Papers

Clinical CT and UHRCT cannot resolve individual trabeculae, whereas synchrotron radiation microCT (SRuCT) provides high-resolution references but is not applicable for in vivo imaging. The two domains differ by a 32x resolution gap, are only coarsely paired, and exhibit severe physical differences including partial volume effects, noise, and artifacts. Existing super-resolution networks and pretrained-prior methods (GLEAN/StyleGAN2, Stable SR/LDM) underperform because they target pixel generation—diverse details and SSIM/PSNR—and do not model these physical differences. Pixel generation for a 32x resolution gap is intrinsically ill-posed. We propose a paradigm shift from pixel generation to topological inference: deterministically predicting invariant microstructures from macro-scale low-resolution inputs, evaluated by morphological parameters. The core of our 2D morphology learning lies in training on 2D slices while evaluating on 3D morphological parameters, ensuring that the learned representations capture true three-dimensional trabecular topology rather than 2D pixel statistics. We realize this paradigm via structural dual super-resolution, coupling forward physical degradation (micro-to-macro) with inverse structural inference (macro-to-micro) through structural duality constraints. The method is an end-to-end, few-shot, compact structural dual network (SDN), comprising a bidirectional modeling network, a multi-scale structural consistency discriminator, and four structural duality constraints. On the test set, SDN achieves morphological parameters largely consistent with SRuCT across six metrics, with SSIM reaching 0.8. Trained on 3.2 um SSRF data, the model generalizes well to 3.25 um isotropic BSRF data from an independent source, validating cross-source generalization and confirming trustworthy structural inference rather than pixel generation.

Physical trajectories contain more than snapshots of a system: they also reveal how its states evolve under governing conditions. However, representation learning for parametric partial differential equations (PDEs) has largely relied on reconstruction-based objectives that emphasize recovering observed physical fields. In this paper, we investigate predictive representation pretraining as an alternative to reconstruction-based learning. We find that predictive representations preserve rich physical information, yet this advantage alone does not ensure accurate field evolution. Based on these observations, we introduce PDE-JEPA for parametric PDE dynamics. Specifically, we first train an encoder using a masked-latent prediction to capture the underlying regularities of PDE dynamics. To explicitly adapt the pretrained representation toward a more dynamics-aligned state space, we then introduce a geometry projector that aligns latent trajectory geometry with the evolution geometry of physical fields. Finally, building on this geometry-aligned latent space, we further develop a physics-structured latent predictor that decomposes the dynamics into parameter-independent evolution and parameter-dependent response components. Extensive experiments on nine widely used PDE benchmarks demonstrate that our framework outperforms existing state-of-the-art methods by an average of 33.4% in-distribution, while achieving an average improvement of 51.4% when extrapolating to unseen governing parameters. The project page is available \href{https://tanpig-x.github.io/PDE-JEPA/}{here}.

As LLMs increasingly act as delegated agents, they are expected to protect principals' interests when interacting with external parties. Standard alignment objectives, such as helpfulness, harmlessness, and honesty, do not specify how agents should protect principals' strategic interests under delegation. We formalize Agent Loyalty as an information-control property requiring agents to prevent Exploitable Information Leakage (EIL) and resist Manipulative Information Uptake (MIU). We introduce LoyalAgent-Bench, comprising 10,298 samples across 42 subscenarios and six domains, and an online GRPO framework that trains against a LLM opponent to generate mechanism-specific reward signals. Experiments show that loyalty is not guaranteed by general capability or existing alignment, with measurable EIL and MIU gaps under zero-shot evaluation, while our trained 8B models reduce per-response leakage in single-turn exchanges by 31-44pp and improve task utility, evidence faithfulness, and decision accuracy by up to 11pp, 49pp, and 29pp, respectively. For the trained Qwen3-4B model, no degradation is observed on out-of-distribution benchmarks in math and narrative reasoning.

