Peer Revenue Equity in science

Papers

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

All of arXiv

Sunday 30 August

Computer Vision

GridFlow: Structured Latent Flow for Seamless City-Scale 3D Point Cloud Generation

Generating realistic 3D city environments from remote sensing data is important for simulation, urban planning, and mixed reality, yet existing point cloud generation methods are limited to single objects or bounded indoor scenes and cannot handle the scale, seamless tiling, and partial observability challenges of city-scale generation. We present \ours{}, a multi-stage framework that generates dense, colored point clouds ($10^5$ points per $150\text{m}{\times}150\text{m}$ tile) at city scale, conditioned on satellite imagery, semantic segmentation maps, and digital surface models (DSM). A Grid-Aligned VAE encodes each tile into a topology-preserving latent grid where tokens correspond to fixed spatial regions, enabling spatially coherent multi-modal conditioning and compact latent-space edge consistency that implicitly aligns thousands of boundary points for seamless cross-tile generation. A conditional rectified flow model synthesizes geometry latents from the fused multi-modal conditions, and an orientation-aware diffusion colorizer separately handles satellite-visible horizontal surfaces and occluded vertical façades. To support standardized evaluation, we build on public 3D data sources to introduce City3D-MultiGen, a benchmark of $163$K densely annotated tiles from Melbourne and London with aligned point clouds, satellite images, semantic maps, and elevation data. Experiments show that \ours{} outperforms adapted point cloud generation baselines across all geometry metrics and produces visually coherent colored point clouds with seamless boundaries over arbitrarily large urban extents. Our benchmark details are available at https://huggingface.co/datasets/e32/City3D-MultiGen
Read the full abstract

All authors Xinyu Wang, Muhammad Ibrahim, Atif Mansoor, Ajmal Mian

Strongly Correlated Electrons

Thermal Hall Signatures of Distinct Schwinger-Boson Flux Sectors on the Honeycomb Lattice

Thermal transport offers a bulk probe of charge-neutral excitations in frustrated magnets. It remains challenging, however, to determine whether such transport can distinguish different parton mean-field structures of the same spin system. This question is particularly relevant on the honeycomb lattice, where distinct $0$-flux, $\pi$-flux, and chiral $\pi/2$-flux sectors have been proposed. In this paper, we compare self-consistent saddle-point solutions in these three sectors for a frustrated $J_1-J_2$ Heisenberg model with a next-nearest-neighbor Dzyaloshinskii-Moriya interaction and a perpendicular magnetic field. Using finite-temperature Schwinger-boson mean-field theory and the Kubo formula for bosonic Bogoliubov-de Gennes systems, we evaluate the intrinsic spinon thermal Hall conductivity. At a common parameter point where all three branches remain gapped, the $0$-flux response is positive and the $\pi$-flux response is negative for $D>0$ and $h>0$ under the sign conventions used in this paper, whereas the $\pi/2$-flux response changes sign with increasing temperature. We further show that a projective symmetry combining a sixfold rotation with time reversal forces the zero-field response of a fixed chiral $\pi/2$-flux domain to vanish. Within the mean-field regime examined in this paper, the sign and temperature dependence of $\kappa_{xy}/T$ therefore provide a flux-sensitive transport signature.
Read the full abstract

Instrumentation and Detectors

A high-speed anamorphic pupil-conjugate slit spectrograph for rapid, self-luminous sources

We present a slit imaging spectrograph whose wavelength registration is a property of the instrument rather than of the pointing of the fore-optic or the position of the source. An internal telecentric 4f relay places the system stop within the instrument while a cylindrical lens places the slit aperture at a plane where transverse position encodes source angle only (an $A = 0$ conjugate in the ABCD matrix formulation). As a result, the lateral motion of the source or the pointing of the fore-optic does not cause the spectral image illuminating the sensor to wander, thus eliminating any non-physical registration of chemical species not present in the event and the need for on-the-fly recalibration. Additionally, the use of a high-speed CMOS camera eliminates temporally averaged spectral lines and aliasing noise. We characterize wavelength calibration (He discharge lamp, quadratic fit, RMS residual 0.12 nm, verified with Ne discharge lamp, same calibration, RMS residual 0.13 nm), spectral resolution (R $\approx$ 639, FWHM 0.92 nm at the optimal slit width setting, R $\approx$ 783, FWHM 0.75 nm at the instrument profile floor), and pointing invariance (He 587.6 nm line wander 0.29 nm RMS, 1.03 nm peak-to-peak, a nearly 100x reduction relative to the same instrument with the aperture stop at the fore-optic pupil). We perform combustion spectroscopy on ignited nitrocellulose at 10 kHz, 99 $\mu$s exposure time, identifying Na, K, Ca, CaOH, and Rb chemical species using the unchanged wavelength calibration transferred from the helium discharge lamp.
Read the full abstract

