for toric varieties with constant global sections
Let $X$ be a smooth toric variety over a number field $k$ such that the global sections of $X$ are constant. We show that $U(k)$ is dense in $U(\mathbf A_k)^{\text{Br}_1 (U)}$ for any open subset $U$ of $X$ such that the codimension of $X\setminus U$ in $X$ is at least 2.
Fluid Dynamics
Inclination-Induced Crossover in the Velocity Scaling of Lubrication-Mediated Droplet Motion
We experimentally investigate the creeping motion of water droplets sliding along an inclined surface immersed in a viscous oil. The droplet velocity, width, and height are measured over a broad range of droplet sizes and inclination angles. The velocity increases with droplet radius according to a power law, but the corresponding exponent depends strongly on the inclination angle. At small inclinations, the exponent is close to 3/2, whereas at larger inclinations it progressively approaches 9/4. The results therefore reveal a continuous inclination-induced crossover in the velocity-size scaling of lubrication-mediated droplet motion. A scaling analysis based on Landau-Levich-Derjaguin film formation and viscous dissipation in the dynamic meniscus reproduces these limiting behaviors, which correspond to the quasi-spherical and pancake scaling regimes, respectively. Remarkably, the evolution of the velocity scaling is substantially more pronounced than the corresponding evolution of the macroscopic droplet shape. While the velocity data progressively approach the pancake scaling law at large inclinations, the global droplet dimensions evolve in a more complex manner than expected from a simple pancake-shape picture. In particular, the approach to pancake-like velocity scaling occurs even when the measured droplet dimensions remain far from the corresponding pancake-limit geometry. These observations suggest a crossover in the dominant lubrication-dissipation mechanism beneath the droplet that is not directly reflected in the global droplet morphology. The results identify inclination angle as a key control parameter governing lubrication-mediated droplet motion and highlight the distinction between global shape evolution and local dissipation dynamics in liquid-liquid systems.
Read the full abstract
Programming Languages
URIUM: A Programming Language for a Practical Open Course on Compiler Design
This paper presents the definition of a simple programming language used as the basis for developing a practical compiler design course. The course explains step by step how to build a compiler, from the initial analysis stages to code generation. The developed compiler generates code for various processors (MIPS, Intel, and RISC-V) and operating systems (MS-Windows and Linux). The course can be adapted to different levels of difficulty and can be used as a starting point for explaining more advanced topics.
Read the full abstract
High Energy Physics (Phenomenology)
Minimal-Length Deformations of Chiral Quark Matter in a 2+1-Flavor NJL Model
In this paper, a minimal-length inspired deformation of hot and dense chiral quark matter is formulated within a \(2+1\)-flavor Nambu-Jona-Lasinio framework. The deformation is based on a complete Jacobi-consistent three-dimensional quadratic GUP algebra and modifies the microscopic phase-space density of states while leaving the quasiparticle spectrum unchanged, leading to a thermodynamically closed mean-field theory with coupled light and strange gap equations. The analysis shows that ultraviolet phase-space suppression weakens dynamical chiral symmetry breaking, lowers the light and strange constituent masses, and shifts the light-sector chiral transition boundary, including its crossover and critical endpoint toward lower temperature and chemical potential. The numerical results are obtained from a single self-consistent solution of the coupled gap equations, and explicit endpoint comparisons show that the temperature-dependent and chemical-potential dependent results are mutually consistent. The result is a systematic effective-model study of how minimal-length inspired ultraviolet deformations reorganize coupled chiral dynamics and bulk quark-matter thermodynamics.
Read the full abstract
Algebraic Topology
On the Formality of Configuration Spaces of $\mathbb{R}^{n'} \times \mathbb{C}^{n}$
This paper presents a complete classification of the formality of configuration spaces of $\mathbb{R}^{n'} \times \mathbb{C}^{n}$. We define a constructible de Rham-Dolbeault cohomology theory which provides a constructible CDGA (commutative differential graded algebra) model of $\Conf_m(\mathbb{R}^{n'} \times \mathbb{C}^{n})$. For $(n'=0,n\ge2)$ or $(n'=1,n\ge1)$, the CDGAs are non-formal. For $n'\ge2,n\ge1$, we establish an explicit quasi-isomorphism between the constructible CDGA and its cohomology by using a diagrammatic CDGA of admissible diagrams and a regularized configuration space integral, which leads to the formality. As an application, we show that the local operator algebra of a topological-holomorphic field theory on $\mathbb{R}^{n'} \times \mathbb{C}^{n}$ ($n'\ge2,n\ge1$) is homotopically equivalent to a higher dimensional analog of vertex algebras.
