Peer Revenue Publish or profit

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Monday 7 September

Machine Learning

Constrained Online Learning with Noisy Constraint Values

We study constrained online convex optimization with adversarial constraints and conditionally unbiased, finite-variance observations of constraint values and gradients. Under common feasibility, our \LEDGER\ algorithm attains $O(\sqrt T)$ expected regret and $O(\sqrt{T\log(eT)})$ expected budget violation, the largest cumulative overspend over any window. It uses a reflected exponential potential, clipped signed observations, and predictable adaptive regularization, with one feedback triple and one projection per round. Neither a Slater condition, independence between feedback channels, nor an absolute constraint-value bound is needed. A Gaussian testing lower bound proves that the budget rate has optimal horizon dependence under square-root regret at fixed positive noise, including the logarithm. The same obstruction holds for terminal violation, so the logarithm is not a cost of maximizing over windows; an $O(\sqrt T)$ budget bound instead forces linear regret. In contrast, fixed positive Gaussian value noise yields a joint regret–hard-violation lower bound of $\Omega(\min\{\sigma,1\}T/\log^2 T)$, even with exact gradients in one dimension. The hard-violation construction matches arbitrarily many moments while preserving a feasible-endpoint gap and constant endpoint probabilities. Together, the bounds separate uncertainty about hard feasibility from learnable signed budgets. Deterministic restarts remove the horizon input without changing either upper rate.
Read the full abstract

Geophysics

ANADEF: A Nested-Permutation Alarm for Dual-Parameter Earthquake Forecasting

Spatially resolved stress proxies and rate-based seismicity models are increasingly combined for regional earthquake forecasting, yet formally testing their non-redundancy remains largely unaddressed. We present the Nested-Permutation Alarm for Dual-Parameter Earthquake Forecasting (ANADEF) pipeline, integrating a stress-sensitive Gutenberg–Richter $b$-value field, estimated via a penalized 2D B-spline inversion, with a stationary background rate ($\mu$) from space–time ETAS stochastic declustering. Applied to the Zagros Fold–Thrust Belt using an 18-year catalog ($n=40{,}731$, $M_{\mathrm{N}}\geq1.5$, 2006–2024) under a two-stage protocol with non-overlapping training (2006–2014) and target (2015–2024) windows, the model achieves, for $M_w\geq5.0$ ($N=55$), a retrospective Area Skill Score $S=0.69$ (95% CI: 0.64–0.73), reducing alarmed area from $\tau\approx0.38$ ($\mu$-only) to $\tau\approx0.28$ while retaining hit rate $\nu=92.7\%$; since thresholds are calibrated on the evaluation catalog, these are in-sample, not out-of-sample, estimates. A nested permutation procedure re-optimizing thresholds within each null realization tests whether $b$-value adds information beyond $\mu$, absorbing the optimization bias. Significance is null-model dependent: cell-wise randomization yields $p=0.012$, while a conservative spatial-structure-preserving null gives weaker, non-significant evidence ($p=0.057$)—incremental stress information is suggestive but not unambiguously established. Calibrated thresholds were frozen and applied to updated 2015–2024 fields, generating an unvalidated, forward-looking spatial alarm template for 2025–2029. These results establish a reproducible, statistically transparent methodology for testing, rather than assuming, complementarity between stress-sensitive and rate-based predictors, offering a candidate operational template pending prospective validation.
Read the full abstract

Differential Geometry

Hitchin sections on $3$-dimensional Sasakian manifolds

By using non-abelian Hodge correspondence on compact Sasakian manifolds, we investigate analogous constructions of Hitchin sections of rank $2$ on $3$-dimensional compact Sasakian manifolds. We obtain canonical deformations of $\widetilde{SL_{2}(\R)}$-Sasakian structures giving a diffeomorphism from the vector space of basic quadratic differentials onto a connected component of the space of equivalent classes of sasakian structures of negative basic first Chern classes and topologically trivial CR structures equipped with an appropriate manifold structure.
Read the full abstract

Combinatorics

Removable matchings in $2$-connected graphs

A matching $M$ of a $2$-connected graph $G$ is removable if $G-M$ is $2$-connected. We prove that, for every integer $d\ge4$, every $2$-connected graph with at least $2d$ vertices and minimum degree at least $d$ has a removable matching of size $d$. This improves the previously known lower bound $d-2$ for $d\ge5$ to the optimal value $d$, answering a question of Li, Zhou, Fujita, and Mao. We also show that a $2$-connected graph on at least six vertices with minimum degree at least $3$ has a removable matching of size $3$ if and only if it is not a wheel. Our proofs use edge contraction and the structure of minimal non-removable matchings.
Read the full abstract