Helium dimer in its metastable triplet electronic state, He$_2(a~^3\Sigma_u^+)$, is a promising candidate for direct laser cooling and precision measurements. Interactions and collisions involving He$_2(a~^3\Sigma_u^+)$ remain unexplored in ultracold and cold energy domains. Here, we present the first theoretical investigation of intermolecular interactions and collision dynamics of He$_2(a~^3\Sigma_u^+)$+Ne van der Waals complex. For the ground electronic state, $X~^3\mathrm{A}'$, we employ a spin-restricted coupled cluster method with single, double, and noniterative triple excitations. The resulting interaction potential exhibits an extremely shallow well with a depth of 4.612 cm$^{-1}$. We construct a three-dimensional vibrationally averaged potential energy surface. Using this surface, we perform full close-coupling calculations including fine-structure levels of He$_2(a~^3\Sigma_u^+)$ to investigate collisional excitation and quenching induced by Ne. We analyze the low-energy scattering dynamics and model resonant features associated with quasi-bound states supported by interaction potential. Propensity rule for fine-structure conserving transitions is established. We additionally consider He$_2$+Ar collisions and compare their energy dependence and resonance structure with those of He$_2$+Ne. The He$_2$+Ar system exhibits a richer resonance structure, highlighting the sensitivity of low-energy scattering to interaction potential and reduced mass. Calculated PES and scattering properties provide a reliable benchmark for description of He$_2$+Ne van der Waals complex and may be useful for interpretation of future experimental spectra and collision measurements.

Practical deployment of large language model (LLM) agents requires strong task performance at affordable inference cost. For long-horizon agentic tasks, this performance-cost trade-off can be improved through within-task large-small model collaboration, as smaller models can handle some stages even when they cannot solve the full task. In this paper, we introduce RSI-router, a routing framework that constructs subtask-level model assignments and model-specific skills through recursive self-improvement over accumulated experience. Each iteration consists of four stages: Subtask Mining derives subtask definitions and identification rules from training trajectories; Routing Strategy Evolution proposes and evaluates diverse model assignments; Model-Specific Skill Evolution compares routed and large-model-only trajectories to diagnose failures and develop reusable execution skills; and Pareto-Optimal Router Selection updates the Pareto population using historical and newly generated routers while retaining dominated routers as experience for subsequent evolution. Routing between DeepSeek-V4.1-Flash and Qwen3.5-9B, RSI-router consistently surpasses the DeepSeek-only baseline at roughly half the inference cost (48.3%) across five agentic benchmarks. In particular, on ALFWorld, ScienceWorld, and WebShop, it cuts inference cost by 74.7-82.2% while simultaneously improving performance; on Terminal-Bench 2.0, it achieves a 16.7% relative performance gain at 18.0% lower cost. Moreover, RSI-router establishes a stronger performance–cost Pareto frontier than 9 routing methods.

Learning local geometry enables graph neural networks (GNNs) to adapt how they compare and integrate neighborhood information. However, estimating geometry from aggregated representations can overlook variation among individual messages and dependencies across feature dimensions. We propose GeoF, a recurrent framework that jointly evolves node features and propagation geometry through message-passing feedback. Each node maintains a local symmetric positive-definite geometry, initialized from a structure-aware prototype atlas and parameterized in block log-triangular coordinates. At each step, the geometry determines neighborhood weights, while triangular frame transport maps transformed source messages into the target node's local coordinates before aggregation. Weighted second-order statistics of residuals between aligned messages and the transformed target state capture directional variation and within-block dependencies, yielding a geometric update target. A shared controller learns complementary corrections through task supervision. A bounded log-triangular update combines these corrections, the target, and the previous geometric state while preserving positive definiteness. The geometry governs subsequent propagation, closing the feedback loop. With parameters shared across recurrent steps, task-specific readouts support node classification, link prediction, and graph classification. Experiments on benchmark datasets show that GeoF consistently outperforms state-of-the-art GNN baselines.

