Statistical Mechanics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Monday 9 November
The kinetic approach for q-distribution is extended through a "bottom-up" procedure to the corresponding entropic formula. In this way, the validity of Tsallis' formula, S_q(W) = k_B ln_q(W), is established from first principles. The resulting q-entropy is additive and the zeroth law of thermodynamics is automatically satisfied because the conventional identification beta = 1/(k_B T) remains valid. Entropy, energy, and all relevant quantities for both discrete and continuous cases are evaluated using standard (linear) averages. The "bottom-up" and "top-down" approaches yield the same q-distribution for velocities. The standard results are naturally recovered in the limit q -> 1.
Software Engineering
Exploring public-private data sharing practices through the lens of Open Data Ecosystems: A multiple case study on municipal and regional freight logistics
Background: Freight transport is expected to increase, contributing to air pollution. There is, therefore, a critical need for increased efficiency that can reduce emissions. Data sharing is one approach, where actors within Open Data Ecosystems (ODE) collaborate and share data among stakeholders to plan and operate more efficient and sustainable freight transport. Aim: In this study, we explore how public-private data sharing practices in municipal and regional last-mile freight logistics in Sweden may improve efficiency. Method: Following a prolonged participation process, we conducted a multiple-case study involving three different cases, built upon initial meetings, semi-structured interviews, supported by a thematic analysis, and validated through a follow-up workshop. Results: The findings show that data sharing in public-private freight logistics (ODE-PPF) is shaped by interrelated socio-technical aspects, including platform design, roles of actors, handling of sensitive data, data quality, and procurement processes. We synthesize our findings in an actor model and a conceptual model, highlighting how these aspects contribute to the functioning of ODE-PPF, and derive seven proposition sketches as a foundation for future research. Conclusion: Data sharing is a central mechanism enabling integration and value creation in last-mile logistics. It depends on the interaction between technical and organizational aspects. Public actors can undertake a key role as a neutral keystone actor by orchestrating the ecosystem and facilitating data sharing through governance practices such as procurement.
Read the full abstract
Systems and Control
KPG193-ext: A Synthetic Korean Power Grid Test System for Stability Analysis
Growing shares of inverter-based generation and constrained transmission expansion increase the need to assess stability across operating conditions. Production-cost models provide chronological commitment and dispatch, but stability analysis requires corresponding electrical detail and device dynamics. This paper introduces KPG193-ext, an extension of the synthetic Korean power-grid test system KPG193 that links chronological operating points with detailed electrical and dynamic models. The extension adds synchronous-machine, controller and renewable-converter models, together with explicit voltage levels, generation terminals, inter-voltage and step-up transformers, and reactive compensation. The resulting 731-bus network retains the original 193 regional locations. Daily unit commitment followed by hourly AC optimal power flow establishes 8,760 operating points for 2025, each distributed as paired steady-state and dynamic-model files. Selected-hour studies examine generation-loss frequency response, steady-state flow-voltage characteristics, and rotor-angle response with special protection system actions. The results reveal differences in delivered frequency support, internal voltage stress along regional transfer paths, and the angle-frequency tradeoff of corrective generator tripping. The database provides a common basis for examining the stability implications of commitment, dispatch and corrective action across daily and seasonal operating conditions.
Read the full abstract
Number Theory
Nontrivial Zeros of the Riemann Zeta Function and Its Derivative
The nontrivial zeros of the Riemann zeta function lie at the heart of analytic number theory. Through the explicit formulas, they govern the distribution of prime numbers: the closer all nontrivial zeros lie to the critical line $\Re(s)=\frac12$, the more uniformly the primes are distributed. The Riemann Hypothesis-asserting that all nontrivial zeros satisfy $\Re(s)=\frac12$-is equivalent to the optimal error bound $O(\sqrt{x}\log x)$ in the Prime Number Theorem. Beyond number theory, the zeros of $\zeta(s)$ and its derivative $\zeta'(s)$ are deeply intertwined: the horizontal distribution of $\zeta'$-zeros near the critical line encodes information about the spacing of consecutive $\zeta$-zeros. This paper establishes some results concerning the nontrivial zeros of the Riemann zeta function.
Read the full abstract
Computer Vision
Occupancy Modelling of Biased Landslide Inventories: When Does Separating Occurrence from Recording Improve on Neutralisation?