Classical Physics

A purely mechanical system realizing a Coulomb-like interaction

We solve in closed form a one-dimensional relativistic system: two masses interacting only through elastic collisions with a massless mediator bouncing between them. Momenta, times, and positions are hyperbolic functions of the collision index. The mediator energy, interpreted as the pair's effective potential, obeys an exact discrete Coulomb law, $V\propto 1/r$, with a Lorentz-invariant action as coupling. A massive Newtonian mediator instead transmits a $1/r^{3}$ force; one adiabatic invariant traces both laws to the mediator's dispersion relation. Continued to negative mediator energy, the closed forms turn trigonometric, binding a one-dimensional mechanical analog of the Coulomb atom.
Read the full abstract

Atomic and Molecular Clusters

Core-Level Spectroscopy Decodes Bond-Alternation Dynamics of Cyclo[18]Carbon

The advent of X-ray free-electron lasers and high-harmonic generation has made time-resolved X-ray spectroscopy a powerful tool for probing local atomic environments, yet whether localized core excitations can report on global collective distortions remains open. Cyclo[18]carbon (C$_{18}$), with its polyynic ground state (D$_\text{9h}$) and cumulenic transition state (D$_\text{18h}$), provides an ideal model to address this long-standing issue in bond-length alternation (BLA) dynamics. Mapping two-dimensional potential energy surfaces by first-principles simulations, we find that core ionization symmetrizes the ground-state double-well potential along the BLA coordinate. Our calculated X-ray spectra reveal remarkable sensitivity to bond-length variations: C1s ionization potentials vary by up to 2.4~eV across the BLA coordinate (1.1–1.4~Å), with a 0.9~eV variation for minima predicted by different functionals, while NEXAFS $\pi^*$ peaks shift by up to 4~eV across the same coordinate. These predicted signatures provide a quantitative spectroscopy–structure dictionary for decoding transient structures in future ultrafast X-ray experiments and monitoring bond-alternation dynamics in real time.
Read the full abstract

All authors Minrui Wei, Zeyu Liu, Sheng-Yu Wang, Jun-Rong Zhang, Lu Zhang, Guoyan Ge, Weijie Hua

Materials Science

Restoring the Surface Magnetic Gap in MnBi$_2$Te$_4$

A widespread experimental realization of quantized anomalous transport in the intrinsic magnetic topological insulator MnBi$_2$Te$_4$ is hindered by its elusive surface magnetic gap. Uncovering the origin of the gapless states is essential for accessing its topological properties. Here we show that surface defects lower the electrostatic potential, drive topological surface states into subsurface layers, suppress exchange interactions, thereby closing the gap. Tuning the surface electrostatic potential via external electric fields or interfacial fields in van der Waals heterostructures restores the expected gap and enables control of its topology. This is confirmed by model calculations and validated in defective MnBi$_2$Te$_4$ films interfaced with polar insulators, explaining the enhanced quantum anomalous Hall effect under AlO$_x$ capping observed in recent experiments. Our theory identifies electrostatically driven surface-state delocalization as a competitive origin of gap suppression and proposes displacement-field engineering for robust quantized transport.
Read the full abstract

Artificial Intelligence

A visual large language foundational model for medical image recognition using clinician-contributed online resources

Large language models (LLMs) have demonstrated strong capabilities across diverse domains, showing considerable potential in medicine. However, their application in medical settings remains limited by the scarcity of visual question answering (VQA) datasets that capture clinical reasoning and explicit image-text alignment. Here, we leverage de-identified medical images and expert commentaries shared through clinician-oriented online resources. By combining an advanced LLM with clinician-in-the-loop verification, we established a rigorous pipeline to construct ThoughtMed-1M, a long-form medical VQA dataset containing over one million VQA pairs and designed to capture structured clinical reasoning and medical image-text alignment. To demonstrate its utility, we developed a FOundational LLM Trained on ThoughtMed-1M (FOLTMed). FOLTMed achieved state-of-the-art performance across 42 medical VQA benchmark datasets, with a macro accuracy of 85.4 percent. It also generated more clinically coherent responses on the ThoughtMed-1M test set, outperforming state-of-the-art models by 3 to 5 percent across factuality and similarity metrics, highlighting a scalable paradigm for advancing research on clinically grounded multimodal LLMs.
Read the full abstract

All authors Lingxuan Hou, Yuhua Xie, Yue Hu, Yan Zhuang, Junqi Li, Chengzhi Xia, Binh Phu Nguyen, Abubakar Siddique, Minh Nguyen, Yao Hou, Yanju Bao, Kexin Liu, Ke Chen, Jianjun Sun, Zeqi Li, Trung Nguyen, Jiangli Lin

