Classical Analysis and ODEs
Mean values, order, and convergence in the Hardy space of Dirichlet series
Ole Fredrik Brevig and Athanasios Kouroupis
Let $\mathscr{H}^2$ denote the
Hilbert space of Dirichlet series with square-summable coefficients. It follows from a
theorem of Bohr that if $f$ in $\mathscr{H}^2$ has a bounded analytic continuation to the right half-plane, then its norm can be computed from the mean values \[\|f\|_{\mathscr{H}^2}^2 = \lim_{\sigma\to 0^+} \lim_{T\to\infty} \frac{1}{2T} \int_{-T}^T \lvert f(\sigma+it) \rvert^2\,dt.\] We investigate to what extent the assumption that the analytic continuation be bounded can be relaxed, and obtain optimal results in two different directions. The first direction concerns the order of the analytic continuation, and yields as a corollary a version of the classical Landau–Schnee convergence
theorem for $\mathscr{H}^2$. The second direction concerns the abscissa of convergence, and resolves a problem posed by McCarthy.
Read the full abstract ▾
High Energy Physics (Theory)
Knizhnik-Zamolodchikov equations and the analytic structure of strings in AdS$_3$
Sergio Iguri, Nicolas Kovensky and Julian H. Toro
The pole structure of string correlators in AdS$_3$ encodes essential information about the AdS$_3$/CFT$_2$ duality. The corresponding residues for spectrally flowed three-point functions have been extracted from their exact expressions and shown to match the predictions of the
holographic CFT. Away from the tensionless point, the latter is defined perturbatively in terms of a deformed symmetric orbifold. For $n$-point functions, such a matching has so far only been achieved for a specific subset of poles, and only by means of free-field techniques. Here we develop an alternative approach. We show that, starting from the localization properties of worldsheet correlators, one can solve the Knizhnik–Zamolodchikov (KZ) equations on the corresponding locus and perform the integration over worldsheet
moduli explicitly for any $n$ and for arbitrary values of the spectral flow charges. In this way we reproduce, directly from the exact worldsheet theory and including the normalization, the residues previously obtained with free-field methods. We expect this to provide a systematic framework for computing the remaining residues and, ultimately, for determining the full analytic structure of string amplitudes in AdS$_3$.
Read the full abstract ▾
Instrumentation and Methods for Astrophysics
Refactoring the SIXTE simulator: Towards a more modular code base
Christian Kirsch, Maximilian Lorenz, Lea Dauner et al. (+7 more)
The SIXTE (SImulation of X-ray
TElescopes) software is a general end-to-end simulation toolkit for X-ray observations, covering the full observation process from source
photon generation to detector readout and the production of high-level output files. It is the official simulator for existing and future X-ray missions, such as eROSITA, NewAthena, THESEUS and AXIS. Originally being designed as a simulator for eROSITA, the addition of new instrument and
telescope types over several years has made the original code base increasingly difficult to maintain. As such, we have refactored the code, changing
languages from C to C++ and switching to a more modular software design to facilitate the implementation of new models. This proceeding highlights some of the design choices used during the refactoring as well as its effects on maintenance and new feature development one year after release of the refactored code base.
Read the full abstract ▾
All authors Christian Kirsch, Maximilian Lorenz, Lea Dauner, Esin G. Gülbahar, Karan Pal, Neo Reinmann, Jakob Stierhof, Philipp Thalhammer, Thomas Dauser, Jörn Wilms
Robotics
Terrain-Aware Autonomous Planetary Exploration for Exteroceptive-Proprioceptive Mapping with Quadruped Scouts
Alberto Sanchez-Delgado, João Carlos Virgolino Soares, Victor Barasuol et al. (+1 more)
Autonomous planetary exploration requires
robots to navigate unknown, uneven terrain while assessing risk, traversability, and energetic cost. Quadruped scouts are well suited for this task because they can traverse irregular surfaces and gather mobility-relevant information during locomotion. This paper presents a terrain-aware exploration framework that combines exteroceptive and proprioceptive mapping for a quadruped
robot in lunar-like environments. An onboard RGB-D camera builds
robot-centered elevation maps, estimates
geometric traversability, and derives
navigation costs for autonomous planning. In parallel, proprioceptive measurements provide interaction-aware terrain cues that complement geometry-based assessment. Local maps are incrementally registered into a global multi-layer representation, which is used by an exploration module to select targets in unexplored regions of interest. The targets are reached by an autonomous
navigation system that guides collision-aware motion using the available map and cost layers. Simulation results on NVIDIA Isaac Sim show autonomous exploration, map expansion, and spatial association between terrain geometry and
robot-terrain interaction. Subsequent
navigation using this information exhibits lower average Cost of Transport (CoT) than initial exploration.
Read the full abstract ▾
All authors Alberto Sanchez-Delgado, João Carlos Virgolino Soares, Victor Barasuol, Claudio Semini
Mesoscale and Nanoscale Physics
Dynamical transition in non-Hermitian Chern insulator
Zhi-Qiang Zhang, Yuan-Hang Ren, Ming Lu et al. (+1 more)
We unveil a peculiar dynamical transition for the propagation of wave packets in non-Hermitian Chern
insulators, where the
evolution of the topological wave packets at the edge is not solely determined by the topological Chern number. Unlike the Hermitian Chern
insulator, where a wave packet initiated at the edge propagates along the system boundary, here it may instead penetrate into the bulk. This behavior is attributed to the competition between the localization induced by conventional
topology and the non-Hermitian skin effect. Specifically, when the former dominant, the wave packet will evolve along the boundary; otherwise, it will spread into the bulk. These features demonstrate that while the generalized-Brillouin zone framework reliably predicts topological phase transitions, it does not on its own specify the dynamical transitions.