Landslide susceptibility models are trained on inventories that record where landslides were seen, not where they occurred. The established remedy fits bias-describing variables and neutralises them when mapping. We adapt occupancy-detection modelling from ecology, treating several inventories as imperfect observers of one event and separating occurrence from recording, and ask when this improves on neutralisation. We use near-complete inventories of the 2018 Hokkaido Eastern Iburi, 2015 Gorkha and 2010 Haiti earthquakes with simulated reporting bias, and three real Gorkha inventories. Compared with standard practice, the occupancy model reduces road bias in 24 of 25 paired draws and raises AUC by 0.08-0.14 on Gorkha and 0.03 on Haiti, but only 0.01-0.02 on Iburi. Its median predicted-to-true landslide rate is 0.99 (range 0.44-1.89), against 0.24-0.30 for the baselines. Rescaling the baselines with the same anchor area corrects their median but leaves larger errors (range 0.20-4.34). Cell-level probabilities remain too extreme (calibration slope 0.59). Compared with the best neutralised model, AUC is higher on Iburi (by 0.03-0.07) and lower on Gorkha and Haiti (by 0.01-0.05). Transplant experiments tie the Iburi advantage to the structure of its true landslide pattern, not to its terrain or road network. In all eight settings with a low interaction gap, computable from biased records, neutralisation was at least as accurate, including a prespecified test on Haiti; a high gap did not guarantee an advantage. The real Gorkha inventories are slope-biased, not road-biased, and there the model brings no meaningful gain. Report-based catalogues remain the main untested case.
Read the full abstract
Machine Learning
Patching Up Circuit Learning via Parameter Sharing
Probabilistic circuits (PCs) have emerged as generative models guaranteeing tractable probabilistic inference. However, tractability comes at the price of larger model sizes than in other generative models. As such, to scale PCs to high-dimensional data such as images, previous work has devised several parameter sharing techniques as well as fast evaluation protocols that consider learning PCs on independent image patches. In this work, we revisit this practice to highlight a surprising learnability issue: models that assume independence counterintuitively can yield better likelihoods than models that do not. To break this apparent bottleneck and to scale PCs further, we introduce a recursive parameter sharing construction (RePA) that fosters efficiency without assuming independence. We show empirically that RePA can greatly reduce time and memory consumption while yielding competitive likelihoods. Inspired by these results, we further generalize the approach to data modalities that do not inherently exhibit patch-structure (NoPA), such as tabular data, showing similar efficiency improvements, with marginal changes in likelihoods and sample quality on seven tabular datasets. Code available at https://github.com/april-tools/RePA.
Read the full abstract
Hardware Architecture
To SIMD or Not to SIMD: Comparing RVV vs. CGRA SoCs for Non-Linear Functions in AI
Non-linear functions are an important class of operations in modern large language models. However, chip vendors building for AI have so far relied on traditional vector or scalar processing units for non-linear functions. The performance of these units has not kept pace with matrix multiplication units (MMUs), which can be found in both high-performance and edge-computing devices from almost every vendor. Reconfigurable units such as FPGAs and CGRAs have been proposed in the literature, but have not been systematically compared with traditional units backed by MMUs. In this paper, we build two SoCs: one using an RVV vector processor and the other using a latency-insensitive (elastic) CGRA, both coupled with an MMU and driven by a single-core CPU. We then propose mappings for common feed-forward networks on our SoCs and find that the CGRA reaches a maximum speedup in clock cycles of 7.12x over RVV on the activation stage of simple point-wise non-linear functions like SiLU, whereas the RVV reaches a maximum speedup of 9.8x over the CGRA on complex non-linear functions such as $\mathrm{softmax}$. Over the enclosing feed-forward block the corresponding maxima are 3.16x and 4.24x.
Read the full abstract
Computers and Society
Frontier AI Safety and Governance 2026: Risk Understanding and Coordinated Governance
Frontier AI Safety and Governance 2026: Risk Understanding and Coordinated Governance is a key outcome of the AI Safety and Governance Program of World Internet Conference Specialized Committee on AI in 2026. Drawing on the latest developments in frontier AI capabilities and applications, the report systematically examines the evolving risk landscape, analyzing key risk areas, their severity and response priorities. Building on a review of existing global governance practices and an assessment of frontier AI governance readiness, it identifies key governance gaps and weaknesses and proposes directions and practical mechanisms for coordinated responses. The report provides a reference for government departments, international organizations, businesses, research institutions and other stakeholders. It aims to strengthen the international community's understanding of and consensus on frontier AI risks, enhance preparedness and collective response capacity, and advance the safe, reliable and responsible development of frontier AI to better serve the common well-being of humanity.
