Machine Learning
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Sunday 13 September
On-policy distillation (OPD) has become a standard approach for transferring capabilities from large teachers to compact students. Its cost, however, is dominated by autoregressive student rollouts and scales poorly in multi-turn agentic settings. Existing acceleration methods truncate or relocate the supervision signal according to fixed, offline budgets, despite substantial variation in teacher-signal reliability both within and across trajectories. Our empirical analysis on $\tau^2$-bench reveals a clear structure in this variation: informative supervision is concentrated in the prefix of each turn, and, most importantly for multi-turn agentic training, the cross-turn loss of teacher endorsement is temporally locked to the student's first erroneous action rather than accumulating gradually over turns. Building on these findings, we propose STRIDE (Stop-and-Restart on-policy Distillation acceleration), which combines two complementary techniques: adaptive early stopping, which terminates a rollout once the cumulative teacher log-probability falls below an out-of-distribution threshold, and a prefix buffer, which caches high-quality prefixes and restarts generation at the weakest correct turn. Together, these mechanisms induce a data-driven curriculum that progressively extends coverage to later turns. On $\tau^2$-bench retail, our method matches full-trajectory OPD and exceeds the 30B teacher at a $3.73\times$ speedup, surpasses the baseline itself at $2.34\times$, and retains a $4.51\times$ speedup under cross-domain multi-teacher training. As a supplementary generalization test beyond the agentic setting, STRIDE outperforms full OPD on AIME 2025 at a $5.10\times$ speedup and on AIME 2024 at a $3.08\times$ speedup; averaged across the two evaluations, both fixed-budget truncation baselines remain below full OPD.
Functional Analysis
Commuting analytic and co-analytic Toeplitz operators on the harmonic Bergman space
By constructing an appropriate quadratic form, we prove that if an analytic Toeplitz operator commutes with a co-analytic Toeplitz operator on the harmonic Bergman space, then at least one of the symbols is constant. This resolves an open question posed by Choe and Lee in 1999.
Computer Vision
SCOUT-SLAM: Structurally-Coupled Dual Uncertainty-Aware 3DGS SLAM in the Wild
Recently, 3D Gaussian Splatting SLAM (3DGS-SLAM) has gained significant momentum in simultaneous localization and 3DGS scene reconstruction. In real-world scenarios with rapid camera motion and cluttered dynamic environments, existing methods rely on the stability of the underlying scene reconstruction to model uncertainty. This leads to a circular dependency between camera tracking accuracy and reconstruction quality: reconstruction instabilities degrade uncertainty modeling, which affects accurate camera tracking and static scene reconstruction. To address this, the paper proposes SCOUT-SLAM, a structurally-coupled dual-uncertainty framework in which both uncertainties are estimated from a shared base network. A low-rank adaptation of this network, trained on multi-view feature consistency, estimates a tracking uncertainty that does not depend solely on the reconstruction quality. A spatially-adaptive prior modulates the network's training objective so that reconstruction instability does not inflate uncertainty on static regions, keeping the shared representation intact for both branches. Evaluations on dynamic benchmarks (TUM RGB-D, Bonn Dynamic, Wild-SLAM MoCap) demonstrate that SCOUT-SLAM achieves state-of-the-art camera tracking accuracy and artifact-free static scene reconstruction.
Read the full abstract
Robotics
REVOLVE: An Automated Closed-Loop Framework for Evolving Robot Manipulation with Minimal Human Intervention
Recent advances in data-driven robot manipulation policies have substantially improved task execution and generalization. However, real-world deployment still relies heavily on humans for failure assessment, correction, and environment reset, while models often fail to continually learn from failures and corrective experience. We present REVOLVE (Robot Evolving via Orchestrated Loops, Verification, and Experience), an automated closed-loop framework for evolving robot manipulation with minimal human intervention. Built on a unified software platform, REVOLVE integrates data collection, policy training and deployment, failure recovery, and continual learning into a single closed-loop workflow. Its Automated Reset and Correction (ARC) architecture automatically resets the environment and intervenes to correct policy failures. Dual-Loop Evolution (DLE) continually improves the manipulation policy and agent by feeding real-world interaction and failure–correction data back into policy learning and using an external mismatch memory to refine agent judgments. Experiments across four real-world manipulation tasks show that, after five iterations, REVOLVE improves average policy success rate by 18.5% and agent judgment accuracy by 8.5%, while reducing human effort in data collection and deployment testing by 94.4% and 95.1%, respectively. These results demonstrate that REVOLVE transforms real-world deployment into a closed-loop learning process that continually accumulates and uses execution experience, enabling continual evolution of both the policy and supervisory model with substantially less human intervention.