Computation and Language

HiVe: Beyond Static Prompts for Multitask Learning via Hierarchy-based Vertical Mixture-of-Experts

As large language models (LLMs) continue to scale, parameter-efficient fine-tuning (PEFT) has become a practical alternative to full-parameter adaptation. Prompt tuning is effective, but existing approaches either use flat prompt structures or hierarchical structures with fixed prompt composition, limiting adaptive prompt specialization. To address this limitation, we propose HiVe, a prompt tuning framework that models prompts at multiple levels and enables input-dependent specialization. HiVe constructs a prompt hierarchy by leveraging inter-task relationships during training, and employs a vertical mixture-of-experts (V-MoE) mechanism at inference time to compose prompts up to the level of specialization required for each input. Experiments show that HiVe consistently outperforms strong prompt tuning baselines across diverse tasks.
Read the full abstract

Optimization and Control

A Unified Perspective on Conformal Prediction and Wasserstein Distributionally Robust Optimization for Uncertainty Quantification

Uncertainty quantification from finite data is central to machine learning, optimization, and automation systems, where decisions must remain reliable under limited samples and test-time distribution shift. Conformal prediction (CP) and distributionally robust optimization (DRO) offer two complementary approaches: CP constructs data-dependent prediction sets with distribution-free finite-sample validity under exchangeability, while DRO optimizes worst-case performance over an ambiguity set around an empirical distribution. We develop a unified probabilistic perspective on CP and DRO by viewing both as ways to turn finite calibration data into a data-dependent quantile estimator that a test score falls below with high probability. From this perspective, CP and DRO correct the empirical quantile along two coordinates of the same family of estimators: CP inflates the quantile level, whereas DRO shifts the quantile value through an ambiguity radius. Both methods provide the same calibration-conditional guarantee for the true distribution, requiring the target coverage to hold with high probability over the calibration sample. Their constructions differ, however: CP uses a closed-form, distribution-free level correction, while DRO uses a value-space correction whose certified radius depends on properties of the unknown distribution and additionally guarantees coverage uniformly over the ambiguity set. This distinction emerges in the tails of the score distribution. Because CP relies on sparse upper-tail order statistics of the calibration samples, its level inflation barely moves the estimator when those samples are dense near the target quantile but overshoots when they are sparse, whereas a well-chosen DRO radius corrects in value space and may avoid this overshoot.
Read the full abstract

All authors Kehan Long, Yiqi Zhao, Pol Mestres, Lars Lindemann, Nikolay Atanasov, Jorge Cortés

Number Theory

Difference of the modular function $\omega_{2}(\tau)$, revisited

Adapting the analytic method of Gross and Zagier, Roskam proved a prime-factorization formula for the norm of the difference of two level-two Weber singular moduli. Independently, Yang and Yin obtained an equivalent formula using Borcherds lifts. More precisely, the formula concerns the norm of \[ \omega_{2}\left(\frac{-1+\sqrt{d_{1}}}{2}\right) - \omega_{2}\left(\frac{-1+\sqrt{d_{2}}}{2}\right) \] for coprime negative fundamental quadratic discriminants $d_{1},d_{2}\equiv1\pmod 8$, where \[ \omega_{2}(\tau) = 2^{12}\frac{\eta(2\tau)^{24}}{\eta(\tau)^{24}}, \] and $\eta(\tau)$ denotes the Dedekind eta function. In this work, we revisit this formula from the perspective of arithmetic intersection theory and give a new proof.
Read the full abstract

Mesoscale and Nanoscale Physics

Josephson-Phase Reversal of Non-Bloch Andreev Propagation

Non-Hermitian control is difficult in solid-state systems due to fixed dissipation. Here, we show that in a spin-orbit-coupled planar Josephson junction, changing only the Josephson phase reverses the non-Bloch propagation of a low-energy Andreev band and switches its boundary accumulation between junction edges. The phase reshapes the band's spin and electron-hole composition, causing a fixed reservoir to unequally attenuate counterpropagating modes. Weak-loss theory links this phase-controlled loss imbalance to boundary-selected complex momenta, offering an in situ route to reconfigurable non-Hermitian transport in superconducting platforms.
Read the full abstract

Risk Management

Recovering Posterior Beliefs in Credit Risk: A Latent-State EM Extension of the Information-Geometric Framework

This paper develops a latent-state framework for recovering borrower-level posterior beliefs in credit-risk analysis. Creditworthiness and financial fragility are represented as latent dimensions, while observed borrower scores follow a finite Gaussian mixture model and default depends on the latent profile. Borrower-specific probabilities of default are obtained by averaging profile-specific default probabilities over the posterior distribution of latent states. A joint Expectation–Maximization procedure is used to estimate the mixture structure and the profile-specific default probabilities from observed score–default pairs. After estimation, predictive posterior beliefs are computed using the observed scores alone, thereby preserving the information available before default realization. A controlled simulation experiment evaluates the recovery of structural parameters, posterior beliefs, and borrower-level probabilities of default. Posterior distributions are interpreted as points on the probability simplex, and their recovery is assessed using both conventional error measures and information-geometric divergences. The results provide a controlled benchmark for studying the interaction between latent economic structure, posterior uncertainty, and credit-risk prediction.
Read the full abstract