Rings and Algebras

$\delta$-Mock-Novikov Algebras: Structural Theory, Operad and Poisson-Type Constructions

We introduce $\delta$-mock-Novikov algebras as a parameter-dependent mock analogue of Novikov algebras. For $\delta=1$, the commutator of a mock-Novikov algebra defines a Malcev algebra. Every $\delta$-mock-Novikov algebra satisfying $\mathcal{A}^2\subseteq\operatorname{Ann}(\mathcal{A})$ is shown to be differentially special. At the operadic level, the binary quadratic operad governing $\delta$-mock-Novikov algebras is quadratically self-dual but not Koszul. We prove that finite-dimensional $\delta$-mock-Novikov algebras are nilpotent in characteristic zero and classify those of dimension at most four. Finally, we study the associated Poisson-type structures and establish relations and constructions among them via polarization, depolarization, and tensor products.
Read the full abstract

Machine Learning

From Synthetic Priors to Model Behavior: Structural Coverage in Tabular Foundation Models

Tabular foundation models (TFMs) are commonly pretrained on large collections of procedurally generated synthetic tasks, yet it remains unclear how well these synthetic pretraining priors support the downstream tasks on which the models are evaluated. We study this question from a distribution-level attribution perspective. We recover or reconstruct the synthetic data generators of four TFMs and compare their generated tasks with datasets from two widely used tabular benchmarks. Each dataset is represented by a common set of structural descriptors capturing schema, feature distributions, dependence structure, response properties, and feature–response relationships. In this space, we measure how broadly and repeatedly each synthetic prior reaches benchmark tasks using structural coverage and normalized density, and examine whether stronger local support is associated with better predictive performance. We find substantial differences across synthetic pretraining priors: some generators provide consistently broader and denser support for benchmark tasks than others. Moreover, stronger synthetic-to-benchmark support is generally associated with better relative model performance. These results suggest that structural coverage provides a useful diagnostic for characterizing synthetic pretraining priors and relating their data-generating assumptions to downstream model behavior.
Read the full abstract

All authors He Zhao, Ryan Thompson, Daniel M. Steinberg, Ashfaqur Rahman, Edwin V. Bonilla, Cheng Soon Ong

Geophysics

The 2016 Mw 7.0 Kumamoto Earthquake Sequence, Japan revisited: Insights from Spatio-Temporal Analysis of Seismicity Parameters

Understanding how crustal faults accumulate strain, nucleate ruptures, and redistribute post-seismic stress is fundamental to seismic hazard assessment. The 16 April 2016 $M_w$ 7.0 Kumamoto earthquake, rupturing the Futagawa–Hinagu fault zone in the Beppu–Shimabara graben, central Kyushu, Japan, offers a premier dataset for tracking these processes across a full earthquake cycle. Using the JMA catalog (2014–2018) with dynamic completeness estimation ($M_c \approx 2.15$–$2.25$), we examine the Gutenberg–Richter $b$-value, the 3-D hypocentral fractal dimension ($D_c$), and the seismicity-rate anomaly ($Z$-value). During the one- to two-year preparatory phase, $b$ declined from a baseline of $1.35 \pm 0.10$ to a precursory minimum of $0.59$, while $D_c$ contracted from $0.85 \pm 0.15$ to $0.70$–$0.80$, recording stress concentration and microfracture coalescence onto a narrow nucleation zone. A coherent negative $Z$-value anomaly (-1.6 to -2.0) developed along the graben, strengthening with integration time—consistent with, though not proof of, progressive fault locking. Depth-sliced volumes show the low-$b$ locked core ($b \le 0.65$) was stratified at $10$–$12.5$~km depth and sharpened within the final four months before failure. Rupture reversed this within weeks: $b$ surged to $1.25$–$1.35$ and $D_c$ expanded to $2.04$–$2.11$, consistent with coseismic stress drop and aftershock activation, followed by recovery over 2.5 years. Cross-sections reveal sharp aftershock localization alongside two unrelaxed asperities ($b \approx 0.80$–$0.95$) near the Hinagu termination and Mount Aso, consistent with positive Coulomb stress loading. This framework resolves asperity locking, release, and healing better than any single metric; however, since anomalies were identified retrospectively, they should be read as evidence of coherent behavior rather than a validated forecast tool alone.
Read the full abstract

All authors Hamzeh Mohammadigheymasi, Muhammed Hossein Mousavi, Marjan Tourani, Nasrin Tavakolizadeh