Read the full abstract
Game Theory
Fine-Tuning Autobidders with Group Relative Policy Optimization
Automated bidding (autobidding) is a core component of modern online advertising systems. Within this component, advertisers delegate sequential bid decisions to algorithms that must maximize campaign value while adhering to constraints such as a limited budget and a target cost-per-click (CPC). One of the approaches to resolve the autobidding problem is to formulate it as a Markov decision process and use reinforcement learning (RL) to train a bid generation function. The standard RL framework is actor-critic, which consists of an actor network that generates actions and a critic network that estimates the value of those actions. In our setting, the action is typically a bid or related pacing multiplier, and the value is the expected return from the auction given the bid. However, the alternating training of actor-critic RL models leads to instability and reduced robustness to noise. To address these issues, we adapt the Group Relative Policy Optimization (GRPO) framework to the autobidding setting. This framework is a critic-free policy-gradient method originally developed for large language model post-training, where the ground-truth target is unknown. The autobidding setting shares this property, since the optimal bid is unknown in advance. Moreover, GRPO in the LLM domain is used to fine-tune the pre-trained model, and we use the same technique to enhance the performance of the strong heuristic baseline. We empirically compare Autobidding GRPO with actor-critic models, simple heuristics, and controller-based methods on the BAT, iPinYou, and AuctionNet benchmarks. Extensive experiments show that Autobidding GRPO consistently outperforms baselines in clicks and is the best or second-best method in conversion volume.
Read the full abstract
Machine Learning
Performative Privacy: When Differential Privacy Maximizes Utility
Privacy-preserving learning is often motivated by the idea that protecting users' data can preserve trust and thus participation, improving utility in the long term. However, this claim has not been formalized so far. In parallel, performative learning provides a framework for studying learning systems whose deployment affects the data they later observe. In this work, we bring these two perspectives together and introduce performative privacy, where data leakage reduces future participation. We study a simple model where agents repeatedly contribute data for mean estimation but may leave the system when their data is leaked. Privacy is implemented through differentially private mechanisms, creating a trade-off between estimation noise and future participation. We show, through a theoretical study of the dynamics and numerical experiments, that a finite privacy budget can outperform non-private estimation in the long term when the feedback loop between leakage and participation is sufficiently strong. This provides first evidence that differential privacy can be optimal not only as a protection mechanism, but also from the perspective of long-term utility.
Read the full abstract
Symplectic Geometry
Semitoric systems with $2n$ Focus-Focus singularities
In this paper, we provide new explicit examples of semitoric systems. More precisely, for any natural number $n$, we give a semtioric moment map $F_n$ with $2n$ focus-focus singularities defined on a symplectic manifold $(M_n,\omega_n)$ of diffeomorphism type $\mathbb{CP}^2\sharp(2n+1)\overline{\mathbb{CP}}{}^2$.
Read the full abstract
Signal Processing
CP-Aware OFDM-Based OOK Signaling
This letter addresses the challenge of cyclic prefix (CP) problem in orthogonal frequency division multiplexing (OFDM)-based on-off keying (OOK) generation for low-power Internet of Things (IoT) systems. We propose a CP-aware waveform design that generates the OOK signal over the entire CP-OFDM symbol duration, potentially breaking subcarrier orthogonality after passing through the channel if the CP is not matched. To enable such a mechanism, OOK has to be generated independently and a modification to the existing OFDM-based OOK generation block is needed. Additionally, a block-wise time shift can be applied to minimize CP mismatch. Analytical discussions and simulations confirm that the resulting interference on legacy subcarriers remains controlled, while OOK detection performance significantly improves, albeit with trade-offs that should be carefully considered in practical deployments.
Read the full abstract
Computer Vision
UniLipi: A Unified Multi-Script OCR for Historical Indic Manuscripts
Optical character recognition (OCR) for handwritten Indic manuscripts is essential for large-scale digitization and computational access to manuscript heritage. However, existing approaches are typically developed for one script at a time and require substantial script-specific customization. This limits scalability and practical deployment across diverse collections. We present UniLipi, a unified multi-script OCR model for handwritten Indic manuscripts trained jointly across 13 Indic scripts within a single framework. UniLipi directly handles realistic manuscript conditions, including extreme variation in line geometry, large variation in line length, and partial interruptions caused by non-textual manuscript entities such as holes, stains, or pictorial illustrations. To operate effectively under ultra low-resource conditions, the model leverages script-aware synthetic manuscript data generation, substantially reducing reliance on large volumes of real annotated data. Beyond historical manuscripts, we show that UniLipi serves as an effective foundational pretrained model. Specifically, its learned representations enable good OCR performance for contemporary Indic handwriting and extend to several non-Indic scripts, including Tibetan, Italian, Latin, and Chinese scripts. In addition to transcription, UniLipi predicts script identity and per-line native character counts, supporting practical manuscript cataloging workflows.