Read the full abstract
Quantum Physics
Linear equations mod $n$ are pseudo-telepathic
We prove that the quantum monad in dimension $2n$ admits no natural transformation to the polymorphism clone of linear equations modulo $n$. Consequently, for every $n\geq 2$, there exists an unsatisfiable system of linear equations over $\mathbb{Z}_n$ whose constraint system game admits a perfect finite-dimensional quantum strategy. As a corollary, we completely characterise pseudo-telepathic constraint languages in finite dimension. The proof combines a result of Harding, Jager, and Smith on group-valued measures on subspaces of Hilbert spaces with the polymorphism-minion characterisation of quantum pseudo-telepathy.
Read the full abstract
Cryptography and Security
LLM Agent Capabilities Should Follow Task Intent and Context Source
LLM agents take real actions, including executing code, modifying files, calling services, and delegating tasks, driven by context sources: user requests, tool results, documents, shell outputs, Skill and MCP instructions, memory. Unlike traditional systems, where capability is predefined, the least-privilege capability an agent needs is dynamic, depending on its task intent: what it wants to do and how. This creates a security and safety challenge: all inputs enter one shared planning channel with equal influence, by adversarial injection or accidental scope widening, yet a user's explicit request and a document's extracted text carry different trust and should not share authority to choose a destination or widen access. Existing defenses constrain operations and information flow; we study task-scoped, multi-source authority composition. We argue that agent capabilities should be scoped to the current task intent, not a sandbox or session lifetime, and that no single context source defines a complete capability. IntentCap composes capabilities from four such sources: user intent, workflow instructions, tool schemas, and runtime environment, with field-level ownership and monotonic narrowing. Each source contributes specific fields, none can fill another's, and the lease only narrows the user's authorized authority, never widens it. IntentCap uses an LLM to generate short-lived leases from these sources, validated by a deterministic checker before any side effect commits and enforced by tool- and OS-level information flow policies. Evaluation shows IntentCap blocks tested violations without rejecting benign actions, each source boundary is independently necessary, and the checker generalizes across tool, execution, placement, and delegation boundaries.
Read the full abstract
Combinatorics
Subset-Sum Density Realization in Locally Finite Abelian Groups
Let $G$ be a countable locally finite abelian group, and let \[ G_1\leq G_2\leq\cdots, \qquad \bigcup_{i\geq1}G_i=G, \] be any filtration of $G$ by finite subgroups. For $A\subseteq G$, let $\mathcal P(A)$ denote the set of all finite subset sums of elements of $A$, and let $2G:=\{2g:g\in G\}$. We prove that $|2G|=\infty$ if and only if for every filtration and every interval $[\alpha,\beta]\subseteq[0,1]$, there exists $A\subseteq G$ such that the set of limit points of \[ \left( \frac{|\mathcal P(A)\cap G_i|}{|G_i|} \right)_{i\geq1} \] is exactly $[\alpha,\beta]$. We also prove a positive-density result that does not require $|2G|=\infty$: if $G$ is infinite and $\mathcal P(A)$ has positive upper density along the given filtration, then there is an infinite set $B\subseteq \mathcal P(A)$ such that $B+B\subseteq \mathcal P(A)$. Combining this with the realization theorem, we show that, when $|2G|=\infty$, every interval $[\alpha,\beta]$ with $0\leq\alpha\leq\beta\leq1$ and $\beta>0$ can be realized by a set $A$ with this additional property.
Read the full abstract
Signal Processing
Rate-Splitting for Service-Aware Multi-Orbit Orchestration in 6G TN-NTN Networks
Integrated terrestrial–non-terrestrial networks (TN–NTNs) are a key enabler of sixth-generation (6G) connectivity across terrestrial, geostationary Earth orbit (GEO), medium Earth orbit (MEO), and low Earth orbit (LEO) domains. However, existing NTN frameworks do not jointly address service-aware serving-domain selection, multi-orbit coordination, and hierarchical multi-stream transmission. This paper proposes a service-aware multi-orbit rate-splitting multiple access (SA-MO-RSMA) framework that integrates global-common, orbit-common, and user-private streams with service-intent-aware serving-domain selection and resource orchestration. The framework operates as a system-level orchestration layer over existing 3GPP NTN architectures. A multi-criteria Orbit–Service Utility Function (OSUF) is introduced, together with metrics for service continuity, switching stability, power-normalized transmission efficiency, and synchronization robustness. Numerical results show that SA-MO-RSMA improves average per-user service utility, service continuity, switching stability, and transmission efficiency over the considered benchmark schemes, demonstrating its potential for coordinated service-aware multi-orbit TN–NTN operation.