Differential Geometry

Minimal Two-Spheres and Manifolds with Positive Isotropic Curvature

We improve the Micallef–Moore index estimate for harmonic two-spheres in $n$-manifolds with positive isotropic curvature to the sharp bound $n-3$. Combining with recent work, this completes the diffeomorphism classification of closed manifolds with positive isotropic curvature in the remaining dimensions five and six.
Read the full abstract

High Energy Physics (Phenomenology)

Inelastic $B-L$ scalar dark matter and the LUX-ZEPLIN event

We show that our previously proposed inelastic scalar dark matter (DM) model in gauged $U(1)_{B-L}$ symmetry naturally accounts for the $2.6\sigma$ high-energy recoil event recently reported by the LUX-ZEPLIN (LZ) collaboration. To satisfy the thermal relic abundance, TeV-scale scalar DM $S$ with $m_S \simeq m_{Z'}/2$ for resonant annihilation and the gauge coupling constant $g_{B-L} \sim 0.5$ are required, where $m_{Z^\prime}$ is the mass of $U(1)_{B-L}$ gauge boson $Z^{\prime}$. The resulting predicted DM-nucleon inelastic cross section of $\sigma_\mathrm{SI} \sim 10^{-45}$ cm${}^2$ is in excellent agreement with the LZ event. This scenario can be decisively tested for future collider searches for $Z^\prime$ boson resonance.
Read the full abstract

Materials Science

Scalable machine learning framework for multiphase identification from powder X-ray diffraction

X-ray diffraction (XRD) is the primary tool for identifying crystalline phases following synthesis, but automated phase identification remains challenging, particularly for multiphase samples with overlapping peaks and experimental artifacts. While deep-learning methods have been proposed to improve upon classical search-match algorithms, most formulate phase identification as a single closed-set classification problem, requiring one shared model to discriminate among all candidate phases. Here we introduce GALAXI, which instead decouples the identification task into independent one-versus-all binary classifiers that each specialize in recognizing a single phase. These pre-trained classifiers first narrow the search space to a small set of plausible phases, which are then evaluated through Rietveld refinement to identify the combination of phases that best explains the full diffraction pattern. On a curated set of experimental patterns, GALAXI identifies the correct phases with a micro-F1 score of 0.935, outperforming classical search-match and prior deep-learning models. The method remains robust to common experimental artifacts, including low impurity phase fractions, small crystallite size, peak shifts, sample displacement, and texture, and performs well when applied to time-resolved in-situ XRD data from solid-state reactions. Moreover, because the phase-specific models are independent, GALAXI can expand to large reference libraries without retraining existing models. This modular architecture enables us to train classifiers for 64,594 structures from the Crystallography Open Database and deploy them through a public web interface at https://galaxi-xrd.com.
Read the full abstract

All authors Xinyang Tong, Ethan Jin, Jiahan Xu, Aditya Rao, Pengcen Jiang, Nathan J. Szymanski

Econometrics

Sign Restrictions and Supply-demand Decompositions of Inflation

Sign restrictions on the slopes of supply and demand curves are often used to identify historical decompositions in structural vector autoregressions. I show that the identifying power of these restrictions depends on both reduced-form parameters and realised forecast errors. Consequently, unlike many other structural objects, the strength of identification cannot be assessed from reduced-form parameters alone. Empirically, identified sets for historical decompositions of US inflation are typically largely uninformative, both in aggregate and in most expenditure categories. Existing inflation decompositions are therefore sensitive to auxiliary assumptions used to select among observationally equivalent models.
Read the full abstract

Statistics Theory

Smoothed Picard Hamiltonian Monte Carlo

We develop a new low-accuracy sampler, called smoothed Picard Hamiltonian Monte Carlo, which combines Gaussian smoothing, Picard iteration, and higher-order discretization. For a log-concave target $\pi \propto \exp(-V)$ in dimension $d$ satisfying $0 \prec \alpha I \preceq \nabla^2 V \preceq \beta I$, with condition number $\kappa := \beta/\alpha$, smoothed Picard HMC returns a sample with $\sqrt \alpha\,W_2(\cdot,\pi) \le \varepsilon$ using $\widetilde O(\kappa^2 + \kappa^{7/6} d^{1/6}/\varepsilon^{1/3})$ gradient queries. We also prove stronger $W_q$ bounds, and then develop an algorithmic framework, the recursive warm start generator, to upgrade these $W_q$ bounds to stronger divergence guarantees. This produces a warm start for the proximal bouncy particle sampler, introduced in a companion work, leading to a high-accuracy log-concave sampler with complexity $\widetilde O((\kappa^{7/6} d^{1/6} + \kappa^{1/2} d^{1/4})\mathrm{polylog}(1/\varepsilon))$.
Read the full abstract