Read the full abstract
Systems and Control
SafeLink-Agent: Agentic Maintenance for Adaptive Bitrate Controllers over Dynamic Starlink Networks
Low Earth orbit (LEO) satellite broadband, represented by Starlink, is making high-resolution video streaming feasible beyond fixed terrestrial coverage. However, Starlink access links change across time and regions, exposing adaptive bitrate (ABR) streaming to shifting throughput tails, latency, volatility, and handover conditions. Existing ABR controllers are usually designed, tuned, or trained for specific network conditions, making it difficult to handle newly exposed hard Starlink profiles. This paper proposes SafeLink-Agent, an agentic maintenance framework for ABR controllers over dynamic Starlink networks. SafeLink-Agent summarizes exposed failures and uses a large language model (LLM)-based agentic patch proposer to generate candidate patches, while replay verification determines whether each patch can be safely committed. The framework supports both rule-based controllers and learned controllers under the same maintenance workflow. Experiments on real Starlink networks show that SafeLink-Agent reduces the severe-session ratio of RobustMPC from 2.60% to 0.40% and reduces cumulative severe sessions from 45 to 7 in rolling maintenance. For learned controllers, verified adaptive auditing lowers the average severe-session ratio from 39.01% to 9.79%. These results demonstrate that agentic maintenance can improve ABR robustness under dynamic Starlink access conditions.
Read the full abstract
Computational Physics
Promoting Green Computing in High Energy Physics and Astrophysics
This research topic aims to explore and promote strategies for green computing in high-energy physics and astrophysics. The primary objectives include assessing current energy consumption patterns, identifying opportunities for resource optimization, and fostering cross-disciplinary collaborations to enhance sustainability. Key questions to be addressed include: What are the current energy demands in these fields? How can existing technologies be leveraged for more efficient resource usage? What synergies can be identified between high-energy and astrophysics projects to promote green computing?
Read the full abstract
Computer Vision
Locate Anything in Videos: Rethinking Efficient Generative Spatio-Temporal Video Grounding
Spatio-temporal video grounding (STVG) requires models to identify when a referred event occurs and localize the target entity throughout that interval. Existing multimodal large language models typically serialize dense localization trajectories autoregressively, causing decoding latency to grow with tube length and allowing localization errors to propagate across time. We introduce Parallel Tube Decoding (PTD), a generative formulation that decomposes grounding into a temporal block followed by time-conditioned spatial blocks decoded simultaneously. This removes both token-level and trajectory-level dependencies, reducing the sequential decoding depth to a fixed $1 + 1$ rounds, independent of tube length. To enable parallel spatial generation, we introduce Decoupled Block Attention, which preserves access to shared video-query context while eliminating cross-box dependencies, together with localization-aware policy optimization for temporal boundaries and spatial geometry. On VidSTG, PTD reduces Tube Completion Latency by 79x and increases spatial decoding throughput by 92x over standard autoregressive decoding, while also improving grounding accuracy. With a compact 4B backbone, our model performs favorably well on VidSTG and HC-STVG, and generalizes zero-shot to temporal grounding, grounded VideoQA, and referring video object tracking. Our results show parallel tube generation is an efficient and effective alternative to autoregressive localization in videos.
Read the full abstract
Computer Vision
EXPOSE: Explainable and Domain-Robust Embeddings from Pathology Vision Foundation Models using Sparse Autoencoders
Vision Foundation Models (VFMs) are widely used in computational pathology but remain sensitive to domain shifts arising from variations in staining, tissue preparation, and scanner hardware. A key limitation is that VFM embeddings entangle biological with domain-specific information, hindering cross-domain generalization. We propose Explainable Probing of Cross-Domain Sparse Embeddings (EXPOSE), a framework that uses Sparse Autoencoders (SAEs) as an explainable bottleneck to identify and suppress domain-specific components in VFM embeddings. We train a sparse representation of VFM features, use a linear classifier to identify domain-specific latent dimensions, and mask these features prior to downstream relapse prediction without retraining the backbone model. Experiments on a large prostate cancer dataset with multiple acquisition domains show that SAE features capture both domain- and task-specific information, which are partially disentangled in the latent space. Removing domain-specific features improves cross-domain performance and increases embedding robustness as measured by the Domain Robustness Index (DoRI). Code is available at https://github.com/imsb-uke/expose .