Read the full abstract ▾
All authors Zhi-Qiang Zhang, Yuan-Hang Ren, Ming Lu, Hua Jiang
Computer Vision
From Scores to Samples: Elastic Forcing for Autoregressive Video Generation
Chi Zhang, Yueyi Liu, Shi Haoyang et al. (+5 more)
Few-step autoregressive
video generation commonly relies on Distribution Matching
Distillation (DMD), requiring a bidirectional
diffusion teacher and an online fake-score model. We instead learn the rollout distribution directly from reference
videos, eliminating both score models during post-training. Our framework minimizes maximum mean discrepancy (MMD) in frozen self-supervised
video representation spaces, using a hybrid Nyström–Monte Carlo
estimator to balance approximation bias and sampling variance.
Memory-efficient replay and gradient subsampling make this objective practical. Using the same architecture and initialization as Self-Forcing, our 1.3B model improves the VBench Total score from 83.80 to 84.64 while retaining 17 FPS. Removing auxiliary score models also enables 14B post-training on eight H200 GPUs. Beyond
distillation, learning from reference
videos enables the acquisition of new visual styles, semantic concepts, and spatial priors without a target-specific
diffusion teacher.
Read the full abstract ▾
All authors Chi Zhang, Yueyi Liu, Shi Haoyang, Ruichuan An, Haoyu Li, Yuhang Wu, Sen Cui, Miao Liu
Computer Vision
AHMAD: Adaptive Hybrid Multi-task Vision Learning with Assisted Distillation for Keypoint Detection
Mohammad Mahdi, Nedyalko Prisadnikov, Yuqian Fu et al. (+3 more)
Generalist multitasking
vision models aim to unify multiple
vision tasks within a single framework, enabling more efficient and versatile learning. However, handling diverse
vision tasks – spanning dense and sparse predictions – remains challenging due to their inherently varying output structures. In this paper, we propose AHMAD, a simple yet effective framework for generalist multitask learning that integrates different key
vision tasks: semantic
segmentation, instance
segmentation, depth estimation, keypoint detection, and object detection. Our approach incorporates these five tasks into a unified structure: a shared encoder-decoder with several lightweight task-specific projectors. Under the multitask learning paradigm, we observed a complementary performance gain, achieving a state-of-the-art PQ of 53.1 and an mIoU of 66.5 for COCO-val panoptic and semantic
segmentation, respectively. Additionally, for top-down keypoint detection, which typically incurs high computational overhead due to multiple forward passes, we introduce a knowledge
distillation-based method that enables a single forward pass over the entire image, greatly improving efficiency. Ultimately, our model delivers a lightweight yet effective generalist multitask learning framework, demonstrating strong performance across five
vision tasks.
Read the full abstract ▾
All authors Mohammad Mahdi, Nedyalko Prisadnikov, Yuqian Fu, Carmelo Scribano, Danda Pani Paudel, Luc Van Gool
Quantum Physics
Seungcheol Oh, Chaemoon Im, Daeyeun Kim et al. (+4 more)
The output of a parameterized
quantum circuit (PQC) can be expressed as a finite
Fourier series whose accessible frequencies are fixed by the data-encoding gates. While the encoder determines which frequencies can appear, the corresponding
Fourier coefficients depend on how the trainable and entangling gates are arranged. Although existing studies provide metrics for characterizing how gate structure affects
Fourier coefficients, they do not translate these analyses into an explicit design criterion specifying how gates should be arranged to make a target coefficient reachable. In this paper, we provide such a criterion. Using the adjoint action of the encoding generator, we decompose operator space into two-dimensional invariant planes indexed by frequency and show that each
Fourier coefficient is exactly a sum of bilinear projections of the effective state and observable onto the planes at that frequency. Because, for Pauli encodings, high-frequency planes are spanned by mixed multi-
qubit Pauli strings, a target coefficient can contribute to the output only when local rotations and entangling layers are arranged so that both the effective state and observable acquire support on one of its planes. For Pauli readouts and commuting two-
qubit entanglers, this yields a
circuit design rule that specifies, for a given interaction graph, a placement of local rotations and entangling layers that makes a target
Fourier coefficient reachable. Using a single encoding layer, we validate the proposed design through a placement ablation, regression tasks from PDEBench and physics-informed Maxwell field modeling, and further demonstrate its robustness to moderate simulated gate noise and damping.