Long-horizon agent benchmarks typically report how far an agent progresses, but do not identify whether its performance comes from the foundation model, scaffold, responsibility scope, match-control granularity, or horizon. We introduce FromPitch2Board, a deterministic football-management benchmark that studies five configurable factors through controlled comparisons on a single simulator, using paired seeds and a frozen calibration. We evaluate four foundation models and four agent scaffolds. In the Model Track, Coach points Z-scores span 0.19, while Manager points Z-scores span 0.68, with GPT-5.6 showing a sharp rise in passivity under responsibility expansion. Its responsibility ladder rises from 46.1 to 58.1 points with recruitment, then falls to 46.8 under full management, localizing the regression to the final responsibility boundary. Across that boundary, its skipped-decision rate rises from 1.1% to 57.9%. Within the Flash-Pro pair crossed across every scaffold, scaffold choice changes Manager points Z-scores by up to 0.48 relative to the fixed stateless scaffold. The 3Y cohort shows a directional reversal in mean ranking between years one and three, while a selected Claude Code+Pro configuration peaks in year three and remains below that peak, showing that responsibility scope and horizon expose behavior changes that a single headline score conceals.

Let $\mathcal{X}$ be a projective, geometrically irreducible, nonsingular algebraic curve of genus $g\ge2$ over an algebraically closed field of characteristic two. We classify cyclic subgroups of $\Aut(\mathcal{X})$ of order $N\ge2g+1$. Besides Kummer extensions of odd degree ramified over three points, precisely two cases occur: $y^2+y=x^m$, with $N=2m=4g+2$, and $y^2+y=c/(x^m+1)$, with $m$ odd, $c\ne0$, and $N=2m=2g+2$. In particular, $4\nmid N$. We then classify groups $H$ with a cyclic subgroup of index two and $|H|>4g+4$. They are precisely the groups $C_k\times D_{2M}$ on the curves $y^k=x^M+x^{-M}$, where $k,M\ge3$ are odd and coprime, and $2g=M(k-1)$. Their possible orders are $4g+2M$, where $M$ runs over certain odd divisors of $g$, and $|H|\le6g$. No curve in this family is ordinary. Hermitian curves occur in the family, and we give two sufficient conditions for maximality over finite fields. For dihedral groups the sharp bound is $4g+4$ in even genus and $4g$ in odd genus. The equality cases are given by explicit Artin–Schreier equations, and in each case the dihedral group is the full automorphism group.

Operational technology systems face exposure to scanning and protocol-specific attack tools, where compromise risks disrupting physical processes rather than just data. Traditional OT defenses rely on blocking and filtering under strict patch constraints, while tarpitting delays scanners through sustained protocol-level interaction. Modbus TCP and IEC-104 carry no native authentication or integrity protection. Application-layer tarpitting, which stalls scanners inside a legitimate protocol exchange, has been studied for IT and IoT protocols, but not for OT/ICS, whose session semantics (state machines, exception codes) create stalling opportunities IT/IoT lack, and whose patch-constrained environments need exactly this kind of alternative defense. We present StallGrid, to our knowledge the first application-layer tarpits for OT/ICS, stalling scanners via Modbus Exception Codes 0x05/0x06 and prolonged residence in IEC-104's connected state machine. Five variants (three Modbus TCP, two IEC-104) ran simultaneously for 24 days online, logging 6,709 sessions, 2,039 cumulative per-tarpit unique IPs, and over 2,000 hours of accumulated connection engagement. GreyNoise enrichment attributes 87–97% of stall time to malicious-tagged IPs, just 22–30% of connecting addresses; protocol-level behavior further correlates with malicious classification, a fingerprinting signal beyond raw stall time. Shorter induced delay (1.5s) yielded more total engagement than longer delay (3s), observed across both protocols. These results position protocol-native tarpitting as a practical, low-cost complementary defense for OT/ICS environments where patching remains infeasible.