All authors Fan Chen, Sinho Chewi, Jianfeng Lu, Matthew S Zhang

Statistics Theory

Accelerated High-Accuracy Sampling from a Warm Start via the Proximal Bouncy Particle Sampler

We study the problem of sampling from $\mu(\mathrm{d}x)\propto e^{-V(x)}\,\mathrm{d}x$ on $\mathbb{R}^d$, where $V$ is $\alpha$-strongly convex and $\beta$-smooth, and write $\kappa:=\beta/\alpha$. We design and analyze the Proximal Bouncy Particle Sampler (Proximal BPS), a new sampler that combines ideas from the proximal sampler and the bouncy particle sampler. From a warm start initialization with $ O(1) $ Rényi divergence w.r.t. $\mu$, Proximal BPS returns a sample whose law is $\varepsilon$-close to $\mu$ in total variation distance using $\widetilde O(\sqrt\kappa\,d^{1/4} \,\mathrm{polylog}(1/\varepsilon))$ gradient queries in expectation.
Read the full abstract

All authors Fan Chen, Sinho Chewi, Jianfeng Lu, Matthew S Zhang

Probability

$q$-deformed polyanalytic Ginibre point processes:construction and central limit theorems for linear statistics

Based on the representation theory of the $q$-CCR algebra, we develop a representation-theoretic construction of the $q$-deformed polyanalytic Fock spaces, their reproducing kernels, and the associated $q$-deformed polyanalytic Bargmann transforms. The associated determinantal point processes provide $q$-deformations of the pure and full polyanalytic Ginibre ensembles. Moreover, we determine their limiting density and establish central limit theorems for the fluctuations of their linear statistics as the number of particles tends to infinity and the deformation parameter $q$ tends to 1 simultaneously. Remarkably, although the $q$-deformation changes the macroscopic density and the size of the droplet, the limiting covariance structures coincide, after spatial rescaling, with those of the classical polyanalytic Ginibre ensembles.
Read the full abstract

Commutative Algebra

Regularity of symbolic powers of complementary edge ideals

Let \(G\) be a finite simple graph on \(n\) vertices, and let \(I_c(G)\) be its complementary edge ideal. We determine \(\reg(I_c(G)^{(t)})\) for every \(t\geq1\) in terms of the number \(c(G)\) of nontrivial connected components of \(G\) and the presence of a \(K_2\)-component. Consequently, \(\reg(I_c(G)^{(t)})\leq \reg(I_c(G)^t) \) for all \(t\geq1\), with equality for every \(t\) if and only if either \(c(G)=1\), or \(c(G)=2\) and \(G\) has a \(K_2\)-component. We also characterize Serre's condition \((S_2)\) for \(R/I_c(G)\), and classify the graphs for which \(R/I_c(G)^{(t)}\) is Cohen-Macaulay for every \(t\geq1\).
Read the full abstract

Strongly Correlated Electrons

Spin–orbit-enhanced correlation sensitivity and anomalous magnetic response in non-dimerized $4d^4$ ilmenite CdRuO$_3$

We report the synthesis and physical properties of CdRuO$_3$, a nominal Ru$^{4+}$ $4d^4$ ilmenite with edge-sharing RuO$_6$ honeycomb layers. Powder X-ray diffraction establishes a crystallographically non-dimerized $R\bar{3}$ structure with equivalent Ru-Ru bonds and a strongly distorted RuO$_6$ environment. The compacted-pellet resistivity is nonmetallic but non-Arrhenius, while the heat capacity contains a finite residual linear term. Matched nonmagnetic calculations show that PBE+$U$ without spin-orbit coupling remains metallic or semimetallic up to $U_{\mathrm{eff}} = 3$ eV, whereas PBE+SOC+$U$ exhibits a strong $U_{\mathrm{eff}}$ dependence and opens a direct gap of approximately 55 meV at $\Gamma$ for $U_{\mathrm{eff}} = 2.5$ eV. Spin-orbit coupling therefore markedly enhances the correlation sensitivity of the non-cubic Ru $t_{2g}$ manifold. After subtraction of a dilute Curie-Weiss defect contribution, the susceptibility remains weakly nonmonotonic and is inconsistent with both an ordinary Pauli response and independent spin-only $S = 1$ moments. CdRuO$_3$ thus realizes the non-dimerized structural branch predicted for ruthenium ilmenites, but not a simple robust multiorbital metal.
Read the full abstract