Read the full abstract ▾
All authors Seungcheol Oh, Chaemoon Im, Daeyeun Kim, Soohyun Park, Vaneet Aggarwal, Mohsen Heidari, Joongheon Kim
Quantum Physics
Temporal trade-offs in high-dimensional entanglement: a comprehensive noise model for optimal time-bin QKD protocols
Alexandra E. Bergmayr-Mann, Gláucia Murta and Marcus Huber
High-dimensional time-bin
entanglement is known for having high potential for
quantum key distribution (QKD) applications, being easily implementable, robust and may offer better keyrates than simple
qubit protocols. The temporal encoding, combined with limited clock resolution, however, implies a trade-off: Is it better to send more low-dimensionally encoded
photons or few high-dimensional ones? Answering this question presents a hard challenge as it depends on the entire context of the protocol, including the production rates, losses, dark counts, timing jitters and many more. We answer this question for single Franson based interferometers and asymptotic key rate and shared
entanglement as main figures of merit. We present a flexible noise model incorporating all relevant parameters and find that there are indeed regions where high-dimensional encoding still outperforms any
qubit based version when the pair arrival rate is limited, for instance through loss or limited pump power.
Read the full abstract ▾
Cryptography and Security
Beyond Scalar Probes: Exploiting Vector-Valued Outputs in ReLU Networks For Signature Extraction
Gorka Abad, Claude Carlet, Ermes Franch et al. (+2 more)
We revisit cryptanalytic extraction of ReLU networks from a
geometric and algebraic perspective. Rather than restricting
attention to a single output component, we study the full vector-valued behavior across adjacent linear regions. This leads to a rank-one characterization of Jacobian differences that recovers the usual row-signature information while also revealing complementary column-side information. Our experiments show how this additional structure can be used in ex- traction and improves numerical estimation under different numerical- precision regimes (float64, float32 and float16). We extend the analysis beyond the high-precision and output-rounding settings commonly con- sidered in the literature towards the low-precision settings encountered in many practical settings.
Read the full abstract ▾
All authors Gorka Abad, Claude Carlet, Ermes Franch, Stjepan Picek, Vincent Rijmen
Computation and Language
Who Is Left of Whom? Tracing Spatial Evidence and Role Binding in Relative-Position Reasoning
Yingjin Song, Denis Paperno and Albert Gatt
High instance-level accuracy can mask inconsistencies in spatial
reasoning when objects exchange positions or their roles are reversed in the query. The internal representations supporting relative-position
reasoning remain poorly understood. We investigate two complementary components of this process: tracking object locations in the input and representing their query roles. Across three VLMs with visual or textual inputs and their
language-model backbones, activation patching reveals a staged progression from early-layer source representations through intermediate-layer query-object representations to late-layer answer states. Targeted interventions further establish causal links along this progression: manipulating source-side representations shifts location information at query-object mentions and ultimately alters relation predictions. Beyond object-location information, we also identify a stable query-side direction associated with the roles of the two objects in the comparison. Steering along directions estimated on synthetic scenes generalizes to natural-image
benchmarks, improving accuracy and both forms of paired consistency in most settings without retraining. Our findings reveal complementary components of relational
reasoning across visual and textual settings and show how targeted interventions can improve the consistency of models' behavior.
Read the full abstract ▾
Statistical Mechanics
Out-of-equilibrium finite-time percolation transitions and spinodal-like behaviors after quenches across magnetic first-order transitions of Ising systems
Andrea Pelissetto, Davide Rossini and Ettore Vicari
We analyze the out-of-
equilibrium relaxational dynamics of
ferromagnetic Ising-like systems driven across their low-temperature
magnetic first-order transition (FOT) line, by sudden and slow variations of an external homogenous
magnetic field $h$. As a paradigmatic example, we consider a two-dimensional Ising system driven across its low-temperature FOT line by suddently quenching $h$ from $h_i<0$ to $h>0$, inducing a transition from the negatively to the positively magnetized phase. We show the emergence of a dynamic percolation transition at a finite critical time $t_c(h)$ along the post-quench
evolution for finite (sufficiently small) values of $h$, which is marked by the percolation of the largest positive-
spin cluster and the antipercolation of the largest negative-
spin cluster. This out-of-
equilibrium percolation transition displays a finite-size scaling behavior as in the standard random-percolation case. However, while the fractal dimension of the percolating clusters is consistent with the random-percolation value, the exponent controlling the approach to criticality differs and depends on $h$. We also show that the percolation transition marks the passage from the metastable negatively-magnetized phase to the stable positively-magnetized one. Therefore, in the small-$h$ limit, the percolation critical behavior is related to the spinodal-like behavior of the magnetization, implying that the percolation time $t_c(h)$ exhibits a spinodal-like exponential dependence on $h$. The existence of percolation transitions is likely a generic phenomenon at
magnetic Ising-like FOTs. For example, we observe an analogous behavior in dynamic (Kibble-Zurek-like) protocols entailing slow variations of the
magnetic field driving the crossing of the FOT line.