Read the full abstract
High Energy Physics (Phenomenology)
Interpretation of the excited $\Omega_c$ family via mass and width in the chiral quark model
In this work, the mass and decay properties for low-lying orbital excitations of $\Omega_c$ baryons are investigated within the chiral quark model combined with the $^3P_0$ decay mechanism. Following the LHCb Collaboration's observations, we extend our previous study of the $\Xi_c^{\prime}$ system to the $\Omega_c$ system, with all model parameters inherited from our earlier work. Our results consistently describe the observed $\Omega_c$ spectrum: $\Omega_c(3000)$, $\Omega_c(3050)$, $\Omega_c(3065)$, and $\Omega_c(3090)$ are well described as $\lambda$-mode $1P$ states with $J^P=1/2^-$, $3/2^-$, $3/2^-$, and $5/2^-$, respectively. $\Omega_c(3119)$ cannot be described as a pure three-quark excitation. Furthermore, the $\Omega_c(3185)$ and $\Omega_c(3327)$ may be assigned to the $2S$ and $1D$ states, respectively. These results offer a coherent picture of the low-lying $\Omega_c$ spectrum and motivate further experimental tests.
Read the full abstract
High Energy Physics (Theory)
Emergent Einstein-Cartan gravity from a spinor loop
In Einstein-Cartan spinor gravity under the irreversible vierbein postulate, only the spinor among the first-order variables has a kinetic term of its own at tree level. Any nonzero background vierbein breaks the local-Lorentz (LL) symmetry spontaneously. Our previous work showed that the spinor loop generates the kinetic and mass terms of the LL gauge field, manifesting LL as a hidden local symmetry. Here we show that the spinor loop also generates the vierbein two-point function, while respecting the background gauge invariance and satisfying the Ward-Takahashi identities of the true gauge invariance, for both the general-coordinate and LL symmetries. The scheme-independent logarithmic divergence assembles the familiar induced-gravity ingredients: a cosmological constant, an Einstein-Hilbert and a Weyl-squared term for the metric sector, and a mass and a higher-derivative kinetic term for the totally antisymmetric torsion. In the vierbein sector, this torsion is carried by the derivative of the antisymmetric part of the linear vierbein fluctuation. This part is the would-be Nambu-Goldstone boson of the spontaneously broken LL symmetry, and its couplings in the minimally coupled action are all fixed by general-coordinate and LL gauge invariance. The five induced terms form a single effective action, manifestly invariant under both symmetries, completing the vierbein sector of the hidden-local-Lorentz emergent-gravity scenario.
Read the full abstract
Quantum Physics
Derivatives of Quantum Randomness: Separating Pseudorandom Unitaries from Pseudorandom (Function-like) States
Quantum computation gives rise to new pseudorandom primitives for states and unitaries, including pseudorandom state generators (PRSGs), pseudorandom function-like state generators (PRFSGs), and pseudorandom unitaries (PRUs). In this paper, we show a full unitary oracle separation between PRFSGs and PRUs. The separation holds between the strongest state notion and the weakest unitary notion: even adaptively secure, quantum-accessible PRFSGs do not imply non-adaptively secure, forward-only PRUs, even when their implementations are allowed to be non-unitary and use an arbitrary number of ancillary qubits. This reveals a fundamental distinction between pseudorandomness for quantum states and for quantum unitaries. Our main technical idea is to view a candidate PRU construction with access to state generation oracles as a map from the underlying oracle states to implemented unitaries, and to study the derivatives of this map. These derivatives are inherently low rank, and we exploit this low-rank structure to distinguish the resulting unitaries from truly random ones. We believe this differential perspective may be useful for studying other structural questions about quantum states and unitaries.