Materials Science

Valley-Enabled Intrinsic Dresselhaus Spin-Orbit Coupling in Silicon

We develop a symmetry-based theory of spin-orbit-valley coupling in silicon that reveals an intrinsic source of Dresselhaus spin-orbit coupling independent of interfaces or external electric fields. Treating the valley degree of freedom as a symmetry-carrying quantum degree of freedom, we show that the Dresselhaus interaction is necessarily valley off-diagonal and that a bulk contribution proportional to the valley Pauli matrix $\tau_1$ is symmetry allowed. Tight-binding calculations yield a bulk coupling more than an order of magnitude larger than typical interface-induced spin-orbit coupling; achieving the same energy scale through the interface-induced mechanism would require electric fields roughly 50 times larger than typical fields. We further derive the symmetry-allowed spin-valley couplings generated by magnetic-field gradients and show how they account for the valley-dependent Zeeman splitting observed in micromagnet experiments. A slight tilt of the background magnetic field out of the plane produces an additional isotropic contribution linear in $B_z$, providing an experimentally accessible signature of the corresponding coupling constant. Finally, we predict a spin-independent micromagnet-induced valley splitting in the $\tau_3$ channel, which is distinct from the $\tau_{1,2}$ channels generated by alloy disorder and therefore remains robust against disorder-induced cancellation. These results establish valley symmetry as a fundamental ingredient in the spin-orbit physics of silicon and provide new mechanisms for controlling and probing spin and valley degrees of freedom in silicon quantum devices.
Read the full abstract

All authors Johannes L. P. Steinschuld, Hendrik J. Bluhm, Lars R. Schreiber, Seyed Akbar Jafari

Earth and Planetary Astrophysics

Probing Triton's Space Environment and Internal Structure: An Integrated Detection-and-Interpretation Framework

Triton, Neptune's largest moon, is a prime ocean-world target. Constraining ocean thickness, composition, and conductivity is essential for habitability assessment, but magnetic induction alone cannot resolve the thickness-conductivity degeneracy, and magnetic perturbations from Triton's space currents can obscure the internal induction signal. We present an integrated detection-and-interpretation concept linking four physically consistent calculations. Using `PlanetProfile', we construct a common radial interior structure (temperature, density, conductivity, seismic-wave speed). We then use `MoonMag' to compute the degree-one magnetic-induction response from that conductivity profile at the synodic, rotational, and orbital periods. We perform a multi-fluid `SWMF' simulation with the induced dipole as the inner-boundary condition and develop a Coulomb-gauge Poisson reconstruction to isolate space-current magnetic fields. Finally, we develop the `TritonSeis' workflow, three-dimensional seismic forward modeling plus hierarchical travel-time inversion, to constrain the ice-ocean and ocean-rock interface depths. We find that induction is substantially more sensitive to ocean conductivity than to layer thickness, and that space-current fields are comparable in amplitude to the internal induction signal. A five-station synthetic recovery test resolves both interfaces to first order, with errors of +8.4% for the ice shell and -12.5% for the ocean. Under a conservative noise assumption, the minimum detectable magnitudes are approximately 3.8-4.6 at epicentral distances of 100-1000 km. The Poisson reconstruction and end-to-end seismic recovery are, to our knowledge, the first such quantitative demonstrations for Triton. Coordinated magnetic, plasma, and seismic measurements are complementary and can break the conductivity-thickness degeneracy, providing a framework for future Triton exploration.
Read the full abstract

All authors Jiansen He, Chuanpeng Hou, Haoen Xie, Jiaqi Li, Tianhang Chen, Hong Zou, Xuzhi Zhou, Hui Li, Yan Li, Fuchuan Pang, Bingkun Yu, Hui Huang, Tong Wang

Computer Vision

InspectorGPT: A Comparative Reasoning Enhanced VLM for Comprehensive Industrial Anomaly Detection