Read the full abstract
Quantum Physics
Postselection-loophole-free Bell test under strict spacetime constraints
Entanglement gives rise to correlations between distant quantum systems that cannot be explained by local realistic theories. Bell inequality violations provide a direct way to reveal these correlations and certify nonlocality, especially when the relevant experimental loopholes are closed. Time-bin encoding, in which quantum information is encoded into well-defined temporal modes, is a commonly used platform for distributing photonic entanglement in optical fibers. Yet loophole-free Bell tests with time-bin entanglement have received comparatively little attention, owing in part to the postselection loophole introduced by conventional interferometric measurements. Here, we demonstrate a fiber-based platform for Bell tests with time-bin entanglement that simultaneously closes the locality, freedom-of-choice, and postselection loopholes. We observe a CHSH violation of $S=2.583 \pm 0.002$, exceeding the local-realistic bound by over 265 standard deviations. Notably, this rigorous certification of nonlocality is achieved at a separation distance of $49.0 \pm 0.7$ m, substantially shorter than previous photonic Bell tests addressing comparable space-time constraints. Beyond its foundational significance, our results demonstrate time-bin entanglement as a viable route towards practical device-independent quantum communication and a future quantum internet.
Read the full abstract
High Energy Physics (Phenomenology)
System-size dependence of strangeness production from p+p to Pb+Pb: quantitative tests of the $K^+/\pi^+$ horn
The pronounced maximum ("horn") in the $K^+/\pi^+$ excitation function observed in central Pb+Pb collisions is well established feature of collision-energy dependence, but its dependence on system size remains poorly understood. We confront hadronic (SMASH) and partonic (PHSD) transport, Glauber core-corona, canonical-ensemble suppression, and the two-phase statistical model of the early stage (SMES) with $K^+/\pi^+$ data spanning the complete NA61/SHINE system-size ladder (p+p, Be+Be, Ar+Sc, Xe+La) together with NA49 Pb+Pb and STAR Au+Au, quantifying every comparison by $\chi^2$. No single framework describes the full range: core-corona is preferred for intermediate systems, SMES with canonical strangeness conservation for heavy systems, and none reproduces the small-system data. Neither transport model generates the horn, with or without partonic degrees of freedom. The data exhibit a step-like enhancement of $K^+/\pi^+$ between Be+Be and Ar+Sc, quantified within the core-corona framework by a step from $\fcore \approx 0.22$ (Be+Be) to $\fcore \approx 0.58$ (Ar+Sc). The strangeness saturation factor $\gs$ rises in two distinct steps, separating geometric core formation from thermodynamic equilibration. The $(K^+/\pi^+)/(K^-/\pi^-)$ double ratio, which at fixed $\sqrts$ cancels the system-independent $\mu_B$ contribution, shows a step between light and heavy systems at a global significance of $3.7\sigma$, providing evidence for genuine strangeness enhancement beyond the pair-production baseline. Preliminary Xe+La data favor core-corona ($\chindf = 5.8$) over SMASH ($\chindf = 23.3$); finalized spectra will sharpen the CC/SMES discrimination at the $27\%$ level.
Read the full abstract
Machine Learning
Beyond Flat Netlist: Hierarchical Graph Representation Learning for Scalable Analysis of Sequential Circuits
Circuit Representation Learning (CRL) offers a powerful paradigm to guide and optimize core Electronic Design Automation (EDA) tasks, but its practical adoption is hindered by the immense scale of industrial netlists and a failure to explicitly model register-level temporal dynamics. To overcome these barriers, we introduce DeepSeq3, a novel hierarchical framework that abstracts circuits into a two-level representation: fine-grained combinational subgraphs partitioned by flip-flops (FFs), and a high-level Super-Node Graph (SNG) that models the register-transfer structure. A dual Graph Neural Network (GNN) architecture learns representations at both levels, capturing local Boolean logic and global state transitions. Crucially, we introduce a state-centric pre-training scheme that predicts the reachability between FF states, endowing the model with a deep understanding of temporal behavior. Demonstrated on large-scale benchmarks, DeepSeq3's approach yields superior scalability and richer representations, reducing bounded model checking (BMC) solving time by 18% while guaranteeing correctness.