Read the full abstract
Theoretical Economics
Designing Signals for Deterrence
States and alliances spend enormous resources signaling resolve to deter adversaries. Two canonical strategies are burning money (sunk costs) and burning bridges (audience costs). Large audience costs can deter effectively, but they may be infeasible and, even when feasible, can increase the likelihood of war relative to burning money. We propose a signaling design in which the defender retains the flexibility to fight or back down but commits ex ante, before learning its resolve, to a rule mapping realized resolve into costly actions. When audience costs are limited, this design raises defender payoffs while reducing the likelihood of war. We discuss implementation through laws and automated algorithms. We also characterize defender payoffs when audience costs are chosen endogenously from a feasible set, showing that lowering the maximum feasible audience cost can raise defender payoffs, and we compare these payoffs with those under the standard burning-money setting without commitment.
Read the full abstract
Computer Vision
TTDF: A Two-Stage Framework for Reliable Surgical Phase Transition Detection
Reliable workflow transition detection is important for context-aware surgical assistance and downstream decision support. However, online surgical phase recognizers primarily focus on frame-wise accuracy and temporal consistency, rather than the reliability of workflow transition events. Directly converting phase changes into events is unreliable: temporal jitter and workflow-illegal switches produce false or duplicate events, while persistent, workflow-consistent candidates may remain incorrect. To address this limitation, we formulate reliable workflow transition detection as a distinct event-level task operating on outputs of a frozen online phase recognizer. We propose the Two-Stage Transition Detection Framework (TTDF), a causal framework that progressively filters transition candidates. Transition Candidate Extraction (TCE) first applies a minimum-duration requirement and a workflow-graph constraint to remove false candidates caused by temporal jitter and phase transitions not allowed by the workflow graph. Specifically, a candidate is retained only if the predicted target phase persists for a minimum duration and the ordered phase pair belongs to the workflow graph's allowed transition set. TCE thereby produces a high-recall candidate set without additional training. Transition Candidate Verification (TCV) suppresses remaining false candidates using phase-posterior shifts and visual-change cues from frozen DINOv2 features. Events are assessed using a phase-pair-aware one-to-one matching protocol. Experiments on Cholec80 show that TTDF reduces false transition emissions while preserving recall and controlling decision delay.
Read the full abstract
Statistical Mechanics
Thermodynamic Realizability of Hidden Markov Processes: Attainment, Observable Certificates, and the Price of Architecture
What is the least dissipative finite Markov machine that can reproduce a given stochastic process exactly? A basic compactness worry is that a minimizing sequence might lower its cost only by sending microscopic rates to infinity, never converging to an actual finite-rate machine. We prove that this escape is impossible at fixed hidden dimension. States of vanishing stationary occupation can be traced out, and infinitely fast conductance classes can be contracted, without increasing entropy production or losing the limiting observed path law. Hence the exact-law cost $V_n(P)$ is attained and lower semicontinuous at every fixed state count, with no cap on rates or mean activity. Adding a positive price $\kappa$ per used hidden state yields an attained optimum over all finite dimensions and an exact representation by globally calibrated lower certificates built from finitely many observable path expectations. The lower envelope $G_\kappa(P)=\min_n[V_n(P)+\kappa n]$ defines an architecture phase diagram whose slope is the selected state count; divergence of that count as $\kappa\downarrow0$ is equivalent to failure of any finite machine to attain the all-dimension infimum $V_\infty(P)$. For the two-block renewal family $P_q$, we derive complete finite-dimensional phase-type cancellation equations, give an explicit four-phase counterexample to a tempting restricted parameterization, and prove strictly positive minimal dissipation at every feasible finite dimension for all $0<q<1$. These results reduce the remaining problem to one sharp question: can growing hidden architecture drive the dissipation of $P_q$ to its all-dimension infimum without any finite optimizer, and in the strongest scenario, to zero?
Read the full abstract
Optimization and Control
On the Homogenization of Norm and Time Optimal Controls for Parabolic Equations with Oscillating Coefficients
We study homogenization of norm-optimal and time-optimal controls for a one-dimensional heat equation with rapidly oscillating periodic diffusion coefficients. The controls act on a fixed subinterval and are subject to pointwise constraints in space and time. We first derive final-state observability and null-controllability estimates that are uniform with respect to the oscillation scale. Combining these estimates with periodic homogenization and an abstract theory of optimal controls for parabolic systems, we prove convergence of the norm-optimal values and controls to their homogenized counterparts. We then deduce convergence of the optimal times and time-optimal controls. The controls converge weakly-star in the space of essentially bounded functions and strongly in every Lebesgue space with finite exponent, after extension to a common time interval when necessary.