Read the full abstract ▾
Quantum Physics
Improving the Stability of the Hierarchical Equations of Motion for Open Quantum Systems with Strong Coupling to Structured Bosonic Baths
Salvatore Gatto, Samuel L. Rudge, Bokang Hou et al. (+2 more)
The hierarchical equations of motion (HEOM) approach is one of the most powerful methods avail- able to simulate the dynamics of open
quantum systems. Although the HEOM approach has been shown to be stable and efficient in many parameter regimes and physical scenarios, it is also well known that the unavoidable finite truncation of the hierarchy can lead to fundamental numeri- cal instabilities for bosonic environments. In this work, we apply recently developed techniques to analyze these issues. We use an auxiliary Fock-space picture of the hierarchy to first identify the unbalanced hierarchy-raising terms responsible for the strong transient amplification associated with the non-normal algebraic structure of the HEOM generator. We then apply a non-
unitary similarity transformation that converts these terms into balanced combinations of hierarchy-raising and hierarchy-lowering terms, thereby suppressing their numerical impact. We first demonstrate the efficacy of this approach via a
spin-
boson model with a simple Brownian oscillator spectral den- sity and subsequently extend the analysis to a structured spectral density, showing for both cases that the transformed HEOM approach allows simulations of much larger system-bath coupling than would otherwise be possible.
Read the full abstract ▾
All authors Salvatore Gatto, Samuel L. Rudge, Bokang Hou, Eran Rabani, Michael Thoss
Machine Learning
Universal Approximation of Measure-to-Measure Operators by Pushforwards
Takashi Furuya, Nicholas H. Nelsen and Frank Cole
Many learning tasks map an input distribution to an output distribution. A natural way to model such an operator is to transform each input sample using a continuous function that may depend on the entire input distribution, and then take the distribution of the transformed samples. This defines a measure-dependent pushforward model and includes measure-theoretic formulations of
transformers. We ask when such models can approximate arbitrary continuous operators between spaces of probability measures. We first show that universal approximation fails when atomic inputs are allowed: some continuous measure-to-measure operators that split or redistribute atomic mass cannot be approximated arbitrarily well by deterministic pushforward models. We then introduce the uniform level set condition, which requires a continuous measure-dependent scalarization whose shrinking level set neighborhoods carry uniformly vanishing mass over the input family. This condition is satisfied, in particular, by compact families of absolutely continuous measures. On every compact family satisfying this condition, we prove that any continuous measure-to-measure operator with outputs of finite $p$-th moment can be uniformly approximated, in the $p$-Wasserstein distance, by continuous measure-dependent pushforwards. Combining our
theorem with existing approximation results for measure-dependent in-context maps yields universal approximation by measure-theoretic
transformers. We also extend the framework to continuously-varying source measures, yielding a corresponding universality result for a class of pushforward models that are closely aligned with cross-
attention architectures.
Read the full abstract ▾
Robotics
ForVis: An In-Field Dataset and Benchmark for VIO Using Under-Canopy UAV Flights in Forests
Arman Kiani, Masoud Ataei, Elvis Gyaase et al. (+4 more)
Visual-inertial Simultaneous Localization and Mapping (VI-SLAM) for
UAVs remains difficult to evaluate in real forest environments, where motion, illumination changes, repetitive vegetation, and vibration can all affect estimation. We present ForVis, an in-field
dataset and
benchmark for evaluating VI-SLAM during
UAV flight in forest environments. The
dataset contains twelve flights across open meadow, above-canopy, and under-canopy conditions in each environment. In total, it provides 563.8s of flight over 1096.8m of trajectory, recorded simultaneously with an Intel RealSense D435i and an OAK-D Pro Wide together with inertial and flight-controller data. We
benchmark seven open-source VI-SLAM systems over 504 runs. The results show that sensor choice has a larger effect on trajectory error than the spread between
algorithms: all seven methods achieve lower median error on the OAK-D Pro than on the D435i. ForVis is intended to support evaluation of speed, accuracy and robustness for VI-SLAM in challenging forest flight.
Read the full abstract ▾
All authors Arman Kiani, Masoud Ataei, Elvis Gyaase, Jeffrey Eiyike, Aaron Weiskittel, Prabuddha Chakraborty, Vikas Dhiman
Distributed Computing
TopoEP: Topology-Aware Load Balancing for Expert-Parallel MoE Training
Jiacheng Zhu, Xie Zhao, Gongming Zhao et al. (+3 more)
Dynamic routing creates severe load imbalance in large-scale expert-parallel Mixture-of-Experts (MoE) training, turning GPUs that host hot experts into stragglers. As each MoE layer waits for its slowest rank, these stragglers prolong the expert-parallel stage and reduce overall training efficiency. Existing expert-parallelism load-balancing (EPLB) systems commonly compute load-balancing plans on the CPU, incurring device–host data transfers and cross-rank synchronization that make scheduling at every layer and microbatch expensive. Their planning formulations also overlook the hierarchical communication costs of modern scale-up and scale-out GPU clusters. We present TopoEP, a GPU-native,
topology-aware load-balancing system for large-scale MoE training. At each MoE layer and training microbatch, TopoEP converts the current routing result into hot-expert replication and
token-rerouting decisions and executes the resulting plan without data-dependent host synchronization, reducing critical-path overhead. To generate these decisions, TopoEP uses a deterministic GPU solver that performs inter-node placement followed by intra-node refinement, allowing all ranks to independently produce bitwise-identical plans. On a 32-GPU NVIDIA H800 cluster, integrating TopoEP with Megatron-LM improves end-to-end training throughput by 6.2%–11.4% across three representative MoE models.