Object navigation requires an embodied agent to find an object in an unseen environment under partial observability and a limited motion budget. Existing methods primarily optimize where the robot should go next by ranking candidate destinations. In contrast to these methods, we present SOR-Nav, a hierarchical navigation system that explicitly arbitrates between continuing to explore the current context and abandoning it for a more promising reachable region. First, an autonomous semantic exploration system is built that accumulates persistent 3D object clusters and organizes reachable frontiers into a cluster decision graph to provide an efficient search abstraction. Then, SOR-Nav uses a context-gated LLM-driven object-search supervisor to evaluate the suitability of the current search context and decide whether to continue exploration or perform cross-region relocation to another reachable frontier cluster. Across the complete, unfiltered validation sets of HM3D-v1, HM3D-v2, and MP3D, SOR-Nav achieves the strongest reported Success Rate (SR) and Success weighted by Path Length (SPL) on all three benchmarks. On MP3D in particular, it more than doubles the previous best SPL from 18.1% to 38.5% while increasing SR from 50.7% to 61.8%. Nested HM3D-v2 ablations validate the proposed decision structure, while a continuous three-target physical deployment demonstrates persistent ObjectNav operation in real-world scenarios.

Topological magnetization textures play a central role in modern magnetism. For many applications, spin textures with different topological charges $Q$ in the same system are particularly attractive. Although the coexistence of skyrmions and antiskyrmions ($|Q| = 1$) has been reported in inversion-symmetric magnets, extending this to high-$Q$ ($|Q| > 1$) textures with arbitrary charge remains elusive. Here, using an atomistic spin model parameterized from first-principles calculations, we predict the emergence of sub-20 nm high-$Q$ textures in Janus monolayers, van der Waals magnets of growing interest. We explore skyrmion and antiskyrmion rings as well as skyrmion bags with $|Q|$ up to 5, and characterize their nucleation mechanisms, thermal stability, and collapse pathways. We find that higher-order spin interactions (HOI), which extend the conventional bilinear-exchange Hamiltonian, are essential for stabilizing these high-$Q$ spin textures. The rings remain thermally stable at zero magnetic field. HOI substantially enhance their energy barriers while leaving their size nearly unchanged and stabilize them even in the absence of Dzyaloshinskii-Moriya interaction. Skyrmion bags, in contrast, nucleate only in the presence of HOI. In particular, the four-spin three-site interaction is the key ingredient preventing high-$Q$ textures from collapsing into the ferromagnetic state. Finally, we identify previously unreported parity-dependent collapse mechanisms for high-$Q$ textures. Our results establish HOI as an overlooked mechanism for high-$Q$ nucleation.

The Doppler velocity and bisector of photospheric lines show significant center-to-limb variation. This variation has been considered to be caused by the projection angle of the granular convection which has a corrugated surface. This study aimed to show the spectral line shifts up to limb and investigate how these shifts are caused by granular convection. We analyzed the Fe I 630.15 nm absorption line by spatially resolving the granulation very close to the limb using datasets of polar observations obtained with the Spectro-polarimeter (SP) of the Solar Optical Telescope (SOT) onboard Hinode. We also analyzed the Fe I 630.15 nm emission line, which was observed within 1" above the limb. This study revealed the following: 1) the spatially averaged spectral lines showed an increase in redshift toward the limb, then became constant, and finally became nearly zero, 2) the Doppler velocities of granules, which were classified as bright and non-magnetic regions, showed as increase in redshifts toward the limb, 3) the bisector of granules changed more than that of the intergranules, and 4) the emission lines showed small blueshifts. These results indicate that the increase in redshift toward the limb is caused by the increase in the redshift of granules. Because the bisector reflects the atmospheric velocity along the line-of-sight, our observations can be interpreted as suggesting that the line-forming region in granules is geometrically thicker than that in intergranules.