Read the full abstract
Quantum Physics
The classical capacity of generalized amplitude-damping channels and its strong-converse exponent
The classical capacity of a quantum channel generally requires regularization because input codewords may be entangled across channel uses. We remove this regularization for generalized amplitude-damping channels (GADCs) and determine their exact strong-converse exponent throughout the full parameter range. The exponent is a single-letter transform of the sandwiched Rényi Holevo information and is attained by product input codewords with collective decoding. The key ingredient is a decomposition of the joint output into positive operators whose triangular factors have identical scalar Gram matrices after norm compression. A sharp triangular Schatten-norm compression inequality then establishes maximum output norm multiplicativity for diagonally sandwiched GADCs tensored with arbitrary completely positive maps. This yields additivity of the sandwiched Rényi Holevo information at every order above one. Consequently, the classical capacity equals the known one-copy Holevo information and admits a scalar optimization. Thus, entangled input codewords improve neither the capacity nor the optimal exponential decay of decoding success above capacity.
Read the full abstract
Quantum Physics
Dynamical Readout of Measurement Statistics and Emergent Entanglement-Like States in Classical Networks
Can a classical network not only encode a quantum-like state, but also read out its measurement statistics through its own collective dynamics? Here we introduce a network-native readout scheme for a structured classical network whose community modes form an effective two-qubit state space. Network connectivity selects a dominant collective mode that encodes the state, while a fixed set of ten elementary connectivity perturbations probes its collective response. The resulting shifts of the dominant growth rate form a complete basis of expectation values for the real two-qubit sector, from which joint-outcome probabilities associated with arbitrary real projectors are reconstructed by linear combination. Once these ten spectral responses are measured, the same data generate correlations over a continuous family of measurement settings. As a benchmark, for an encoded Bell state four selected correlations give $|S|=\chsh>2$, whereas a separable product-state reference remains within the Clauser–Horne–Shimony–Holt (CHSH) bound $|S| \leq 2 $. These results establish an informationally complete dynamical readout of effective two-qubit states directly from the spectral response of a classical network.
Read the full abstract
Computer Vision
ESAFusion: LiDAR–4-D Radar Fusion via Local Geometric Complementation and Multiscale Adaptive Interaction for 3-D Object Detection
LiDAR–4-D radar fusion combines accurate spatial geometry with motion and reflectivity cues from radar, offering a promising solution for 3-D object detection in complex driving environments. However, sparse radar observations and differences in spatial sampling between the two modalities complicate reliable cross-modal complementation. Moreover, the relative importance of modalities and feature scales varies across spatial regions, making adaptive fusion challenging. To address these challenges, we propose ESAFusion, an evidence-aware and scale-adaptive framework that combines local geometric complementation with multiscale adaptive interaction. Specifically, we introduce an Evidence-Aware Radar Selection (ERS) module to suppress radar clutter using motion and observation-quality evidence while retaining foreground confidence for subsequent fusion. Then, the Pillar-Level Complementary Encoder (PCE) improves cross-modal complementation under mismatched spatial sampling using local geometric support from neighboring LiDAR pillars. We further design an Intra- and Inter-Scale Adaptive Fusion (ISAF) module to adaptively adjust the contributions of different modalities and feature scales in bird's-eye-view (BEV) space. Extensive experiments on the View-of-Delft (VoD) dataset show that ESAFusion achieves the highest mean average precision (mAP) among the compared methods, reaching 74.60% in the Entire Annotated Area and 88.89% in the Driving Corridor. It also attains the highest average precision (AP) for Cyclist among these methods in both regions while running at 19.23 FPS. Evaluations on VoD-Fog further demonstrate robustness under progressively degraded LiDAR observations. The source code will be made publicly available at https://github.com/SenJieHu549/ESAFusion.
Read the full abstract
Analysis of PDEs
A Quantitative Gidas–Ni–Nirenberg Theorem for the Fractional Laplacian
We establish a quantitative Gidas–Ni–Nirenberg theorem for the pure fractional Dirichlet problem \[ (-\Delta)^s u=k(x)g(u)\quad\text{in }B_1, \qquad u=0\quad\text{in }\mathbb R^n\setminus B_1, \qquad 0<s<1. \] For positive bounded solutions under a two-sided $L^\infty$ normalization, we prove \[ \mathcal D(u)\le C\,\mathcal D(k)^\gamma,\qquad 0<\gamma<1, \] where $\mathcal D$ measures spherical oscillation and outward radial increase. To the best of our knowledge, this provides the first quantitative Gidas–Ni–Nirenberg estimate for the pure fractional Dirichlet problem. The key new ingredient is a shift-uniform, weak Harnack inequality for shifted antisymmetric supersolutions, which overcomes the exterior-tail obstruction in the quantitative moving-plane argument. This mechanism extends to nonseparable nonlinearities.