Read the full abstract ▾
All authors Jiacheng Zhu, Xie Zhao, Gongming Zhao, Hongli Xu, Yao Fei, Jin Fang
Mesoscale and Nanoscale Physics
Extracting the anyonic exchange phase from co-tunneling through an open quantum dot
Felix Puster, Matthias Thamm and Bernd Rosenow
The elementary excitations of fractional
quantum Hall states obey anyonic statistics, described by an exchange phase $\theta$ intermediate between the bosonic and fermionic values. We study an interferometer in which three
chiral edge segments are pairwise connected by three
quantum point contacts, so that, depending on the applied voltages, one of the segments acts as an open
quantum dot: its spectrum is continuous, and tunneling through it is resonant without any gate tuning. This setup realizes the co-tunneling mechanism of a recently proposed antidot interferometer in a particularly simple setting. Using a non-
equilibrium Keldysh calculation to third order in the tunneling amplitudes, we show that the interference current contains a co-tunneling contribution in which two quasiparticles are exchanged, so that for Laughlin states the exchange phase $\theta=\pi\nu$ appears as the phase shift between the Aharonov-Bohm oscillations measured in two voltage configurations. We obtain a closed analytic expression for the interference current at finite temperature and finite separation between the
quantum point contacts and show that, provided voltage-induced changes of the enclosed area are sufficiently screened, the exchange phase can be extracted from a robust phase-plateau difference. Alternatively, simultaneous measurements of the interference currents at two drains avoid this screening assumption and time-of-flight phases can be quantified analytically, allowing the exchange phase to be extracted by a controlled extrapolation with a few-percent systematic error.
Read the full abstract ▾
Robotics
Robot Tool Design from Scratch via Behavior-Aware Hierarchical Optimization
Yinghan Chen, Xiyao Tian, Yizan Dai et al. (+2 more)
The ability to design a tool for a task marks a level of intelligence beyond merely understanding, selecting, or using one. Existing methods for robotic tool design typically optimize a tool's continuous shape and action within a structure that is prescribed or generated beforehand, so the structure itself stays outside the physical optimization loop. We study task-driven tool design from scratch, where tool structure, shape, and action are all derived from the desired physical outcome. Here we show that the three elements can be designed jointly by HOT, a hierarchical optimization whose upper level searches over discrete tool structures with BASS, while lower-level physical optimization evaluates their task behavior and returns milestone progress as behavioral evidence for the search, ultimately providing jointly optimized shape and action. On four tool-use tasks with distinct physical functions, HOT discovers functional structures after evaluating only a small fraction of search spaces containing up to 56 million structures, and the subsequent refinement of their geometry lowers the task loss on all tasks while preserving success, through deformations that are functionally interpretable. Once 3D printed, the tools accomplish all tasks on a real
robot with the actions found in simulation. Designing tools from required physical effects, rather than a catalog of known tools, is a step toward the open-ended tool making seen in humans and animals.
Read the full abstract ▾
All authors Yinghan Chen, Xiyao Tian, Yizan Dai, Yuyang Li, Yixin Zhu
Hardware Architecture
Analog Computing revisited: A fully analog and minimalistic Damage Detector for Ultrasonic Testing enabling Material-Integrated Structural Health Monitoring
Stefan Bosse
Ultrasonic Testing (UT) is commonly used to detect damage in structures, e.g., metal plates. A sensor acquires Ultrasonic waves, e.g., by using PZT transducers. The time-resolved sensor signal must be processed with analog electronics, e.g., amplified and filtered. Commonly a digitalization follows using an Analog-to-Digital converter, finally processing the digital sensor signal, applying digital signal processing, feature extraction, and Machine Learning by using powerful microprocessor systems. The disadvantages of digital processing systems are their high number of transistors (microchip area), energy consumption, state-dependent processing and therefore sensitivity to energy supply interruption. Beyond silicon electronics, printed organic electronics gains interest. But printed electronics is still limited to low transistor and electronic component counts (typically 100). We will investigate and demonstrate a fully analog signal processing and feature extraction system consisting of an analog
Hilbert transform deriving the signal envelope, simple analog arithmetic calculations for feature extraction, and finally damage classification and regression using an analog Artificial Neural Network. We expect a full damage detection system with less than 100 transistors. We will test our damage detection system with PZT transducer signals from Steel plates with circular defects. The focus of this work is the analog computation of the signal envelope (using all-pass filter networks for approximation of the
Hilbert transform) and the analog feature extraction as well as the prediction of damage, forming an analog computer which can perform in-sensor computation, computing without a digital computer.
Read the full abstract ▾
Optics
Plasmon-Enhanced Second-Harmonic Generation in Atomically Thin Crystalline Silver Nanostructures
Saad Abdullah, Philipp K. Jenke, Andrew P. Weber et al. (+8 more)
The intrinsically weak nonlinear optical response of existing materials, further constrained by
symmetry-forbidden second-order processes in centrosymmetric media, severely limits efficient frequency conversion in deeply subwavelength, ultrathin volumes. Addressing this challenge is crucial for the development of nonlinear nanophotonics. Here, we show that atomically thin, epitaxially grown crystalline silver films circumvent these restrictions through the interplay of vertical electronic
quantum confinement and lateral plasmonic enhancement. We fabricate atomically thin films that exhibit an enhanced nonlinear response associated with electronic
quantum wells, and subsequently pattern them into periodic nanoribbon and nanotriangle arrays sustaining infrared localized surface plasmon resonances. Strong near-field confinement in these structures further boosts second-harmonic generation compared to unpatterned films. Precise control over nanostructure geometry enables spectral tuning of the plasmonic resonance, and consequently, the enhanced harmonic frequency. Our findings establish an approach for activating robust second-order nonlinearities in
quantum-confined metals, where intrinsic size effects and plasmonic resonances act synergistically. The compatibility of high-quality epitaxial growth with microchip fabrication technology offers a scalable route toward ultracompact nonlinear optical components for on-chip frequency conversion, sensing, and
quantum photonic applications.