Downlink communications from base stations (BSs) to unmanned aerial vehicles (UAVs) in sixth-generation (6G) networks require precise beam alignment to overcome severe mobile communication path loss. However, traditional exhaustive beam sweeping relies on discrete codebooks and consumes valuable air-interface resources for online measurements, rendering it inefficient for highly dynamic aerial environments. In this paper, we propose BeamCKMDiff, a physics-informed generative diffusion framework designed to construct high-fidelity continuous beam-aware channel knowledge maps (BeamCKMs). Unlike existing empirical approaches, BeamCKMDiff employs a diffusion transformer (DiT) architecture featuring a dual-path conditioning mechanism. It integrates an analytical line-of-sight (LoS) beam prior with environmental topologies, while simultaneously embedding continuous beamforming vectors through an adaptive layer normalization (adaLN) mechanism. Building upon this differentiable generative mapping, we propose a CKM-based end-to-end continuous beamforming and proactive dual-BS handover algorithm. By analytically propagating gradients through the reverse diffusion process, the proposed framework enables successive convex approximation (SCA)-based beamforming optimization in the computational domain, effectively bypassing physical pilot scanning. Simulation results demonstrate that BeamCKMDiff achieves a normalized mean square error (NMSE) of –23.96 dB, establishing a highly reliable spatial prior. Compared to discrete beam sweeping and LoS-assumed baselines, the proposed CKM-based beamforming adapts to non-LoS (NLoS) conditions, providing accurate beam alignment and higher spectral efficiency. Furthermore, the integrated dual-BS handover mechanism eliminates blockage-induced link outages, ensuring robust connectivity for high-mobility aerial platforms.

While Deep Neural Networks (DNNs) have achieved remarkable progress in cutting-edge domains, their inherent brittleness has become a growing concern. To ensure the reliability and safety of DNN-enabled software, DNN testing has emerged as an indispensable practice. Within this context, test input prioritization is essential for early fault detection and reducing labeling costs. However, it remains challenging to accurately identify failure-inducing inputs. Although decision ambiguity and distributional atypicality are two widely adopted perspectives for characterizing inter-class competition and intra-class typicality respectively, relying on either perspective in isolation inevitably introduces blind spots. In this paper, we propose DuFP (Dual perspective Feature space Prioritization), a KNN density-based test input prioritization approach for DNNs that jointly incorporates both inter-class and intra-class perspectives. The prioritization framework of DuFP is built upon class-conditional density estimation. Based on the estimation results, prediction correctness is characterized by an ambiguity score and an atypicality score, with the former reflecting decision ambiguity and the latter quantifying distributional atypicality. A hybrid uncertainty score is then constructed by integrating both scores to guide the final prioritization. We evaluate DuFP on prioritization and selection tasks across image and text datasets under clean, corrupted, and adversarial scenarios. Experimental results demonstrate that DuFP effectively and efficiently prioritizes fault-inducing inputs and outperforms state-of-the-art approaches.

Future Mars missions will require rover autonomy that can operate across unstructured terrain, changing illumination, atmospheric dust, and limited communication. Simulation is a practical way to study these conditions before deployment, but existing Mars-relevant resources differ in scope, including mission-oriented simulators, fixed analog datasets, task-specific environments, and open robotics interfaces. In this context, we present MarsLab, an open-source, ROS2-native Mars rover simulator for autonomy and navigation algorithm development. MarsLab combines HiRISE-derived and procedural terrain with customizable rock, crater, solar-illumination, and atmospheric-dust settings, and runs a Perseverance-class rover model in NVIDIA Isaac Sim. The runtime publishes RGB, depth, RGB-D point clouds, LiDAR, IMU, wheel odometry, and Ground Truth (GT) pose data through standard ROS2 topics. We demonstrate MarsLab with Simultaneous Localization and Mapping (SLAM) benchmarks across sensing modalities, dust levels, scene geometry, and route length, and with Visual Place Recognition (VPR) benchmarks over repeated Mars Base traversals under illumination and dust changes. The results illustrate how controlled scene variation and shared GT trajectories can be used to compare trajectory-level estimation and image-level place recognition within the same simulator. Our Project Page: https://kimhoyun-robotair.github.io/MarsLab/.