Read the full abstract
High Energy Physics (Theory)
Monopole fission
There is a 4N-dimensional moduli space of SU(2) BPS monopoles with charge N, but introducing a Higgs potential removes this degeneracy and yields a repulsive force between monopoles. However, if a point in the BPS monopole moduli space is sufficiently symmetric, then it flows to a similar non-BPS saddle point solution once the Higgs potential is introduced. Monopole fission, into constituent monopoles with lower charges, results from symmetry breaking perturbations of such saddle point solutions. This fission is investigated numerically, by evolving the gradient flow equations of the Yang-Mills-Higgs theory, with initial conditions and symmetry breaking perturbations created using rational maps between Riemann spheres. The results resemble components of BPS monopole scattering processes obtained within the geodesic approximation for low speed monopole dynamics.
Read the full abstract
Signal Processing
Channel-Adaptive Region Adjacency Graph Carriers for Semantic Image Communication
Semantic image communication seeks to preserve task-relevant scene structure under limited channel resources, but carriers are often dense latent tensors or grid-aligned semantic layouts that do not explicitly encode region-level relations. This work introduces a segmentation-derived region adjacency graph (RAG) carrier, termed channel-adaptive RAG (CA-RAG), for joint source-channel coding-style image communication. Nodes store interpretable region attributes, edges preserve adjacency, channel-adaptive graph simplification (CGS) controls the node budget, and semantic belief propagation refines noisy graph embeddings before diffusion-based reconstruction. On Cityscapes, pre-channel RAG payloads are several times smaller than compressed class-index layouts in a 2,000-image study. Under additive white Gaussian noise at signal-to-noise ratios from 0 to 15 dB, CA-RAG reports higher semantic consistency than deep joint source-channel coding and a same-decoder layout baseline, with comparable perceptual quality. At 10 dB, the full-budget rate-sweep point reaches mean intersection over union (mIoU) = 0.329 at approximately 3.3 x 10^3 channel uses, while the default adaptive-CGS setting reports mIoU = 0.294 at approximately 2.6 x 10^3 channel uses.
Read the full abstract
Computer Vision
Open-UniMo: Towards Unified Motion-Language Understanding and Generation in the Open World
Unified motion generation and understanding is crucial for embodied AI systems that can both synthesize and interpret human actions in open-world environments. Existing motion-language models often treat motion as an auxiliary modality of a language model, leading to text-dominated representations and limited cross-modal interaction. Moreover, the next-token prediction paradigm is not naturally suited to long motion sequences, where autoregressive generation may accumulate prediction errors. To address these challenges, we propose Open-UniMo, a unified Large Motion-Language Model (LMLM) trained on million-scale open-world motion-language data. Open-UniMo promotes modality parity by extending Qwen's vocabulary of about 150K text tokens with 64K motion tokens, enabling motion and language to share a unified token space. We further introduce motion-consistent Chain-of-Thought reasoning as an intermediate representation to bridge language semantics and motion dynamics. Open-UniMo is trained with a two-stage pipeline, where supervised fine-tuning establishes CoT-guided bidirectional motion-language mapping and Group Relative Policy Optimization (GRPO) improves semantic alignment while mitigating cumulative errors in autoregressive motion-token generation. To support comprehensive evaluation, we propose Open-MoBench, a VLM-guided benchmark for assessing text-to-motion (T2M) generation, motion-to-text (M2T) understanding, and bidirectional consistency. Extensive experiments show that Open-UniMo achieves state-of-the-art performance on both conventional metrics and Open-MoBench. Furthermore, ablation studies reveal that M2T understanding is not primarily limited by motion-token vocabulary size; instead, coupling M2T with the learnable T2M generation path yields stronger cross-modal representations, demonstrating that generation can facilitate understanding in AR-based motion-language modeling.