Read the full abstract ▾
All authors Saad Abdullah, Philipp K. Jenke, Andrew P. Weber, Álvaro Rodríguez Echarri, Vahagn Mkhitaryan, Fadil Iyikanat, Frederik Schiller, Philip Walther, J. Enrique Ortega, Lee A. Rozema, F. Javier García de Abajo
Robotics
CoBrush: A Hierarchical Planning Framework for Human-Robot Co-Painting
Dantong Qin, Yike Guo, Qinlin Liu et al. (+2 more)
Embodied co-painting requires a
robot to repeatedly update a shared physical canvas while human intent evolves over interaction. Existing reference-driven painters or reactive assistants are typically optimized for single-shot rendering or sketch completion, limiting their ability to sustain coherent multi-round collaboration or to construct complex, content-rich scenes over time. We present CoBrush, a hierarchical framework that formulates multi-round co-painting as a coordinated semantic, spatial, and execution process. By separating high-level intent inference from spatial grounding and stroke-level control, the system supports progressive scene development on real acrylic canvases. We evaluate the framework through real human-
robot painting sessions, stress tests, and user studies. Compared to single-turn baselines, our approach achieves stronger semantic
alignment, more stable spatial progression, and higher perceived plausibility of
robot actions. These results demonstrate that structured multi-stage
reasoning improves the coherence and robustness of interactive painting and supports the progressive development of content-rich physical artworks.
Read the full abstract ▾
All authors Dantong Qin, Yike Guo, Qinlin Liu, Alessandro Bozzon, Pan Wang
Computation and Language
Spontaneous Context Restoration: How Language Models Recover from Corrupted Inputs
Pranjal Garg and Jacob Beck
Language models sometimes produce correct outputs even when their inputs are corrupted by deletion, replacement, or misspelling. We study the internal processes accompanying this behavior, which we call context restoration, in controlled
attention-only
transformers and five pretrained LLMs (1B-32B parameters) across arithmetic, reading comprehension, and multiple-choice
reasoning tasks. In the
attention-only
transformers, restoration emerges spontaneously despite training exclusively on clean sequences, without corruption training or an explicit denoising objective. We find that context restoration follows a two-phase process: early layers localize effects associated with repair at corrupted positions, while later layers accumulate these effects at uncorrupted positions through the residual stream and ultimately concentrate them at the output position. Repair outcome is predictable from hidden states: cosine
alignment with the clean state is highly predictive in
attention-only models, while linear probes recover additional information in pretrained LLMs. A linear probe using only the corrupted
prompt's first-block hidden state predicts failure with mean ROC-AUC 0.78. This enables failure triage under matched or even partially shifted deployment conditions and may reduce unnecessary verification or computation. Failed examples also show substantially greater nonlinearity along corruption directions. Moderate-corruption finetuning increases corruption tolerance while simultaneously reducing displacement-normalized linearization error, associating improved robustness with a more nearly linear response to corruption.
Read the full abstract ▾
Complex Variables
Strongly linearly convex exhaustion of a class of $\mathbb{C}$-convex domains
Naveen Gupta
Let $D\subseteq \mathbb{C}^n$ be a bounded $\mathbb{C}$-convex domain with $C^1$ boundary whose outward unit normal admits a modulus of continuity $\omega$ satisfying $\lim_{t\to 0^+}\dfrac{\omega(t)}{\sqrt t}$. We prove that $D$ admits an increasing exhaustion by bounded $C^{\infty}$ strongly linearly convex domains. This, in particular, answers a question posed by Azinberg \cite{azin} in affirmative for a class of domains $D$.
Read the full abstract ▾
Computer Vision
Handwritten Text Recognition Lives in the High-Pixel Variance Subspace
Carlos Garrido-Munoz and Jorge Calvo-Zaragoza
In self-supervised pretraining for Handwritten Text Recognition (HTR), pixel reconstruction methods outperform contrastive methods, unlike in natural-image classification. We argue that this difference follows from where discriminative signal lies in pixel space: for HTR, it is concentrated in high-variance directions and largely absent from low-variance ones. This predicts that objectives preserving high-variance pixel content will transfer best. We test six SSL methods from three families (pixel-grounded MIM, JEPA, and contrastive) under matched encoder, data, and evaluation protocols on six handwriting
benchmarks across five
languages. With full labels, pixel-groundrounded SSL achieves the lowest CER on every
benchmark and both frozen probes, exposes per-position character information that other families recover only through the readout, and is the only family to benefit from pretraining on real handwriting. Pixel-grounded representations are also more label efficient. Across
datasets, encoder
alignment with the high-variance pixel subspace predicts CER within every method. With a pretrained LLM decoder, a frozen pixel-grounded encoder is competitive with fully fine-tuned supervised baselines; full fine-tuning achieves the lowest mean CER and ranks first or second on every
benchmark. These results show that the value of pixel reconstruction depends on where discriminative signal lies in the input.