Read the full abstract
Algebraic Geometry
The Global Dimension Function on Stability Manifolds
In this paper, we study the reachability and boundary behavior of the global dimension function on Bridgeland stability manifolds. For a smooth projective variety $X$ of dimension $n$, we prove that the infimum of the global dimension function is $n$. This value is attained when $-K_X$ is ample, is taken by every stability condition when $K_X\in\mathrm{Pic}^0(X)$, and is not attained when $K_X$ is big and nef. We also extend these reachability results to pre-stability conditions. In addition, we construct compactifications of reduced stability spaces of smooth projective curves to which the global dimension function extends continuously. Finally, we discuss algebraic models of the reachability problem, including finite-dimensional algebras and semiorthogonal decompositions.
Read the full abstract
General Relativity and Quantum Cosmology
Photon Spheres and Shadows of Covariant Loop Quantum Black Holes in the general $\mu$-scheme
We investigate black hole shadows and photon sphere properties for two families of covariant quantum-corrected black-hole metrics (hereafter called BH-I and BH-II) formulated within a general $\mu$-scheme, parameterised by a power-law exponent $s$ and an amplitude $\xi$. The extension to general (including non-integer) $s$ is a phenomenological interpolation between the $\mu_0$-scheme ($s=0$) and $\bar\mu$-scheme ($s=1$) and reveals a rich phenomenology masked when $s$ is usually fixed to 1 in previous literature. For BH-I, the photon sphere exists for all parameter values and exhibits an exact cancellation at $s=1$ where the coordinate location $r_{\rm ph}=3M$ is restored for any $\xi$. For BH-II, both the horizon and the photon sphere exhibit critical curves in the $(\xi,s)$ parameter plane; the photon sphere disappears when $\xi$ exceeds a closed-form critical value $\xi_c^{\rm PS}(s)$. We prove analytically that the photon sphere is always unstable ($\lambda_{\rm ph}>0$) throughout the physical parameter space of both metrics, and derive closed-form expressions for the Lyapunov exponent. Systematic parameter scans reveal that for fixed $\xi$ the shadow radius decreases monotonically with $s$ for BH-II and non-monotonically for BH-I. We derive small-parameter analytic expansions for the photon-sphere radius, shadow radius, and Lyapunov exponent, and demonstrate that a single shadow measurement suffers an observational degeneracy in the two-dimensional $(\xi,s)$ space; the degeneracy can be broken by a simultaneous measurement of the Lyapunov exponent. Using Event Horizon Telescope (EHT) measurements, we derive constraints on the $(\xi,s)$ parameter space and find that BH-II is constrained roughly $2$–$4$ times more tightly than BH-I.
Read the full abstract
Optics
AAA rational approximation for scatterometry of nanoscale gratings
We investigate the applicability of AAA rational approximation to scatterometry-related response functions of nanoscale gratings. A silicon line grating with material dispersion is considered. We compare the convergence of AAA rational approximation with that of cubic spline interpolation as the number of frequency samples is increased. The results demonstrate the potential of rational approximation, consistent with the pole structure of the considered response function arising from resonance effects of the grating.
Read the full abstract
Machine Learning
Biologically Inspired Mechanisms for Facilitating Grokking in Multilayer Perceptrons
Grokking is a delayed transition from memorization to generalization that is often accompanied by substantial reorganization of internal representations. This paper studies whether biologically inspired mechanisms, many of which are not commonly incorporated into artificial neural networks, can actively promote this transition by regulating hidden-layer computation at the levels of neuronal activity, response, and effective connectivity. We augment a multilayer perceptron with input gating, structural plasticity, gain modulation, threshold modulation, homeostasis, lateral inhibition, and activation decorrelation, and evaluate these mechanisms through systematic ablations on two established grokking benchmarks: sparse parity and noisy XOR classification. The results show that the mechanisms contribute unequally to generalization. Homeostasis provides the strongest and most consistent benefit, while structural sparsification emerges as the second major mechanism. The remaining biologically inspired mechanisms have smaller or less consistent effects in the present experiments. For both problems, the results support the common principle that explicit regulation of neuron utilization and effective connectivity can improve the emergence of generalizable internal computation. These findings motivate broader investigation of biologically inspired activity regulation and adaptive sparsification, including in large language models, where they may accelerate the development of generalizable representations and reduce the optimization time required for robust generalization.