Read the full abstract
Fluid Dynamics
N-Layer quasi-geostrophic modons in a vertically sheared flow
Modons, or dipolar vortices, are a form of propagating vortex solution consisting of two counterrotating regions of fluid and moving together under self-advection. They are common features of the upper ocean that can remain stable over long times, thereby transporting fluid over large distances. Here, we present a semi-analytical method for finding modon solutions in a vertically-sheared background flow using the multi-layer quasi-geostrophic framework. This method works for arbitrarily many fluid layers and builds on our previous studies of multi-layer modon solutions. We present three example of modons in a background flow, highlighting the effect of varying relative potential velocity between layers and the impact of depth-dependent potential vorticity gradients arising from surface deformation and velocity shear.
Read the full abstract
Materials Science
Lattice dynamical properties and interatomic force constants of transition metal oxide perovskite superlattices
In this study, we investigate the construction of efficient model potentials for perovskite heterostructures by transferring relevant dynamic quantities from bulk systems and short-period superlattices. Through first-principles calculations, we provide a comprehensive analysis and comparison of the dynamical properties of different perovskite-based superlattice structures, including (PbTiO3 )n /(SrTiO3 )m superlattices, (ATiO3 )n /(AO) 2 superlattices, and An+1 Tin O 3n+1 Ruddlesden-Popper phases, where A can be Sr, Ba, or Pb. We demonstrate that as the thickness of these structures increases, the dynamical properties of the perovskite blocks converge rapidly towards their bulk counterparts. We discuss this convergence in terms of phonon density of states in reciprocal space and inter-atomic force constants in real space. Our findings suggest a straightforward strategy for developing transferable model potentials that accurately capture the dynamics of layered oxide compounds with progressively thicker layers.
Read the full abstract
Applied Physics
Mechanisms of Zener diode and ohmic contact resulting from a heavily doped p-n junction: A novel explanation in terms of impurity-photovoltaic-effect resulting from infrared self-emission at room-temperature
In textbooks, a quantum-mechanical tunneling mechanism based on electron-wavefunction is used to explain Zener diode as well as ohmic contact resulting from a heavily doped p-n junction. In this study, an attempt in particle description is made to explain these devices. Any object at room-temperature emits infrared (IR) photons. The process of heavy doping can cause many defects. The self-absorption of the IR emission could be achieved through sub-band-gap excitations due to defects, creating carriers. In a heavily doped p-n junction device, some of the IR-generated carriers diffuse into the junction which has a built-in field. The built-in field sweeps out the carriers, creating IR photocurrent. The reverse leakage current of the p-n junction is regarded as the IR photocurrent according to a precedent where impurity-photovoltaic-effect is used to explain circuit devices (arXiv:2510.18226). Heavy doping results in not only a great many IR-generated carriers but also an avalanche effect at low voltage, producing Zener breakdown. Additionally, the impurity-photovoltaic-effect can make a reasonable contribution to explaining the temperature dependence of Zener breakdown. If the doping is very heavy, the heavily doped p-n junction can be altered to create an ohmic contact, which can also be explained in terms of impurity-photovoltaic-effect. Not only quantum mechanics, but also this photovoltaic mechanism in particle description can explain the above devices. Considering the wave-particle duality, the combination of quantum-mechanical and photovoltaic mechanisms is suggested to explain these devices. Each mechanism plays a big or small role in the explanation.