Multi-agent systems (MAS) are increasingly used to automate enterprise workflows involving multiple specialized agents, external tools, and long-running task execution. Failures may arise from tool degradation, context propagation errors, coordination breakdowns, or repeated agent interactions that prevent task completion. While existing observability frameworks provide traces and logs, diagnosis and remediation are largely performed after execution completes, limiting opportunities for recovery during runtime. We present ResonAct, a runtime self-healing framework that enables continuous monitoring, diagnosis, and remediation of multi-agent systems through streaming operational metrics. ResonAct ingests execution traces, agent interactions, and tool invocations into a streaming analytics layer that continuously derives task progress, context health, and tool reliability metrics. These metrics serve as runtime control signals for detecting anomalous execution patterns and localizing root causes using a structured failure model. Based on the diagnosed failure, ResonAct dynamically selects remediation policies and performs actions. The framework operates as an external control plane, enabling intervention without modifying application agents or orchestration logic. We evaluate ResonAct across enterprise workflow scenarios and AppWorld benchmarks. The results show that the streaming metric-based analysis identifies execution degradations and localizes faults. Furthermore, policy-driven remediation improves task completion rates by up to 10.00 percentage points, with detection precision ranging from 70.59% to 82.91%, recall from 63.09% to 100%, recovery rates from 10.48% to 46.67%, and runtime overhead ranging from $-0.25%$ to 14.12% across the evaluated configurations.

For the fractional heat equation $\partial_t u + (-\Delta)^{\beta/2}u=0$ on $\mathbb{R}^d$ with $\beta\in(0,2)$, Weber and Zacher introduced the optimal constant $C_{\rm LY}(\beta,d)$ in the Li–Yau type estimate $$(-\Delta)^{\beta/2}\log u(t,\cdot)\leq \frac {C_{\rm LY}(\beta,d)}{t},\quad t>0.$$ They asked whether $C_{\rm LY}(\beta,d)$ converges to the classical Li–Yau constant $d/2$ as $\beta\uparrow2$. In this note we prove that the answer is affirmative. More precisely, for every fixed $d\geq1$ there exists a constant $K_d>0$ such that $$\frac{d}{\beta}\leq C_{\rm LY}(\beta,d)\leq\frac{d}{\beta}+K_d(2-\beta)\log\frac{e}{2-\beta},\quad 1\leq\beta<2.$$ Consequently, $\lim_{\beta\uparrow2}C_{\rm LY}(\beta,d)=d/2$. The proof has two main ingredients. First, the Bochner subordination formula, together with a concavity argument, shows that the supremum in Weber–Zacher's formula for $C_{\rm LY}(\beta,d)$ is attained at the origin. Second, the factor $2-\beta$ is derived from a lower bound for the fractional heat kernel profile in the large-distance regime, where the Lévy–Itô decomposition plays an important role.

Quantum conditional mutual information (QCMI) is a fundamental measure of conditional independence, with central applications throughout quantum information theory and many-body physics. Despite its importance, finding a fully quantum Rényi generalization that satisfies all desirable structural properties has remained an open question. In this work, we introduce a quantum Rényi conditional mutual information possessing these desirable properties: it is monotone under local quantum channels, additive under tensor products, monotone in the Rényi parameter, and converges to the QCMI in the limit $\alpha \to 1$. We establish bounds on this quantity in terms of relative entropies to the Petz-recovered state, showing that it provides a distinct measure of approximate conditional independence. As a consequence, we derive recovery-based upper and lower bounds on the QCMI in terms of quantum generalizations of the Kullback-Leibler divergence, improving upon the best known results in certain regimes. Finally, we establish bounds in terms of relative entropies to the sets of Markov and separable states and derive Rényi chain rules, which appear to be novel even in the classical setting.