Industrial anomaly detection is a critical component of modern manufacturing. Most traditional unsupervised methods rely on modelling normal feature distributions, inherently limiting generalization to unknown categories. To improve generalizability, some recent methods incorporate vision-language models (VLMs) for zero-shot detection via text prompts. However, we observe that reasoning-oriented post-training can cause anomaly discrimination to collapse, with some fine-tuned models performing worse than their base VLMs. Existing methods also provide only textual decisions or coarse boxes, without pixel-level segmentation. A more explicit detection principle comes from human inspection: anomalies are identified by comparing a query image with a defect-free reference. Inspired by this, we propose InspectorGPT, a VLM framework centered on comparative reasoning. Given a normal reference and a query image, InspectorGPT compares them to identify discrepancies and perform multiple inspection tasks with detailed reasoning. We internalize this capability through Chain-of-Thought (CoT) fine-tuning and Group Relative Policy Optimization (GRPO) with tailored, verifiable rewards. We further introduce InspectorGPT-Seg for pixel-level anomaly masks. Segmentation supervision improves anomaly discrimination but weakens semantic reasoning, while joint training fails to balance them. We therefore train the two branches separately and combine them through task-vector fusion. Extensive experiments demonstrate superior multi-dimensional performance and generalization to unseen benchmarks, validating comparative reasoning for comprehensive industrial inspection.
Read the full abstract

All authors Weifei Chen, Honghao Zhang, Zhiyuan You, Xinyi Le

Quantum Algebra

Poisson bialgebras by deformations-to-quasiclassical limits

Poisson algebras are the quasiclassical limits of associative algebra deformations of commutative associative algebras. This paper extends this process to the level of bialgebras. We derive Poisson bialgebras as the quasiclassical limits of antisymmetric infinitesimal bialgebra deformations of commutative and cocommutative antisymmetric infinitesimal bialgebras. It might be regarded as the “infinitesimal" version of the quantization process of Lie bialgebras in terms of Hopf algebras. Such deformations-quasiclassical limits process for a Poisson bialgebra is equivalently characterized in terms of the introduced notions of deformations of a matched pair of associative algebras as well as a standard Manin triple of associative algebras, whose corresponding quasiclassical limits are a matched pair of Poisson algebras and a standard Manin triple of Poisson algebras, respectively. We illustrate these equivalent deformations-quasiclassical limits processes via coherent derivations.
Read the full abstract

Statistics Theory

Sturm-Liouville-Type Parity and Oscillation of a Cubic Spline Eigenbasis

We study the eigen-structure of the penalty matrix arising from cubic smoothing splines on equally spaced knots. Using purely matrix-theoretic arguments, we show that its positive eigenvalues are simple, that the associated eigenvectors alternate between even and odd, and that the eigenvector for the $k$th largest eigenvalue has exactly $k+1$ sign changes. The approach provides a direct and transparent alternative to existing variational proofs of the oscillation property. These results show that equally spaced knots support a spline basis with both a parity structure and an oscillation pattern.
Read the full abstract

All authors Shih-Hao Huang, Jephian C.-H. Lin, ShengLi Tzeng, Tzu-Lun Yuan

Algebraic Geometry

Infinitesimal Deformations of Generalized Parabolic Hitchin Pairs

We develop the infinitesimal deformation theory of generalized parabolic Hitchin pairs (GPHs) on irreducible nodal curves. For any GPH $(E,\phi,F(E))$ we associate an explicit three-term deformation complex $\mathcal{C}_{(E_\bullet,\phi)}$ whose first hypercohomology $\mathbb{H}^1(\mathcal{C}_{(E_\bullet,\phi)})$ parameterizes first-order deformations. As the main application, we prove that the coarse moduli space $\mathcal{M}_{\mathrm{GPH}}$ of $1$-stable generalized parabolic Hitchin pairs of rank $n$ and degree $d$ with $\gcd(n,d)=1$ is a local complete intersection, smooth in codimension one, and hence a normal variety. We further show that this moduli space carries a natural Poisson structure.
Read the full abstract

Number Theory

Exact logarithmic Hausdorff and capacity exponents of the non-Brjuno set

Let $\mathcal{N}$ be the set of non-Brjuno real numbers. We determine the exact critical exponent of $\mathcal{N}$ for both logarithmic Hausdorff measure and logarithmic capacity. For the gauges $h_\delta(r)=(\log(1/r))^{-\delta}$, the critical exponent is $2$: the $h_\delta$-Hausdorff measure is infinite locally for $0<\delta\leq2$ and vanishes globally for $\delta>2$, while the logarithmic capacity is positive exactly for orders $0<s\leq2$. In particular, $\dim_{\mathrm{cap}}\mathcal{N}=2$.
Read the full abstract

Materials Science

Phonon-Localization-Driven Decoupling of Dual-Channel Transport for Record-Low Intrinsic Lattice Thermal Conductivity