Read the full abstract
Medical Physics
Sensitivity of diffusion MRI to white matter pathology: a comparison between ROI-based effect size and TBSS analysis under different diffusion protocols, field strengths and post-processing strategies
We investigated how much more advanced diffusion MRI acquisitions and more complex processing increase sensitivity to group-wise white matter (WM) differences due to pathology, and how the choice of statistical analysis affects this sensitivity. Data were acquired with DTI and DKI protocols at 3T and 7T in patients with systemic lupus erythematosus (SLE), athletes with persistent post-concussive symptoms (PPCS), and healthy controls (HC). Seven processing pipelines combining denoising, Gibbs removal, eddy current and motion correction, and Gaussian smoothing were applied. Sensitivity was assessed with two methods: a region-of-interest-based (ROI-based) effect size analysis over 72 WM tracts segmented by TractSeg, quantified by Cohen's d, and tract-based spatial statistics (TBSS), quantified by the fraction of significant voxels in the WM skeleton. The ROI-based analysis found significant group differences for both SLE versus HC and PPCS versus HC, whereas TBSS found differences only for SLE versus HC. Both analyses agreed that the diffusion protocol affected sensitivity more than field strength or post-processing. Regardless of pipeline, DTI-derived fractional anisotropy (FA) yielded a 0.26 higher effect size and 25% more significantly different voxels between SLE and HC than DKI-derived FA, and 3T yielded a 0.2 higher effect size and 21% more significant voxels than 7T. Smoothing reduced sensitivity in both analyses (by up to 12% in TBSS for SLE versus HC). Overall, TBSS appears less suited to localized pathology such as neurotrauma, whereas the ROI-based analysis showed no bias with respect to the nature of the pathology. Advanced protocols such as 3T-DKI and 7T-DTI may need further optimization before surpassing conventional 3T-DTI in sensitivity to WM pathology in SLE.
Read the full abstract
Computation and Language
Can Frontier Auto Researchers Reconstruct Human Research Decisions in AI?
Scientific research has long been driven by substantial human effort, from reviewing prior work and identifying research gaps to designing specific solutions. Recent advances in harness-based agent systems have given rise to increasingly capable Auto Researchers, which can autonomously carry out these tasks and, increasingly, entire end-to-end research workflows. Yet despite their rapidly growing capabilities, how closely the research decisions and outputs produced by auto researchers align with those of human experts remains underexplored. We investigate this gap through the lens of research reconstruction and study: Given comparable research context, how similar are the research decisions and outputs produced by Auto Researchers to those of human researchers in AI research? To answer this question, we introduce AutoResearcher-Human Research Similarity (AutoHRS), a stage-wise evaluation framework for assessing how faithfully auto researchers reconstruct human research artifacts across different stages of the scientific workflow. Using 152 ICLR oral papers as our corpus, we generate 456 research trajectories spanning different stages of the research process. Applying AutoHRS to these trajectories, we find that current Auto Researchers frequently fail to recover key research decisions and their resulting artifacts across multiple stages of the research process.
Read the full abstract
Mathematical Physics
An Algebraic Foundation of Rate-Product Style Quantity Calculus
Using the new notion of a scalable (right) monoid-heap, the quantity calculus in [1] is generalized to include quantities that may not be possible to multiply as well as both (weak) rates and products of quantities. Several examples are given of applications in metrology and physics
Read the full abstract
Machine Learning
UCB-min: Combining Bandits Under Variance Heterogeneity (Extended Version)
Multi-armed bandits are a natural framework for many real-world sequential decision problems, from recommender systems and clinical trials to educational test recommendation and online advertising. The performance of these algorithms depends on how they explore different arms, and exploration strategies range from variance-independent (e.g., UCB1) to variance-aware (e.g., UCB-V). In practice the reward variance of each arm is rarely known and varies widely across arms, and the reward support is often underspecified. In these cases no single rule is uniformly best: variance-independent methods are robust but loose when variances are small, while variance-aware methods are tight only when their variance estimates and the assumed support bound are reliable. Such heterogeneous, unknown variances arise in many settings, for instance when multiple objectives are scalarized into a single reward. We propose UCB-min, a per-arm adaptive meta-algorithm that selects, for each arm, the tighter of a variance-independent and a variance-aware exploration bonus, without prior knowledge of the variance regime and without reward normalization. We prove that, with UCB-V as variance-aware base, its regret is bounded by the sum of the two base bounds at the standard exploration constant, and by their per-arm minimum, up to a constant, for a larger one. We validate it on synthetic data, an educational tutoring task, and two group-recommendation benchmarks (MovieLens, Steam), showing that UCB-min consistently tracks the best base algorithm across diverse variance regimes and datasets.