The upcoming High-Luminosity Large Hadron Collider (HL-LHC) at CERN will deliver an unprecedented volume of data for High Energy Physics (HEP). This wealth of information offers significant opportunities for scientific discovery, but its scale challenges traditional analysis workflows. In this paper, we present CMS analysis pipelines being developed to meet HL-LHC demands. These pipelines build on the broader scientific Python ecosystem, complemented by solutions specifically designed for HEP. A central focus of this work is the Integration Challenge, an IRIS-HEP led effort aimed at assessing the readiness of developed software stack to be used in real world physics analysis and improving the readiness of analysis facilities for the HL-LHC era. The Integration Challenge acts as an end-to-end integration test: by implementing a complete physics analysis pipeline, it evaluates tool interoperability and the overall user experience for analysts. The current pipeline includes columnar data processing, machine learning, statistical inference, and visualization tasks covering a variety of CMS analysis scenarios. In addition, the Integration Challenge explores efficient strategies for delivering skimmed data using diverse tools and data formats, as well as evaluating the ServiceX data-delivery system for HEP analyses. Throughout the testing phase, we also investigated several prototype services, such as histogram-as-a-service capabilities, along with other emerging services that may support future HL-LHC analysis workflows.

We introduce Triangular Resampling (TR), a post-training method for mitigating long-horizon error accumulation in motion diffusion models. Built on FloodDiffusion's triangular denoising schedule, TR addresses the mismatch between ground-truth-derived training windows and model-generated inference states. Replacing only completed motion history leaves this mismatch unresolved in partially denoised states within the active window. TR therefore extends rollout-based training to these states, using ground-truth clamping to limit excessive drift. For each replayed sample, TR draws one denoising threshold, shared across latent positions and replay updates, and replays multi-step triangular denoising without gradient tracking. After each update, states below the threshold are replaced with noise-matched ground truth, while those at or above it retain model predictions. The resulting latent window enters the standard training update. This rollout construction supports both supervised training (TR) and distribution matching (TR-DMD). On 120-second motion generation from HumanML3D test prompts, TR and TR-DMD achieve state-of-the-art FID AUC within their respective non-DMD and DMD comparison groups. Supervised TR reduces FID AUC by 40.9% and FID degradation slope by 55.3% relative to matched post-training without replay.

We develop an effective hadronic theory for heavy-meson decays into energetic pions, organized by the approximate heavy-quark and chiral symmetries. The dynamical degrees of freedom of the theory consist of quasi-static heavy-meson fields coupled to soft and collinear pions, relevant for exclusive $B \to \pi$ transitions at large recoil energy ${E_\pi \simeq M_B/2}$. Our approach exploits the factorization of soft and collinear quarks in Soft-Collinear Effective Theory (SCET), leading to an emergent duplication of chiral symmetry in the two kinematic sectors. Promoting the soft and collinear chiral transformations to local symmetries enables us to systematically match non-local four-quark operators in SCET – governing the soft-overlap contribution to heavy-to-light form factors – onto effective operators constructed from hadronic fields below the scale of chiral symmetry breaking. As a first application, we compute the logarithmically enhanced pion-mass dependence of the $B \to \pi$ form factors at large recoil by calculating the one-loop matrix elements of the leading hadronic operators. We find that the non-local transition currents in our effective theory lead to a pattern of chiral-logarithmic corrections that is not compatible with the existing formulation provided by Hard-Pion Chiral Perturbation Theory.