Numerical Analysis

Data Assimilation with Sparse Observations

Data assimilation by nudging (also called CDA) yields exponentially decaying errors and an infinite predictability horizon if the method parameter is large enough and the observations are frequent enough in time and dense enough in space. We consider the complementary case of moderate parameters and sparse and infrequent observations. We prove that assimilation with any data and any(positive) parameter strictly decreases errors and strictly increases the (now finite) predictability horizon.
Read the full abstract

All authors Nan Jiang, William Layton, Nanda Nechingal Raghunathan, Troy Yang

Astrophysics of Galaxies

A Unified Magnetohydrodynamic Scaling Relation for the Multiphase Interstellar Medium

The interplay of magnetic fields, turbulence, and gravity governs the structural evolution of the interstellar medium (ISM) and the initial conditions of star formation, yet observational gaps have long enforced a broken power-law description of the magnetic field–density relation. Here we assemble a unified dataset spanning ten orders of magnitude in density ($10^{-26}$–$10^{-16}\,\mathrm{g\,cm^{-3}}$) by combining pulsar and Zeeman observations. The unified data are consistent with a continuous magnetic-field evolution organised by the Alfvén Mach number $\mathcal{M}_{\rm A}=\sqrt{E_K/E_B}$. Within this interpretation, the low-density gas is magnetically dominated ($\mathcal{M}_{\rm A}<1$), whereas the high-density gas becomes kinetically dominated ($\mathcal{M}_{\rm A}>1$) as gravity increasingly contributes to the kinetic-energy budget, with magnetic tension continuing to influence the collapse geometry. Within the Gradual Transition interpretation, the empirical break density traces the vicinity of the trans-Alfvénic equipartition point, $\mathcal{M}_{\rm A}=1$. This Gradual Transition model makes three predictions tested here. Its exponent and background field are fixed in advance by turbulent physics and recovered by the fits ($\beta\approx0.15$–$0.21$ against a predicted $0.147$; $B_c\approx2.0\,\mu$G). A dense-gas fit, extrapolated blindly across four decades, passes through the diffuse pulsar data. And, under the adopted scale mappings, the implied magnetic-energy spectrum approaches a $k^{-5/3}$-like scaling on large scales and departs from this extrapolation on small scales, where gravitational compression amplifies the field. The broken power law can therefore be viewed as a piecewise approximation to the continuous magnetic equation of state, with its fitted transition density potentially retaining a physical connection to the onset of gravity-driven motions.
Read the full abstract

Populations and Evolution

Kuramoto Phase Synchronization in Regional Epidemic Dynamics: Two Test Cases from European COVID-19 and Influenza Surveillance

We test whether the Kuramoto model quantitatively describes spatial synchronization in regional epidemic waves, using daily COVID-19 incidence for 400 German Kreise (2021–2023) and weekly ILI rates for 12 European countries (ECDC, 2021–2026). Bandpass filtering and Hilbert-transform phase extraction yield high global order parameters ($\langle r\rangle\approx0.87$ for Germany; $\langle r\rangle\approx0.97$ for Europe), yet both are dominated by a common-mode driver rather than pairwise local coupling. Removing the common mode reveals genuine short-range synchronization: an exponentially decaying residual local order with correlation length $\xi=152\pm6$~km in Germany, and a significant neighbor-vs.-non-neighbor contrast (Mann–Whitney $p=0.005$) in the European panel. We interpret this convergent pattern across two diseases and two spatial scales as a universal two-scale structure: a long-range common-mode field superimposed on shorter-range diffusive coupling. Cross-validation against a companion gauge-mediated Doi–Peliti framework shows that district-level effective screening mass $m_R$ is uncorrelated with local~$r$ (null result), the national $(m_R,r)$ phase portrait independently reproduces a previously reported hysteresis loop, and pairwise phase lags quantitatively match fractional-calculus predictions. A frequency–degree correlation provides the structural signature of explosive synchronization; excess synchronization between districts sharing the same Bundesland implicates shared school-holiday calendars and state-level policy as local coupling drivers.
Read the full abstract

Computation and Language

Dynamic-Programming-Guided Hierarchical BPE and Empirical Analysis of Vocabulary Pruning