Read the full abstract
Computer Vision
dNaViT: Discrete Codes are General-Purpose Visual Representations
Discrete autoregressive modeling, notably through next-token prediction, has become a cornerstone of modern language models. Extending this paradigm to vision allows visual and language models to converge on a unified framework, thereby fostering shared methodology and infrastructure. This work investigates a central question: how far can a finite discrete vocabulary go in representing the richness of the visual world? We present dNaViT, a Discrete Native-resolution Vision Transformer that produces discrete visual primitives analogous to language tokens. Built upon a frozen pretrained encoder, dNaViT applies lightweight residual modulation with a dual-branch objective to inject reconstruction-aware cues into semantic features. By leveraging grouped product quantization, it scales discrete capacity without inflating spatial sequence lengths, while a balanced training recipe maintains both semantic and pixel-level information. Consequently, dNaViT functions as a versatile discrete visual tokenizer, enabling unified encoding across downstream tasks alongside high-fidelity image decoding. Extensive evaluations show that dNaViT matches or exceeds task-specific baselines across 12 visual understanding benchmarks, 4 generation benchmarks, and an image editing benchmark. These findings demonstrate that a compact discrete vocabulary can serve as a universal visual currency and unlock the full potential of fully unified, native multimodal methodology.
Read the full abstract
Signal Processing
Neural Multipath Fields for 6G FR3 Digital Twins: Latent Adaptation from Sparse Data
Wireless digital twins are essential for environment-aware 6G communications, motivating the investigation of frequency range 3 (FR3), an emerging mid-band regime offering a promising balance between bandwidth and propagation characteristics. Neural Radio Frequency (RF) fields provide continuous surrogates for site-specific radio propagation, but adapting a scene-specific field to a new environment can require substantial measurements and retraining. To this end, we propose the Latent-Adaptive Neural Multipath Field (LANMF), a scene-conditioned FR3 model that predicts path-resolved delays, complex channel impulse response (CIR) coefficients, and arrival and departure directions from transmitter-receiver coordinates. LANMF learns shared propagation representations across source environments and adapts to unseen scenes by optimizing only a 16-dimensional scene latent. Evaluated on six ray-traced indoor scenes at 7.5 GHz, latent-only adaptation yields improvements in prediction with just 5-10 target links (0.42-0.83% of available links). Compared with full-network fine-tuning, it updates only 16 persistent target parameters, reducing adaptation time by 28-30% and incremental peak GPU memory by 82-85%. While the proposed solution supports various wireless applications, we demonstrate its utility through beam-candidate reduction using a 64-beam codebook, retaining measurement-based final beam selection.
Read the full abstract
Computational Engineering
Physics-informed neural networks in the small-data regime: A controlled comparison with classical PDE surrogates
Neural surrogates for engineering models are often trained on large collections of simulated solutions. We investigate whether physics-informed neural networks (PINNs) can achieve accurate predictions from scarce labelled observations by incorporating governing equations and prescribed initial and boundary conditions. We introduce hierarchical loss weighting (H-PINN) and its gradient-direction-aware extension (HO-PINN), which adjust the emphasis on physical constraints according to higher-priority losses. Hierarchical PINNs are compared with Vanilla PINN, tuned gradient boosting, support-vector regression, and Gaussian-process regression on a heat equation and three heterogeneous two-phase-flow benchmarks. Experiments use 20–1000 labelled states and ten paired seeds. Each label is one coordinate–state observation, not a complete simulation. At least one PINN variant achieves the highest mean primary-target \(R^2\) in each interpolation and equation-assisted forecasting comparison at 20 and 50 labels. The advantage for water saturation persists across all tested budgets. Hierarchical weighting reduces variability across runs in selected settings. These results demonstrate that known physics can support accurate neural surrogate predictions from very few solution observations, at substantially higher fitting cost than the tested classical regressors.
Read the full abstract
Systems and Control
Scalable Practical Synchronization of Heterogeneous Linear Systems
We study practical state synchronization of heterogeneous linear multi-agent systems over directed graphs using static diffusive state feedback. Persistent mismatches between the agent dynamics generally preclude exact synchronization and are treated as bounded disturbances within an Input-to-State Stability (ISS) framework. We derive a scalable synthesis procedure based on fixed-dimension linear matrix inequalities (LMIs), whose dimension is independent of the number of agents. Local robust dissipation conditions are combined through a small-gain argument to obtain global ISS guarantees and explicit bounds on the synchronization error. The framework is formulated in a unified manner for continuous- and discrete-time systems, while highlighting a fundamental difference in disturbance attenuation between the two domains. In particular, exact ZOH discretization of a continuous-time controller need not preserve closed-loop stability, motivating a native discrete-time synthesis. Numerical experiments on heterogeneous harmonic oscillators illustrate the proposed approach and the continuous/discrete-time distinction.