Read the full abstract
Distributed Computing
Memory-efficient GPU pipelines for real-time non-line-of-sight reconstruction
Non-line-of-sight (NLOS) imaging reconstructs scenes hidden around a corner from indirect light recorded by a single-photon avalanche diode (SPAD). A single reconstruction is a large inverse problem: billions of photon timestamps must be binned, moved through memory, transformed and inverted. As SPAD arrays raise acquisition throughput, reconstruction becomes the limiting stage. We rebuild the GPU execution of two established wave-based algorithms, f-k migration and phasor-fields, for both streaming and offline processing. On the phasor-fields side we assemble the ring-and-radius kernels of previous work once and offline, using the analytic Fourier transform of a ring, so the propagation kernel never exists in dense form at runtime, reducing the memory and bandwidth. We reorganize the pipeline of both algorithms with fused kernels, warp-level photon binning, batched transforms, CUDA graph replay, and FP16 storage applied only where it reduces the actual bottleneck. Our implementations are up to 42x faster than the reference streaming pipeline and up to 14x faster than the fastest published GPU baseline, all while using a fraction of the memory (down to 2.5%), enabling vastly larger and finer reconstructions on the same hardware, or comparable ones within a much lower memory budget. We report an ablation of each implementation choice and propose three denoising strategies enabled by the resulting frame budget for next-generation NLOS video processing.
Read the full abstract
Physics and Society
Benchmarking large language model agent societies against human behavioural distributions
Populations of large language model agents are increasingly used as experimental societies. Three doubts shadow every such result: whether the agents behave like the humans they stand in for, whether a finding survives changes to the apparatus that leave the rules untouched, and whether apparent social dynamics are interaction at all rather than the reproduction of experiments the models have read. This article introduces SILICA, an open instrument that tests all three. Five environments carry published human anchors, each paired with perturbations that re-render the same rules and with variants whose payoffs point away from the memorised result. Twelve open-weight models were run through it on a single consumer graphics card. Agreement with human data is confined to starting points: first-round public-goods contributions fall inside the equivalence margin for eight of eleven models, while no model matches end-state contributions or the human corridor of cooperation. Merely swapping the order in which two actions are listed costs one model 58 points of cooperation. Presenting responders with a fixed schedule of offers shows that only one model, the sole reasoning-trained one, places its acceptance threshold where the incentive requires; two move theirs part of the way, two move them the wrong way, and three never acquire one. Conventions form through a shared prior over the names rather than through negotiation, though negotiation reappears once that prior is disrupted. On the certification ladder defined here, current silicon societies support exploratory claims and no more.
Read the full abstract
Algebraic Geometry
Independence of the Grothendieck classes of twisted symplectic flag varieties
In this short note, making use of the theory of noncommutative motives, we prove that the Grothendieck classes of the twisted symplectic flag varieties are linearly independent and moreover homogeneous-quadratic independent in the Grothendieck ring of algebraic varieties.
Systems and Control
Distributed Model Predictive Control for Optimal Consensus of Constrained Heterogeneous Multi-agent Systems
This paper investigates the distributed optimal consensus control problem of constrained heterogeneous multi-agent systems within a model predictive control (MPC) scheme. Both the control input sequence and the dynamically feasible consensus equilibrium are optimized simultaneously within the proposed MPC framework to improve consensus performance, yielding a coupled constrained optimization problem at each prediction time. A distributed primal–dual algorithm is developed to solve the resulting optimization problem, and locally verifiable conditions are derived to guarantee its convergence. Furthermore, sufficient terminal conditions are established for the proposed MPC framework to guarantee the recursive feasibility and asymptotic consensus of the closed-loop heterogeneous multi-agent systems. Finally, numerical simulations verify the effectiveness of the proposed approach.
Read the full abstract
Machine Learning (Statistics)
Conformal Risk-Averse Decision Making with Optimized Certainty Equivalent Risk Control
We study risk-averse decision making, in which an agent selects actions while being uncertain about the true system state. The risk is measured via optimized certainty equivalent (OCE) metrics, which generalize popular criteria such as mean-variance risk and conditional value-at-risk (CVaR). We characterize the optimal policy under known distributions, and show that it reduces to a prediction set-based solution for the CVaR. This provides an operational interpretation of conformal prediction-type prediction sets. For unknown distributions, we develop a data-driven calibration strategy, based on a synthetic model for the likelihood and held-out calibration data, yielding high-probability control of the OCE risk. The approach is evaluated on two wireless beamforming settings.