A fundamental bottleneck in pushing the intrinsic lattice thermal conductivity of inorganic crystalline solids to its lowest limit arises from the inherent competition between the particle-like propagation (\(\kappa_{\mathrm{L}}^{\mathrm{P}}\)) and wave-like tunneling (\(\kappa_{\mathrm{L}}^{\mathrm{C}}\)) channels. Herein, we demonstrate that phonon localization provides a robust pathway to decouple the dual-channel transport, achieving record-low \(\kappa_{\mathrm{L}}\) in quasi-1D ternary helical crystals. Despite the structural complexity leading to densely populated phonon branches and thus inducing abundant coherent phonons, the weak interchain interactions and heavy elements compress numerous branches into highly localized, nearly dispersionless flat bands. Such strong localization simultaneously suppresses both the diagonal and off-diagonal components of the group velocity, thereby synergistically suppressing \(\kappa_{\mathrm{L}}^{\mathrm{P}}\) and \(\kappa_{\mathrm{L}}^{\mathrm{C}}\). Taking InSeI as an example, the interchain room-temperature \(\kappa_{\mathrm{L}}^{\mathrm{P}}\) and \(\kappa_{\mathrm{L}}^{\mathrm{C}}\) are 0.145 and 0.053 W/mK, respectively, yielding an ultralow total \(\kappa_{\mathrm{L}}\) of 0.198 W/mK. Weaker interchain interactions further drive the room-temperature \(\kappa_{\mathrm{L}}\) of GaSeI and AlSeI to record lows of 0.086 and 0.089 W/mK, respectively; these values even drop to 0.058 and 0.059 W/mK at 900 K. These findings provide useful insights into exploring the thermal conductivity limit in crystals.
Read the full abstract

All authors Zhunyun Tang, Xiaoxia Wang, Jin Li, Chaoyu He, Chao Tang, Mingxing Chen, Tao Ouyang

Statistical Mechanics

Representation Transitions Reveal Predictive Structure in Complex Systems: A Trajectory-Level Reconstruction in a Critical System

Trajectories in complex systems often contain structure that is not captured by the population-averaged, scalar statistics traditionally used to describe them. Here we reconstruct, through independent numerical re-simulation rather than literature review, a systematic search across representational forms - from single scalar statistics, through two independent discrete multi-feature classification attempts, to a trajectory-level structural representation and a diagnostic surrogate test, to a continuous low-dimensional compression - applied to the same underlying trajectory ensemble in a critical complex system. Predictive failure is not explained by insufficient correlation: one discrete representation's own strongest individual feature correlates with the target more strongly ($r=0.588$) than the eventual successful representation's own headline statistic ($r=0.540$), yet fails to separate the ensemble's two most consequential cases - while the continuous representation succeeds, robustly, across eight independent scale-removal folds ($r=0.677$-$0.717$, all $p<0.001$). The determining factor was not correlation strength but whether the representation preserved the structure relevant to prediction. These results indicate that the choice of representation determines whether predictive structure in complex systems becomes accessible - a scientific variable in its own right, not a downstream analysis choice.
Read the full abstract

Numerical Analysis

Efficient Hybrid WENO Schemes for Special Relativistic Hydrodynamics with Adaptive Characteristic Reconstruction

Special relativistic hydrodynamics (SRHD) equations arise in the modeling of high-speed fluid flows encountered in astrophysical phenomena such as jets, supernova explosions, and gamma-ray bursts. Owing to their highly nonlinear hyperbolic nature, solutions often develop strong discontinuities, making the design of stable and accurate numerical schemes challenging. Although Weighted Essentially Non-Oscillatory (WENO) schemes are widely used for such problems, component-wise WENO reconstruction may produce spurious oscillations near discontinuities. On the other hand, characteristic-wise WENO reconstruction provides accurate non-oscillatory solutions for systems of conservation laws, but it involves the computation of eigenvectors in each cell, which leads to high computational cost. In this work, we intend to develop hybrid schemes which maintain the non-oscillatory feature of characteristic-wise WENO while being less costly. We propose three hybrid schemes, namely the H1-WENO, H2-WENO, and H3-WENO schemes, based on a new troubled-cell indicator constructed from the smoothness indicators of the WENO scheme. The proposed troubled-cell indicator effectively distinguishes smooth and discontinuous regions, allowing the hybrid schemes to employ inexpensive reconstructions in smooth regions and the characteristic-wise WENO reconstruction only near discontinuities. Numerical experiments demonstrate that the proposed schemes retain the accuracy and robustness of characteristic-wise WENO methods while significantly reducing the computational cost. In particular, the H1-WENO scheme achieves an approximately 30–40% improvement in computational efficiency compared to the standard WENO scheme in 2D test cases.
Read the full abstract

All authors Rakesh Kumar, Biswarup Biswas, Asha Kumari Meena, Harish Kumar

High Energy Physics (Phenomenology)