Read the full abstract
Machine Learning
Two-Phase Rapid Simulation-Based Inference with Differentiable Simulators
We introduce Rapid Simulation Based Inference (RSBI), a diffusion-based variational approach to likelihood-free Bayesian inference that achieves high sampling efficiency under large prior-to-posterior volumes. RSBI builds on advances in Schrödinger Bridge (SB) diffusion sampling to handle multi-modal posteriors, with an initial likelihood supplied by a distance measure via a differentiable simulator. The key innovation of this initial surrogate likelihood is that it gives a highly efficient proposal distribution, replacing sequential rounds of other methods with an efficient one-shot proposal. Subsequent refinement then transports the proposal to the true posterior as specified by the simulator and the original prior. The SB sampler is warm-started from the proposal, and its energy is corrected by a neural likelihood ratio that is learned only where the proposal places mass. As a variational method, RSBI does not suffer from prior leakage commonly observed in standard sequential posterior estimation methods. To improve mode coverage we optionally utilize the well-tempered meta-dynamics framework, which also encourages exploration of the prior volume. We observe strong performance on standard SBI benchmarks, particularly when the prior volume is increased relative to the ground truth posterior, achieving a performance degradation of only $\sim5\%$ on Two Moons under a highly constrained simulation budget. Additionally, we evaluate RSBI's performance for gravitational wave ring-down posterior estimation, in a real-world physics benchmark inspired by black hole spectroscopy.
Read the full abstract
Analysis of PDEs
Positive solutions of power equations in planar corner domains
We study positive solutions of $\Delta u=\lambda u^k$ on bounded planar corner domains with zero boundary values away from the vertex. For negative-power absorption we identify the admissible interval of Martin masses and prove uniqueness at the minimum mass. Along the ordinary boundary, we identify the weak Hessian endpoint and its sharp tail when the set of zero inward derivative has positive length. For superlinear sources we prove attainment of the largest positive Martin mass in the subcritical range. At criticality we construct a family of singular solutions with zero Martin mass and distinguish its members by their second logarithmic coefficient and the endpoint integrability of their Hessian differences. Above criticality an angular eigenvalue governs the Morse index away from its zero threshold. We also give sharp physical-Hessian asymptotics for finite-mass remainders and joint corner and side estimates for negative-power sources.
Read the full abstract
Computation and Language
Path-Level Structural and Semantic Alignment for Hierarchical Multi-Label Text Classification
Hierarchical multi-label text classification (HMTC) requires models to capture both structural dependencies among labels and the fine-grained semantics of labels at different hierarchy levels. Existing methods often model hierarchy through prediction constraints or representations of individual labels and local relations. Structural and semantic supervision, however, are usually formulated separately. We instead use complete root-to-node label paths as a shared unit for both. Our framework introduces two auxiliary training objectives while retaining a standard multi-label classifier at inference. PathSimNCE aligns document representations with hierarchical paths using hierarchy-aware contrastive targets. These targets provide graded supervision for structurally related paths while preserving explicit supervision for the ground-truth paths. The complementary semantic objective uses LLM-generated, document-conditioned descriptions of the same positive paths. It thereby provides context-specific label semantics beyond static label representations. Experiments on three benchmarks show consistent improvements: structural alignment is particularly beneficial in deeper hierarchies, while semantic alignment provides additional gains for fine-grained labels.