Byte Pair Encoding (BPE) constructs vocabularies through greedy pair merging, but the resulting merge order does not necessarily allocate a fixed model-visible vocabulary optimally for compression. We propose Dynamic-Programming-Guided Hierarchical BPE (DH-BPE), a vocabulary-construction method that combines token exposure under exact minimum-token segmentation with the hierarchical dependencies induced by BPE training. Starting from a modestly overshot BPE candidate vocabulary, DH-BPE uses dynamic programming to measure candidate utility and applies exposure-guided, dependency-aware pruning to select a fixed-size model-visible vocabulary. We compare DH-BPE against Standard BPE and recent vocabulary-optimization baselines, including Pruned BPE, MinGram, and MinGram-PP, in primary evaluations at 12K and 16K target vocabulary sizes, with an additional 18K evaluation against MinGram only. Across the primary 12K and 16K comparisons, DH-BPE consistently improves aggregate compression over Standard BPE, Pruned BPE, and MinGram under a shared exact minimum-token DP encoder. MinGram-PP achieves stronger aggregate compression in the primary comparisons, but DH-BPE outperforms it at overshoot factors f = 2.0 and f = 3.0 in cross-corpus evaluation; at 12K, MinGram-PP reverses this ordering only with the substantially larger candidate pools at f = 4.0 and f = 5.0. Qualitative analysis further shows that DH-BPE balances later, more complete BPE merges with reusable subword components, providing a practical approach to improving vocabulary allocation under a fixed model-visible vocabulary budget.
Read the full abstract

Optimization and Control

Gramian-Informed Framework for Edge Flow Network Analysis Against Nodal Attacks

We study the first-order effects of nodal inputs on edge flows of network systems through controllability Gramians. We consider both discrete- and continuous-time dynamics subject to impulse and step input disturbances. To characterize their impact on network behavior, we introduce the notion of a vulnerability matrix (VM) and provide explicit Gramian-based expressions that reveal the critical role played by the interplay between network topology and dynamics. For the class of directed line networks, we particularize these closed-form expressions in terms of edge weights, input duration, and graph distance to the nodal input. Numerical simulations on directed line and random Erdős-Rényi networks show a tight correspondence between influential nodes identified with the VM and conventional time-domain performance metrics, such as $\mathscr{H}_2$ and $\mathscr{H}_{\infty}$.
Read the full abstract

Robotics

Distributed Secure Learning Control for Large-scale Multirobots under Stealthy Actuator Attacks

Distributed learning control for multirobot systems (MRS) offers significant flexibility in presence of uncertainties but lacks provable performance guarantees. A promising direction involves integrating reinforcement learning (RL) into distributed model predictive control (DMPC), leveraging the strengths of RL in nonlinear policy design and the receding-horizon replanning capabilities of DMPC. However, ensuring secure control within such a learning framework under malicious cyber attacks, particularly stealthy ones, remains a critical challenge, because the distributed policies generation depends on information exchange among neighbors, where compromised agents can rapidly influence the behavior of others through the communication network. This article proposes a distributed secure learning control (DSLC) framework for large-scale MRS under malicious, stealthy actuator attacks. Our framework offers two key features: (i) a unified approach that enables secure learning control across various coordination scenarios and (ii) a game-theoretic distributed learning-based predictive control strategy that learns how to balance the attacker and defender through a differential-game based DMPC framework. Specifically, DSLC employs a distributed attacker-actor-critic architecture to learn the optimal defense and attack policies online within each prediction interval. Unlike numerical optimization-based controllers that calculate open-loop control sequences, our method simultaneously generates adversarial attack policies and corresponding defense policies in analytical closed-loop form. The defense policies could be directly generalized to MRS with varying scales and diverse actuator attack probabilities. The effectiveness and scalability of DSLC are validated through comprehensive simulations and real-world experiments in multiple wheeled robots via various control tasks.
Read the full abstract

All authors Xinglong Zhang, Qingwen Ma, Cong Li, Hui Yin, Changxin Zhang, Yueying Wang, Wei Pan, Xin Xu

Number Theory

Three Irrationality Results for the Dilogarithm: Rhin–Viola and Viola–Zudilin Constructions at $-1/4$, $1/5$, and $-1/3$

We prove the irrationality of \[ \Li_2(-1/4),\qquad \Li_2(1/5),\qquad \Li_2(-1/3). \] The three arguments form a natural progression. The first is an endpoint completion of the five-parameter Rhin–Viola method. Rhin and Viola's 2019 treatment already supplies the negative-argument continuation, permutation invariance and factorial divisor; at $z=-4$ we use three bounded shifts, an elementary fixed-contour estimate and a four-term recurrence to remove the remaining nonvanishing problem in the complex-saddle regime. For $1/5$ we pass to the six-parameter Viola–Zudilin family and apply the Rhin–Viola factorial transformations term by term inside two binomial expansions of the same integral. This inherited divisor crosses the arithmetic threshold that the unrefined five-parameter construction does not reach in our computations. The same crossbreed, combined with the negative-argument continuation, proves the result at $-1/3$. All proof-critical finite inequalities are accompanied by exact executable certificates. Supplementary computations propagate $10{,}000$ exact primitive coefficient pairs for each of $1/5$ and $-1/3$, with complete gcd removal and no failure of the certified smallness or adjacent nonproportionality checks.
Read the full abstract