X(2370) as the pseudoscalar glueball: Another clue from a constituent approach

The pseudoscalar state X(2370) is the clearest glueball candidate experimentally identified so far. From a phenomenological point of view, constituent approaches, in which the lightest glueballs are modelled as bound states of two transverse constituent gluons linked by an adjoint flux tube, reach a good agreement with quenched lattice QCD pure gauge spectrum. Based on the assumption that the total decay width of glueballs or light unflavoured mesons is proportional to the energy stored in the flux tube, a model is developed that linearly correlates the hadron decay width to its mass. The model is first fitted on light unflavoured mesons and then extended to glueballs. Both the mass and decay width of the X(2370) are compatible with the results of the constituent approach. Other experimental glueball candidates in the scalar and tensor channels are also reviewed to show the ability of the developed approach to identify glueballs: The f$_2$(2150) and f$_2$(2300) states appear as promising tensor glueball candidates.
Read the full abstract

Physics and Society

Peer Review at Capacity: An editor's view

From the editor's desk, the pressure on peer review is no longer an abstract concern. It appears in the growing number of invitations needed to secure an appropriate report, the repeated reliance on a small group of dependable colleagues, and the duplication of work when manuscripts move between journals. The problem is not simply that science has too few potential reviewers. It is that reviewing labor is unevenly distributed, weakly recognized, and organized through a publication architecture built for a smaller system.
Read the full abstract

Cryptography and Security

HSMLog: Small Language Model-Assisted Hardware Security Module Log Anomaly Detection with Behavioral Analysis

Hardware Security Module (HSM) logs capture security-critical behavior, but anomalies emerge from relationships across event sequences, keys, object states, sessions, and temporal patterns rather than isolated events. Existing methods separate detection from HSM-specific evidence validation and reporting. In this paper, we present HSMLog, a two-stage framework for HSM log anomaly detection with retrieval-grounded behavioral analysis. In Stage 1, a small language model (SLM) identifies candidate alerts from sliding windows of structured HSM events and performs policy-guided assessment using HSM-specific operational rules. In Stage 2, retrieved policies and historical suspicious-key records strictly predating the alert window, together with candidate-related log context, support conservative candidate review and incident analysis. Evaluated on real industrial HSM background logs augmented with anomaly scenarios co-defined with industrial partners, HSMLog achieves 98.97% precision, 96.00% recall, 98.66% anomalous-event coverage, and a 97.46% F1 score, demonstrating effective anomaly alerting and incident triage in the studied setting.
Read the full abstract

All authors Chia-Hsuan Wu, Dar-Hsin Dustin Wu, Rui Fang, Yi-Ting Lee, Chia-Chih Lin, Ming-Syan Chen

Robotics

Self-Aware Active Learning Enables Continual Improvement in Autonomous Driving

Learning-based autonomous driving (AD) systems can perform reliably in familiar conditions, yet rare distribution shifts and long-tail events remain a major source of abrupt failure. A central limitation is that most agents learn primarily from passive experience and lack mechanisms to estimate when their competence is insufficient, seek timely assistance, and convert safety-critical encounters into targeted improvement. Here we present self-aware guided exploration (SAGE), an active learning framework for post-training adaptation in AD. SAGE learns a predictive world model that generates two online intrinsic signals: fear, which estimates short-horizon predictive risk and model uncertainty, and curiosity, which measures novelty through prediction error. Curiosity adaptively calibrates the intervention threshold for fear, allowing the agent to regulate risk in a context-dependent manner. When predicted fear exceeds this adaptive threshold, the agent transfers control to an expert or fallback policy and uses the resulting takeover trajectories for focused imitation learning. In parallel, fear is integrated into policy optimization and evaluation as a safety-oriented constraint to reduce performance regressions during adaptation. We evaluate SAGE in simulated route-transfer tasks, Waymo-based logged driving scenarios, CARLA occlusion hazards, and real-world mobile robot navigation tests. Across these settings, SAGE improves robustness in novel and safety-critical scenarios, reduces safety violations, and maintains task performance comparable to strong baseline policies. These results suggest that agents can improve after initial training by estimating the limits of their competence, requesting guidance when needed, and learning selectively from rare high-value events.
Read the full abstract

All authors Dong Hu, Chao Huang, Carman K.M. Lee, Dimitrios Kanoulas

Superconductivity

Magnetic contacts on freestanding superconducting LaAlO$_3$/SrTiO$_3$ micromembranes

The superconducting two-dimensional electron gas (2DEG) at the LaAlO$_3$/SrTiO$_3$ (LAO-STO) interface is a promising platform for superconducting spintronics, however, integrating ferromagnetic contacts with the superconducting 2DEG remains challenging. Here, we realize superconducting LAO-STO micromembrane devices contacted by ferromagnetic nickel contacts through a side-contact geometry. Low-temperature transport measurements demonstrate that superconductivity is preserved in the presence of the ferromagnetic contacts. We show that the superconducting state is strongly influenced by the magnetic history of the nickel contacts, which generates a tunable effective magnetic field in the 2DEG. Through an effective field model, the magnetization of the contacts can be inferred from the maximum superconducting response. Our results establish ferromagnetically contacted LAO-STO as a platform for future investigations of spin injection into oxide superconductors and provide a route towards superconducting spintronic devices based on complex oxide interfaces.
Read the full abstract