Read the full abstract
Machine Learning
Optimal Rates for Adaptive Private $k$-PCA
Given $n$ i.i.d. random matrices $A_i \in \mathbb{R}^{d \times d}$ with common expectation $\Sigma$, the goal of Differentially Private Stochastic PCA is to identify a $k$-dimensional subspace capturing the leading variance directions of $\Sigma$, while preserving differential privacy~(DP) for each individual sample $A_i$. \citet{dungler2025iterative} introduced \kdppca, the first algorithm to simultaneously (1) achieve sample complexity $n=\widetilde O(d)$ for sub-Gaussian data, (2) adapt its privacy noise to the intrinsic randomness of the data, and (3) extend seamlessly to any target dimension $k\le d$. However, its sample complexity has a suboptimal dependence on $k$. We propose the first algorithm that achieves optimal sample complexity in both $d$ and $k$, while retaining properties (2) and (3) of \kdppca. Additionally, prior utility guarantees require a minimum sample size with exponential dependence on the eigengap. Our method removes this dependence. It also improves the dependence on spectral parameters, matching known lower bounds for the spiked covariance model. Unlike deflation-based approaches like \kdppca, our algorithm updates the full $d\times k$ subspace jointly rather than one eigenvector at a time. This non-deflation structure simplifies the algorithm, reduces the number of hyper-parameters, and improves computational efficiency, particularly when using block linear algebra libraries.
Read the full abstract
High Energy Physics (Theory)
First-order origins of second-order supersymmetry
One-dimensional second-order supersymmetry is constructed from conserved first-order operators on a surface with a cyclic coordinate. In Witten's mechanics with two fermionic modes, a centrally extended $N=4$ algebra permits a conserved product connecting the zero- and two-fermion sectors. Cyclic reduction gives the second-order supercharge and the correlated quantum corrections in the charge and partner Hamiltonians. Isothermal coordinates identify the auxiliary Hill equation with the separated Laplace or wave equation for the gauge-transformed charge kernel, and express the correction as a Schwarzian derivative. The central momentum fixes the quadratic invariant $C$: the Riemannian construction gives $C\leq0$, and its indefinite counterpart gives $C\geq0$, with a separately chosen positive adjoint after reduction. Diagonalizing the one-fermion intermediate recovers real scalar factorizations for $C\leq0$ and complex ones for $C>0$. Pauli reductions provide a second first-order mechanism; on the hyperbolic plane they reveal a curvature-dependent invariant and additional charges of a regular free-particle–Pöschl–Teller pair in a selected momentum sector. In one dimension, a formally Hermitian two-component intermediate with a positive inner product yields local first-order chains for arbitrary $C$. For a constant superpotential, a real transverse deformation on a flat cylinder also gives a first-order surface realization of $C>0$ with a positive inner product.
Read the full abstract
Quantum Algebra
Holomorphic integer-graded vertex operator superalgebras are purely even
We prove that a simple, self-contragredient, rational, $C_2$-cofinite, integer-graded holomorphic vertex operator superalgebra of CFT-type has zero odd part. We first show that a hypothetical nonzero odd part would be a simple current of categorical dimension $-1$ over a strongly rational pointed VOA with four simple modules. We then prove that every nonvacuum simple current of a strongly rational VOA has strictly positive lowest conformal weight. This leads to a contradiction with Dong-Jiao-Xu's theorem on the positivity of categorical dimension.
Read the full abstract
Solar and Stellar Astrophysics
Parker Solar Probe Flies Through the Sun's Magnetic Poles Near 10 Solar Radii
During the peak of the Sun's magnetic cycle, the solar corona can be complex and dominated by dynamics, while hidden below, the dipole component of the its photospheric magnetic field steadily reverses with respect to the rotational axis. Here, we use Parker Solar Probe (Parker) data from two of its closest approaches in mid-2025 to show that the large-scale field at solar maximum resembled a simple tilted dipole. Similar to how Ulysses provided a complete scan of the solar rotation-aligned dipole at solar minimum, Parker sampled the full range of magneto-latitudes, probing both magnetic poles below the Alfvén surface. Concurrent in situ data at larger distances reveal that both hemispheres contain solar wind with asymptotic speeds near 800 km s-1, yielding a picture similar to the Ulysses latitude scans but now in the equator near solar maximum. This contrasts with the disordered structure observed by Ulysses during its solar maximum latitude scan, which we attribute to the long timescale of data collection. Overall, we find the coronal field is organized as a dipole, but the fast solar wind streams are deformed by non-dipolar substructure: the positive dipole axis is aligned with one coronal hole, the negative centers over a pseudostreamer, and significant mass flux enhancements are observed at magnetic separatrixes. Our measurements suggest that even at solar maximum, the solar corona can still exhibit structure nearly as ordered as solar minimum. Moreover, by sampling rapidly and close to the Sun, the detailed imprint of source region structure can be determined in the solar wind.