2609.06895 · 7 Sep · 28 pages. Ancillary files contain exact verification scripts and certificates for the arithmetic, contour and recurrence arguments, together with supplementary 10,000-term stress tests for Li_2(1/5) and Li_2(-1/3). The numerical stress tests are not used in the proofs

Optics

Pulse-by-pulse programmable synthesis of ultrafast optical waveforms

Programmable control of individual pulses in a high-repetition-rate (typically MHz) ultrafast pulse train is a long-standing goal for optical arbitrary waveform synthesis. Here, we report a programmable pulse-by-pulse shaper that enables deterministic spectral-temporal control of ultrafast pulses at a repetition rate of $\sim$ 20~MHz. By synchronizing an FPGA-driven electro-optic modulation on the stretched waveform in a temporal 4$f$ shaping system, the regime writes pulse-index-dependent spectral phase profiles onto individual pulses. We demonstrate three levels of programmable ability: zero-order phase coding that maps pulse-by-pulse phase sequences into double-slit-like spectral-temporal interference; first-order phase programming that produces arbitrary temporal trajectories with deterministic delay; and fractional-order phase engineering that generates programmable temporal breathing. By launching the shaped pulse train into a nonlinear fiber stage, the programmed temporal breathing is converted into one spectral breathing. We further construct a phase-level-dependent regime map of nonlinear spectral breathing, revealing transitions from weak single-envelope breathing to multi-peak spectral splitting and strongly breathing merged-spectrum dynamics, in agreement with numerical simulations. This pulse-by-pulse spectral-temporal synthesis platform establishes pulse index as a programmable degree of freedom for ultrafast pulse shaping and provides a route toward real-time and pulse resolved optical arbitary waveform synthesis.
Read the full abstract

All authors Shilong Liu, Gabriel Demontigny, Émile Dessureault, Stéphane Virally, Denis V. Seletskiy

Computation and Language

Towards Bridging the Gap Between Offline and Iterative Alignment via Preference Distillation

Direct preference optimization DPO is a promising offline approach for aligning large language models (LLMs) due to its simplicity, computational efficiency, and implicit modeling of human preferences. Interestingly, iterative extensions of DPO have achieved stronger performance on academic benchmarks, raising two key questions: (i) Why do iterative methods generally outperform offline ones? (ii) Can their advantages be incorporated into offline alignment? To answer the first question, our controlled experiments reveal that the explicit preference model, additionally introduced in the iterative procedure, is a key factor behind its superiority over offline methods. This insight leads us to answer the second question affirmatively and propose Distilled Preference Probability Policy Optimization (DP3O), an effective and efficient offline alignment algorithm. DP3O first learns an explicit preference model using a helper class of LLMs and then distills its knowledge into policy optimization. Theoretically, we show that explicit preference modeling admits better estimation error control than implicit formulations, and that DP3O achieves a tighter generalization bound than hard-label DPO through variance reduction. Empirically, we evaluate DP3O on a wide range of chat-based and downstream tasks and show that it outperforms state-of-the-art offline methods, achieves performance comparable to iterative DPO, and reduces training time by about $42\%$, demonstrating both its effectiveness and efficiency.
Read the full abstract

All authors Wenbo Zhang, Wenzhuo Zhou, Hengrui Cai, Zhengling Qi

Probability

Hydrodynamic limit of exclusion processes with killing on weighted Riemannian manifolds via a graph discretization

In the present paper, we consider an exclusion process with killing on a proximity graph constructed from a partition of a geodesically complete weighted Riemannian manifold. We rigorously derive its hydrodynamic equation, which is governed by a heat equation with a killing potential term on the manifold. Within our graph discretization framework, we define an empirical density field for the exclusion process and prove its convergence, under an appropriate space-time scaling, to the unique bounded weak solution of the hydrodynamic equation, provided that the weighted manifold is stochastically complete. Combining local regularity for the parabolic equation with techniques from stochastic analysis on manifolds, we prove the uniqueness of the bounded weak solution and give the explicit representation of the solution in terms of the minimal Schrödinger kernel. Our result requires neither a global lower Ricci curvature bound nor global boundedness of the weight function.
Read the full abstract