Read the full abstract
Machine Learning
SH-WRNN: Implicit Spherical Harmonics Weight Field Routing Neural Networks for Asymmetric Edge Intelligence
Deep learning architectures remain rigidly built upon traditional fully connected layers. While networks scale up, few challenge this foundational root. In this work, we reshape this paradigm by transforming the core synapse weight matrix from static, discrete parameters into a differentiable, continuous field governed by spherical harmonics functions. We introduce the Implicit Spherical Harmonics Weight Field Routing Neural Network (SH-WRNN), which constrains weight matrices within a continuous parametric field instead of optimizing millions of localized discrete weights. When retrieving the weight matrix of the current layer, connection parameters are localized using latitude and longitude on a rectangular plane mapped from the continuous field. The latitudinal coordinate is specified by activated neurons from the previous layer, while the longitudinal coordinate is determined by keys generated from previous layer activations via matrix multiplication. By evaluating intersections on this map, the network dynamically extracts its connection weights on-the-fly. Empirical validation on MNIST demonstrates that under compact configurations of (32, 10, 10) and (32, 3, 10), SH-WRNN achieves robust accuracies of 91.05% and 81.54% within a single training epoch. Furthermore, we propose an asymmetric Surface Baking scheme. Upon convergence, the continuous weight field is baked once into a static parametric surface. By eliminating analytical spherical harmonics calculations during inference and reducing dynamic matrix extraction to high-speed localized memory slicing, this scheme achieves asymmetric algorithmic acceleration with negligible accuracy degradation. This paradigm shift bypasses GPU memory-bandwidth monopolies, opening a novel path to reshape the advantages of CPU computing. Code is available at https://github.com/jzb1111/SphericalHarmonyRoutedNeuralNetWork.
Read the full abstract
Superconductivity
Altermagnetism across the BCS-BEC crossover
We study a two-dimensional paired Fermi system in which altermagnetism produces a spin splitting that depends on momentum. The interaction strength is described through the two-body binding energy, so that the chemical potential and pairing gap evolve self-consistently from the weak-coupling Bardeen-Cooper-Schrieffer (BCS) regime to the strong-coupling Bose-Einstein-condensate (BEC) regime at fixed density. The stability of the uniform paired state is examined by giving the pairs a small center-of-mass momentum and following the resulting change in free energy. We show that at zero temperature, the phase stiffness follows a universal quadratic suppression, $J/J_0 = 1 - (\alpha/\alpha_0)^2$, across the entire fully gapped crossover regime. Deviations from this relation emerge only on the BCS side upon the opening of gapless Bogoliubov pockets, which rapidly reduce the stiffness and can trigger an instability toward finite-momentum pairing. In the BEC regime, this universal quadratic correction corresponds to the altered effective mass of the tightly bound composite bosons. The model therefore provides a simple setting in which to compare altermagnetic pair breaking on the BCS and BEC sides of the crossover.
Read the full abstract
Probability
Least-Variable Harmonic Matrix-Exponential Distributions
Concentrated matrix-exponential (CME) distributions are random clocks approximating a fixed time, with quality measured by the squared coefficient of variation (SCV); a well-known construction builds such clocks from products of cosine-squared terms under shared exponential damping, but the number of these terms, and hence the number of parameters to optimize, grows with the order. Since exhaustive search over the full parameter space becomes prohibitively expensive at high orders, previous work has relied on low-dimensional heuristic parametrizations rather than the true optimum. We introduce a strictly larger class of common-damping harmonic densities, built from an arbitrary nonnegative trigonometric polynomial rather than such a product. For each fixed pair of scalar parameters, coefficient optimization reduces to a single eigenvalue problem, leaving a two-dimensional nonlinear search independent of the order. The enlargement leaves the infimum unchanged from that of the classical cosine-squared construction. Applying this exact characterization at the orders where the three-parameter heuristic allows comparison gives smaller reported SCV values, with larger gains at higher orders. Separately, an explicit harmonic construction gives an $O(N^{-2})$ upper bound on the attainable SCV, where $N$ is the ME representation budget, compared with the Erlang distribution's linear rate $1/N$.