Read the full abstract
Mesoscale and Nanoscale Physics
Gate-tunable supercurrent switching in a three-terminal quantum dot transistor
We propose and analyze a three-terminal superconducting transistor switch in which a gate-defined quantum dot region interrupts a flux-controlled loop. The output supercurrent is switched off completely when the two loop arms couple to the dot with equal tunneling strengths and half a flux quantum threads the loop. The device then operates as a symmetric SQUID at half flux. The condition follows from a single geometric factor $\chi$, which reduces the three-terminal problem to an equivalent two-terminal Anderson impurity model and makes the switching mechanism independent of the Coulomb interaction, the dot level position, the total hybridization and temperature. These enter only through a prefactor and thus decide how tolerant the device is to imperfect barrier tuning. Numerical renormalization group calculations across the full coupling parameter space show that the resonance is embedded in a broad region of elevated switching ratio at intermediate hybridization $\Gamma \approx \Delta$, while both weaker and stronger coupling reduce it to a thin isolated line. Finite temperature smooths the underlying singlet-doublet transition and lowers the supercurrent amplitude, but affects the open and closed states comparably and leaves the switching ratio largely intact. Within a zero-bandwidth two-level model, a second dot level is shown to open a leakage channel that grows as the two levels approach in energy. Raising the output barrier and the realistic presence of inter-level Coulomb repulsion, however, suppress the leakage.
Read the full abstract
Systems and Control
Sensor placement for real-life water distribution networks, ensuring observability under uncertainties
Leakage detection in water distribution networks (WDNs) is crucial for limiting water losses. However, effective detection strongly depends on sensor availability and placement. Most existing sensor placement methods rely on extensive simulations or precise knowledge of hydraulic parameters, which are often uncertain in practice. In this paper, we propose using a scalable sensor placement framework based on structural observability that depends only on the network topology and the underlying pressure-flow interconnection structure. The approach guarantees structural observability independently of hydraulic parameters, and is therefore robust to parametric uncertainties, assuming that the network topology is correctly specified. The method is computationally efficient and its performance is demonstrated on several real-life WDNs provided by Brabant-Water. While the framework ensures full network observability, the required number of sensors can be high in large and cyclic networks, which may limit applicability in real-life and motivates future work on relaxed observability guarantees for sensor placement.
Read the full abstract
Algebraic Geometry
Derivations tangent to polar discriminants
We describe the module of logarithmic derivations along the discriminant of a stable polar representation of a reductive algebraic group. It splits into derivations along the fibres of the quotient map and a free summand lifted from logarithmic derivations along the associated branch divisor. This gives a uniform extension of previous results on full adjoint discriminants. We prove that, for visible stable polar actions of semisimple groups with finite generic stabilizer, the discriminant is a free divisor. For stable periodic gradings of arbitrary order with semisimple degree-zero group, we construct explicit Saito matrices from the infinitesimal action and the gradients of basic invariants, which yields determinantal equations for several exceptional discriminants. In particular, starting with exceptional Lie algebras, we provide examples of free discriminants of degrees 60, 120, 126 and 240 for inner gradings and a degree-72 example for an outer one.
Read the full abstract
Multiagent Systems
Unprompted coordination through shared memory in LLM agents: a locked two-implementation replication and extension
Coordination between large language-model agents has become a central question for AI safety and software engineering, as such agents are deployed in groups that share infrastructure. Existing research typically tells the agents that partners exist or asks them to coordinate. Whether coordination arises through shared state alone, with nothing disclosing that anyone else is present, is the key question for deployments that share caches and memory between agents, yet it has rarely been examined, and to our knowledge not with independent replication; this study does both. Four agents each received a short arithmetic task and a key-value store presented to each as its own scratch space; two held complementary halves of one computation, and no prompt mentioned other agents, sharing or collaboration. For reliability, the environment was rebuilt independently from a written protocol, and both implementations ran under hashed locks on design, data and analysis. The results bear novelty value in three respects. Joint success occurred only through the shared store. Whether a model coordinated depended not on its vendor tier but on whether it wrote its intermediate result and looked at the store again, and one ordinary sentence, "check your workspace again before submitting", switched that second look on and off. The findings held under replication. For practice, state shared between agents is a communication channel and should be separated, namespaced and logged. For research, designs should be locked before the data, and the parse rule, endpoint and parameters reported. Future work should test open-weight models and separate using another agent's output from intending to communicate.