Read the full abstract ▾
Artificial Intelligence
Why Deterministic PRM Guidance Underperforms in Discrete Diffusion Reasoning
Yan Zhan, Shaobo Liu and Zhijun Gao
Discrete
diffusion language models (dLLMs) expose a denoised solution at every step, which makes process reward model (PRM) guidance look like a way to spend compute at test time. We show that once denoising, PRM scoring, and outcome reward model (ORM) scoring are charged in the same budget of forward passes, its deterministic form loses to a much simpler baseline. Our PRMs score intermediate denoising states and are trained on the correctness of the final answer. On Dream-v0-Instruct-7B with 8 candidates per GSM8K problem, keeping the candidate with the highest PRM score at every scoring step reaches 65.18%, while independent sampling plus an ORM reranker trained for the task reaches 75.13%. The gap grows to 12.69 percentage points (pp) with 32 candidates, and is 9.85 pp on MATH and 12.16 pp on MBPP. We trace it to two separable failures. First, guidance prunes on a weak signal: on GSM8K, PRM ROC-AUC falls from 0.77 to 0.54 as the mask ratio rises, a decay that persists when states are relabeled with fresh rollouts, and
pruning lowers the best accuracy reachable from the candidate pool from 81.05% for independent samples to 67.30%. Second, on GSM8K and MATH, the PRM is a poor final judge: a sequential Monte Carlo sampler at the same budget restores that ceiling to 77.89%, yet selecting with the PRM gives 65.48%, on par with deterministic guidance, while a PRM retrained on final states matches the ORM on identical candidates. MBPP separates the two: there the PRM reaches 65.47% when reranking finished programs, on par with the ORM, but 50.88% when it guides denoising. The results point to two targets for dLLM guidance: keep correct partial solutions alive through early denoising, and leave the final choice to a verifier trained on final states. We release the corpus of denoising states with outcome labels and evaluation toolkit for reproducible comparisons at matched compute.
Read the full abstract ▾
Astrophysics of Galaxies
Diverse Metallicities in Protoclusters at $z \sim 3.4$ detected by Deep Learning
Yoshihiro Takeda, Nobunari Kashikawa, Kei Ito et al. (+12 more)
We present spectroscopic follow-up observations with Subaru/MOIRCS of four protocluster candidate regions identified by the deep-learning model PCFNet. In three candidate regions, we confirm four protoclusters at $z =$ 3.316, 3.403, 3.408 and 3.425, exhibiting
galaxy density peaks over five times the uniform distribution. The detection rate of confirmed members aligns with PCFNet's predictions, underscoring its effectiveness. Analysis of spectral energy distributions (SEDs), incorporating up to ten photometric bands $(uu^*grizyJHK)$, reveals that protocluster members lie on the
star-formation main sequence. The masses of the most massive
halos in the protoclusters are estimated from the most massive
stellar mass of the member
galaxies, yielding $\log M_\mathrm{halo}/M_\odot=12.4$-$12.8$, which are consistent with those at this
redshift predicted by cosmological simulations and are comparable to those of the progenitors of typical local clusters ($M_\mathrm{halo}^{z=0}\sim10^{14}M_\odot$). Using the Voronoi Tessellation Monte Carlo technique based on the probability of protocluster membership from PCFNet, we find that our protoclusters exhibit a significantly higher SFRD than the
cosmic SFRD at $z\sim3.4$. Based on the [Oiii]/H$\beta$ emission line ratio, two protoclusters show marginally ($1.4$-$2\sigma$) higher average
metallicities than the field, while the other protoclusters are statistically consistent with the field. These diverse
metallicities highlight the environmental impact within protoclusters detected by PCFNet, likely reflecting the complex opposing interplay between metal-rich gas recycling and metal-poor gas inflow, which may depend on the surrounding environment of each protocluster. This work demonstrates PCFNet's capability to unveil diverse samples of high-
redshift protoclusters, paving the way for a more comprehensive understanding of their role in
galaxy evolution.
Read the full abstract ▾
All authors Yoshihiro Takeda, Nobunari Kashikawa, Kei Ito, Jun Toshikawa, Mariko Kubo, Rieko Momose, Yongming Liang, Takehiro Yoshioka, Junya Arita, Hiroki Hoshi, Shunta Shimizu, Hisakazu Uchiyama, Satoshi Kikuta, Ryo Emori, Kentaro Koretomo
General Economics
Representation Risk in Pretrained Image Encoders
Ardyn Nordstrom, Morgan Nordstrom, Vamuyan Sesay et al. (+1 more)
Applied researchers increasingly convert images into features with pretrained encoders, then use those features in a downstream prediction model. The encoder is often treated as an implementation detail. We show that it can instead be a consequential source of model uncertainty. We call this uncertainty representation risk: plausible pretrained encoders map the same images into different feature spaces and can yield sharply different out-of-sample conclusions from predictive performance. We compare ten modern and legacy frozen encoders across applications involving house prices, racehorse performance, breast-cancer histology, chest radiographs, continuous facial age, and rice disease. With common dimension control, heads, and group-safe splits, validation selects SigLIP 2 for houses, raising test $R^2$ from 0.396 for ResNet50 to 0.629, and DINOv2 for horses, raising $R^2$ from 0.029 to 0.105. No encoder is best in every task. Candidate procedures are constructed using training data and compared on a separate validation partition. The selected procedure reaches 0.658 for houses and 0.979 accuracy for pneumonia. Fixed-split gains are small for horses and rice, while repeated partitions reveal instability in horse feature union. Continuous age selects SigLIP 2 at 4.786 years MAE. The principal representation gaps persist with neural heads, similarly sized DINOv2 and ViT models, and limited adaptation. These results support a simple workflow:
benchmark plausible representations, select on locked validation data, combine only when separate validation evidence justifies the additional cost, and report paired and split-level uncertainty. We implement this workflow in LOOKAGAIN-ML, the software package used to conduct the analyses in this paper.