Read the full abstract
Algebraic Geometry
Nagata's conjecture on a polynomial automorphism in positive characteristic
An automorphism of the polynomial ring $k[x_1,\ldots ,x_n]$ over a field $k$ is said to be $\mathit{tame}$ if it can be obtained by composing affine automorphisms and elementary automorphisms, and $\mathit{wild}$ otherwise. Jung and van der Kulk showed that every automorphism of $k[x_1,x_2]$ is tame. In 1972, Nagata conjectured that a certain automorphism of $k[x_1,x_2,x_3]$ is wild. In 2003, Shestakov and Umirbaev proved this conjecture for $\mathop{\mathrm{char}}\nolimits k=0$. The purpose of this paper is to prove the conjecture for $\mathop{\mathrm{char}}\nolimits k\ge 7$. This is the first time that the existence of a wild automorphism has been confirmed in positive characteristic.
Read the full abstract
Statistics Theory
Two short proofs of incompatibility for correlation matrices
Let $P_d$ be the elliptope of $d \times d$ correlation matrices, and, for a standardised law $F$, let $S_d^F$ be the set of correlation matrices of random vectors with all margins $F$. We give two short proofs. First, an eleven-term representation of $|x|^4$ as a sum of fourth powers of linear forms on $\mathbb{R}^4$ yields $S_{11} \neq P_{11}$ for uniform margins. Second, the six diagonals of the regular icosahedron yield $S_6^F \neq P_6$ for arcsine margins. Together with a result of Devroye and Letac, this gives $S_d^F = P_d$ if and only if $d \le 5$, settling a conjecture of theirs.
Read the full abstract
Disordered Systems and Neural Networks
Why 2D Models Fail to Capture Flow Properties in 3D Heterogeneous Media: A Connectivity Perspective
We analyze how the equivalent hydraulic conductivity, $K_{eq}$, varies with the coarsening scale $\lambda/I$ in 2D and 3D heterogeneous media. Their local hydraulic conductivity, $k({\bf r})$, follows a lognormal distribution in all cases, while spatial connectivity of $k({\bf r})$ ranges from high (HCS), through intermediate or multi-Gaussian (ICS) and low (LCS), to unstructured (NS). Using a stochastic approach, we characterize the full distribution $P[\log(K_{eq})]$, and its moments: mean $\langle K_{eq}\rangle$, log-variance $\sigma_{\log(K_{eq})}^{2}$, and log-skewness $\gamma_{\log(K_{eq})}$, as a function of $\lambda/I$. Significant contrasts appear between 2D and 3D, which we interpret in terms of two key factors associated with spatial dimensionality. We observe that, compared with ICS, LCS and HCS exhibit two distinct features: existing analytical expressions for coarse-graining of $k({\bf r})$ in multi-Gaussian media do not apply, while $P[\log(K_{eq})]$ deviates from Gaussian at all scales $\lambda/I$. Critical path analysis is then used to quantify the connectivity of our samples, and show that $K_{eq}$ exhibits a power-law dependence on it. Our results reveal that connectivity can lead to substantial differences between 2D and 3D macroscopic flow properties, even for the same geostatistical parameters, highlighting the severe limitations of using 2D models to represent 3D flows.
Read the full abstract
History and Philosophy of Physics
Niels Bohr in North America (1923) and his Meeting with Robert Frost
During the autumn of 1923, Niels Bohr stayed for about three months in Canada and the United States where he gave a series of lectures on atomic and quantum theory. This paper provides details about his many meetings and lectures. It deals in particular with his stay at Amherst College, where he had a long conversation with the celebrated poet Robert Frost. Based on new archival sources, the paper includes a more fine-grained and better documented version of the episode than the one offered by literary critics and scholars. It concludes, perhaps somewhat disappointingly, that the Bohr-Frost meeting was of no great importance to either the physicist or the poet.
Read the full abstract
Statistics Theory
Unrestricted conditional maximum likelihood estimation for a Fréchet regression model
Heavy-tailed response variables are routinely modeled as a function of covariates, but the asymptotic theory available for the resulting maximum likelihood estimators sidesteps certain technical difficulties by restricting the parameters to a compact set, which is not the natural parameter space. We consider a Fréchet model whose scale parameter depends log-linearly on a vector of covariates, while the tail parameter is constant, and we study the conditional maximum likelihood estimator of the tail parameter and the regression coefficients jointly. We show that this estimator is consistent over the full, noncompact parameter space, and that it is asymptotically normal, with the inverse Fisher information as limiting covariance matrix. Consistency is obtained by a specific compactification of the parameter space relying on the cosmic closure of Euclidean space. The extension of Wald's consistency theorem needed for that step, from single observations to blocks of observations, is stated in an abstract setting and is possibly of independent interest.