All authors Thies Jansen, Pinelopi Konstantinopoulou, Niklas Martin, Fabio Miletto Granozio, Alessia Sambri, Rasmus Bjørk, Felix Trier, Thomas S. Jespersen

Robotics

Sampling-based Certified Planning with Graphs of Convex Sets

Planners on graphs of convex sets return trajectories that are collision-free by construction, provided the convex regions are collision-free. The region generator only promises that property probabilistically, and no planner in the family verifies it. We report the first measurement of what the gap costs. On a scaled 14-DOF bimanual library, $3.2\%$ of interface samples are in collision, and a search-based GCS planner (\gcsstar) turns that volume error into a $62\%$ answer error: $18$ of $29$ pick-and-place queries return trajectories that drive the arms through the shelves, up to $91$\,mm deep, reported as successes. Repairing the library does not work; a ten times stricter acceptance contract, sums-of-squares certified regions, and uniform margins each destroy the connectivity planning needs before they deliver soundness. We instead build a planner that certifies its answers. It samples the overlaps and shared faces of the decomposition, prunes with an admissible informed bound, and verifies the one candidate each search round proposes, continuously, by a chain of clearance certificate balls with no resolution parameter; failures are repaired with local in-region detours, and the convex polish is re-verified. Head-to-head on all $29$ task queries it delivers zero invalid answers against $21$ for the reference, reaches its first certified answer in $0.11$\,s against $1.59$\,s for the reference's unverified one, and reproduces the reference optimum exactly on every query whose reference answer is physically valid.
Read the full abstract

Robotics

Learning Agile Perceptive Traversal of Sparse 3D Structures for Humanoids

Traversing sparse 3D structures requires humanoid robots to perceive thin, overhanging geometry while executing agile, accurate whole-body motions. We study this problem through monkey-bar traversal, where the robot must jump to the structure, traverse it through sparse bar interactions, and land safely. For this task, we present a reinforcement-learning-based perceptive control system that operates directly on observations from a head-mounted solid-state lidar. To extract task-relevant geometry from the sparse returns, the policy consumes the raw lidar scan through an attention-based encoder with recurrent memory. This policy is obtained by a phase-scheduled teacher- student pipeline that combines privileged experts for jumping up, brachiating, and jumping down. For transfer to hardware, we model lidar noise, battery-voltage sag, and actuator thermal limits, and equip the humanoid with passive hook end-effectors for robust bar interaction. On hardware, the resulting policy completes the full jump-up->brachiation->jump-down sequence in 14 of 15 trials across three bar configurations and reaches brachiation speeds up to 0.5 m/s. Beyond brachiation, the same perception backbone supports a separately trained policy that ducks beneath thin overhead obstacles with 2 cm cross-sections.
Read the full abstract

All authors Efe Ongan, Chong Zhang, Boyang Sun, Andrei Cramariuc, Cesar Cadena, Marco Hutter

Robotics

SmoothRL: Online Reinforcement Learning During Asynchronous Execution

Deploying robot policies in the physical world requires satisfying two fundamental desiderata: reliability and smooth real-time execution. However, deploying state-of-the-art generalist models presents challenges on both fronts. Achieving the precision and robustness required for real-world deployment necessitates sample-efficient online reinforcement learning (RL) to adapt pretrained models. Meanwhile, the increasing scale of robot foundation models has led to higher inference latency. To satisfy real-time constraints under high latency, modern systems adopt asynchronous inference with action chunking, overlapping policy computation with chunk execution to hide latency and enable smooth control. Despite their complementary roles, integrating asynchronous execution with gradient-based online RL remains underexplored. We present SmoothRL, an online RL framework that fine-tunes a pretrained policy within an asynchronous inference loop. SmoothRL follows a value-gradient paradigm, directly updating policy parameters using gradients of the action-value function with respect to policy actions. To enable correct optimization under asynchronous execution, SmoothRL explicitly models the asynchronous inference process during training. Specifically, each generated action chunk is partitioned by frame index into three regions: a committed region, consisting of actions committed by the previous inference cycle; an execution region, containing newly generated actions executed by the robot; and a discarded region, containing actions superseded by the next inference cycle. Gradients are propagated only through the execution region, ensuring policy optimization aligns with the trajectory distribution induced by asynchronous execution. We evaluate SmoothRL on real-world robotic tasks requiring high precision, as well as highly dynamic tasks that necessitate asynchronous execution.
Read the full abstract