Read the full abstract ▾
All authors Ardyn Nordstrom, Morgan Nordstrom, Vamuyan Sesay, Matthew D. Webb
Robotics
Rethinking Causal Action Tokenization with Conditional Annealing in Flow Matching
Chenyu Zhang, Yuhang Cao, Daru Du et al. (+8 more)
Autoregressive
Vision-
Language-Action (VLA) models offer a scalable path to
robot learning, yet existing action tokenizers treat tokenization as a compression problem, producing representations that are semantically misaligned with the autoregressive backbone. We propose CATok, a causal action tokenizer that reframes tokenization as a causally structured generative process. CATok introduces a conditional annealing mechanism that extracts action
tokens by progressively annealing a flow-matching process: each
token is conditioned on all preceding
tokens and encodes the residual reconstruction signal at a specific noise level, establishing a coarse-to-fine causal
token space whose generative semantics are structurally aligned with autoregressive modeling. A
token-conditioned flow-matching decoder built on
Multimodal Diffusion Transformer (MMDiT) reconstructs continuous action chunks from these discrete
tokens with the precision of hybrid
diffusion-head architectures. This discrete bottleneck enforces knowledge insulation by design, cleanly separating high-level semantic
reasoning from low-level motor execution without requiring explicit
attention masking. Extensive evaluations across three simulation
benchmarks and real-world robotic manipulation tasks demonstrate that CATok consistently surpasses existing tokenization methods in both reconstruction fidelity-compression tradeoff and inference efficiency, while improving VLA task success rate and training efficiency, establishing a high-performance, scalable foundation for purely autoregressive VLA systems.
Read the full abstract ▾
All authors Chenyu Zhang, Yuhang Cao, Daru Du, Yingxi Lu, Jing Shao, Ruoqu Chen, Jiajun Liu, Liu Cao, Yicheng Liu, Hang Zhao, Mengdi Xu
High Energy Physics (Phenomenology)
Exploring $U_A(1)$ symmetry restoration in a magnetic field by meson screening masses in the NJL model
Bing-Kai Sheng, Zigeng Ding, Danning Li et al. (+2 more)
We investigate the effective axial $U_A(1)$
symmetry restoration at finite temperature in an external
magnetic field using
meson screening masses, as well as
meson susceptibilities, in a $(2+1)$-
flavor Nambu-Jona-Lasinio (NJL) model. To incorporate inverse
magnetic catalysis, we employ a
magnetic-field-dependent four-
quark coupling constrained by the
lattice QCD
chiral pseudo-critical temperature, with the 't Hooft coupling fixed. In the
lattice-improved NJL model, the longitudinal screening mass difference between the $U_A(1)$ partners $\delta^0$ and $\pi^0$ and their normalized
meson susceptibility difference exhibit axial
magnetic catalysis at low temperatures and axial inverse
magnetic catalysis near the
chiral crossover, whereas the transverse screening mass difference increases with $eB$ at all temperatures. Despite these different responses, the axial pseudo-critical temperatures extracted from the thermal inflection points of these three alternative order parameters decrease with $eB$, implying that a decreasing pseudo-critical temperature can coexist with an increasing transverse screening mass difference at fixed temperature.
Read the full abstract ▾
All authors Bing-Kai Sheng, Zigeng Ding, Danning Li, Xinyang Wang, Lang Yu
Algebraic Geometry
Quasi-$F$-split primitive symplectic varieties in positive characteristic
Haitao Zou
Let $X$ be the good reduction of a projective hyperkähler variety of dimension $2n\geq4$. We prove that $X$ is quasi-$F$-split if and only if it is Frobenius split, equivalently if $\operatorname{H}^2_{\operatorname{crys}}(X/W)[1/p]$ has a slope-zero part. Thus its quasi-$F$-split height is $1$ or $\infty$. The proof combines a Verbitsky slope comparison with a Witt–Euler identity and requires no crystalline torsion-freeness. The same dichotomy holds for primitive
symplectic varieties in characteristic $p$, and Hodge-goodness is open in smooth proper families. Hodge-deformations of
Hilbert schemes $S^{[n]}$ of $K3$ surfaces ($p>n$) and generalised Kummer varieties $K_n(A)$ ($p>n+1$) remain primitive
symplectic, with torsion-free crystalline
cohomology and unobstructed mixed-characteristic formal deformations.
Read the full abstract ▾