All authors Guang Gao, Yuxuan Nong, Baifu Huang, Jianan Wang

Robotics

LARC: Lazy Adaptive Reachability Certification of Robot Manipulator Trajectories

Discrete trajectory checks can miss collisions between sampled robot states. Reachability-based certification bounds motion between states, but uniform time partitions waste computation where clearance is large. We present lazy adaptive reachability certification (LARC), which checks a planned trajectory by bisecting only intervals with an inconclusive clearance test. For piecewise-cubic Hermite joint trajectories, the method bounds link occupancy using midpoint capsules inflated by exact componentwise speed maxima. Certified intervals covering the trajectory provide continuous-time external-obstacle clearance, subject to geometric containment, static obstacles, and a prescribed margin. On 160 AgileX PIPER trajectories from 80 start-goal pairs, LARC matched all decisions of the fixed-fine baseline at depth nine. It used 20328 interval evaluations (24.8% of baseline work), with a median paired speedup of 10.28x. A separate MoveIt/FCL audit checked 158051 states and detected collisions in 21 direct-interpolation controls, none of which LARC certified. The method reduced computation under a shared certificate model, but 27 of 139 sampled-clear trajectories remained uncertified. The sampled audit cannot independently prove continuous-time clearance.
Read the full abstract

All authors Yu Feng, Hao Wu, Yuzhe Wang, Jianshu Zhou

Mesoscale and Nanoscale Physics

Localized orbitals and tunnel couplings from general confinement potentials in gate-defined quantum-dot arrays

Efficient simulation of dense gate-defined multi-quantum-dot arrays requires accurate and scalable modeling methods, compatible with asymmetries and imperfections of realistic voltage-controlled confinement potentials. We present a numerical localization procedure that rotates the eigenbasis of a general one-particle effective orbital Hamiltonian into $s$-, $p$-, $d$-, $\ldots$-shells of localized orbital wavefunctions associated with individual quantum dots. The pairwise tunnel couplings between such states are computed directly as matrix elements of the Hamiltonian. We demonstrate this procedure on a 2D triangular Si-MOS triple-quantum-dot array by obtaining the voltage dependencies of the tunnel couplings and their distributions in the presence of disorder. We discuss the implications of this evaluation method on the many-body calculations, and relate it to the experimental tunnel coupling measurements.
Read the full abstract

Machine Learning

PruneShift: A Framework for Evaluating Decision Reliability in Structured Pruning

Structured pruning uses surrogate objectives because direct task evaluation over every feasible mask is too expensive. Most evaluations report average surrogate error or rank correlation on broadly sampled masks. These summaries do not directly test the mask chosen by the surrogate. We introduce PruneShift, an evaluation framework that separates broad predictive fidelity, fidelity near selector outputs, and the quality of the selected pruning decision. We first prove that Spearman and Kendall agreement can approach one while normalized selection regret remains maximal. We then derive sufficient conditions based on uniform error, selector suboptimality, decision margin, density ratio, and comparison mass. The analysis also yields a finite pool certificate with an explicit excess cost bound. Four studies test different links in this argument. External TextbookQA confirmation is heterogeneous: 7 of 20 simultaneous intervals favor the surrogate-selected mask, 6 favor its fixed comparator, and 7 cross zero. On a fixed Natural Questions pool, strict improvement holds in one of four settings. A controlled QQP experiment supports the proposed coverage mechanism in all 16 prespecified endpoints, although the sufficient bounds are conservative. Finally, a restricted OSSCAR reconstruction study on OPT-125M shows better local than broad fidelity in 68 of 75 primary endpoints. Independent fixed-mask confirmation is inconclusive in 24 of 25 endpoints and favors the comparator in one. These results show why predictive fit, decision reliability, and pruning method quality require separate evidence.
Read the full abstract

Computational Complexity

The Complexity of Coverability-Like Problems in Elementary Object Systems: Data-Nets to the Rescue

Elementary Object Systems (EOSs) are a model in the nets-within-nets (NWNs) paradigm, where tokens in turn can host standard Petri nets. We study the complexity of coverability-like problems, including termination and boundedness, over EOSs. Since coverability and boundedness are undecidable in general on EOSs, we focus on the relevant fragment of conservative EOSs (cEOSs). Our technique interprets cEOSs into the framework of data nets, whose tokens carry data from an infinite domain, thus bridging the nesting and the data-aware paradigms. Specifically, we show that cEOS coverability-like problems are equivalent to the coverability-like problems over an interesting fragment, called channel-$\nu$PNs (c-$\nu$PNs), of data nets that extends $\nu$PN (featuring globally fresh name creation) with restricted forms of transfers with renaming. c-$\nu$PNs remain less expressive than Unordered Data Nets, which feature lossy name creation as well as powerful forms of whole-place operations and broadcasts. These reductions allow us to analyze cEOS coverability taking advantage of known results on data nets. We conclude that the complexity of cEOS coverability is double-Ackermanian, $\mathcal{F}_{\omega 2}$-complete, while termination and boundedness are non-primitive recursive.
Read the full abstract