Reward shaping is fundamental to modern robotic control with deep reinforcement learning (RL), yet practitioners still rely heavily on heuristic principles borrowed from classical optimal control and trajectory optimization. Existing methods rarely distinguish reward terms that are intrinsic to the control objective from numerical regularizers, leading to brittle hyperparameter tuning. To determine which quantities a reward must contain, we study the stabilization control problem with a focus on zeroth-order (configuration) and first-order (velocity) information. We theoretically and empirically demonstrate that policy gradient methods can successfully solve stabilization tasks without first-order reward terms, adding such terms can instead introduce severe sensitivity as their scale grows. Conversely, our findings confirm that reward functions must be zeroth-order complete over goal-relevant coordinates, while the first-order state remains necessary in the policy observation under our low-dissipation assumptions. Overall, these results provide actionable and principled guidance for reward design in robotic RL.

Let $\mathcal E$ be a separable Hilbert space, $B\in H^\infty_1(L(\mathcal E))$, and $\alpha>-1$. Associated with the Toeplitz operator $T_B$ on the vector-valued weighted Bergman space $A^2_{\alpha,\mathcal E}$ are the sub-Bergman spaces $\mathcal A_\alpha(B)$ and $\mathcal A_\alpha(B^*)$. In this paper, we prove that \[ \mathcal A_\alpha(B)\supset A^2_{\alpha-1,\mathcal H(B(0))} \quad\text{and}\quad \mathcal A_\alpha(B^*)\supset A^2_{\alpha-1,\mathcal H(B(0)^*)}. \] This confirms a conjecture of \cite{GLM26}. We also construct an infinite-dimensional two-sided inner function $B$ such that $\|B(0)\|<1$ and \[ \mathcal A_\alpha(B)\approx \mathcal A_\alpha(B^*)\approx A^2_{\alpha-1,\mathcal E}, \] but $B$ is not a finite Blaschke-Potapov product. This example shows that the finite-dimensional rigidity theorem of \cite{GLM26} does not extend to infinite-dimensional coefficient spaces.

Jupiter's auroras are produced by magnetospheric electrons precipitating into its atmosphere. Juno/JADE and Juno/JEDI together measure these electrons from 100 eV to 1 MeV. However, their auroral spectra are poorly reproduced by the widely used kappa distribution, hereafter the classical single-kappa distribution. We aim to statistically characterize the energy flux distributions of precipitating auroral electrons and to derive a new phenomenological 4-kappa distribution better suited to JADE+JEDI data. We build mean energy flux distributions in six auroral sub-regions (polar, main, and outer emissions in both hemispheres), combining JADE and JEDI data from PJ3 to PJ35, mapped to the SIII frame with the JRM33 model including the current sheet. We fit them using MCMC with the classical single-kappa and 4-kappa distributions. The mean distributions show a monotonic power-law decay with an energy cutoff beyond ~500 keV in most sub-regions, although some individual spectra depart from it at intermediate energies. Mean energies range from ~21 to ~119 keV, and energy fluxes from ~9.5 to ~28.7 mW/m$^2$. The classical single-kappa fit fails in all sub-regions, with $\kappa$ converging to the lower bound of its prior range. We therefore introduce the 4-kappa distribution, a linear combination of four classical single-kappa distributions modulated by an exponential high-energy cutoff, which accurately fits all six sub-regions. At equal energy flux and mean energy, transport simulations show that the 4-kappa distribution deposits energy at lower altitude than the classical one and distributes it differently along the vertical, with implications for the UV color ratio, atmospheric chemistry and thermal structure, and ionospheric conductances. This distribution provides an analytical framework for realistic modeling of electron precipitation and UV auroral emissions at Jupiter.

The reliable prediction of passenger train delays is a critical component of railway management. While contemporary research frequently attempts to maximize absolute accuracy by deploying opaque deep learning architectures, the underlying data mechanics driving longitudinal predictive decay remain underexplored. Consequently, this study provides an explainable temporal robustness analysis of network-wide railway delay prediction. Focusing on the Dutch railway network, this research utilizes interpretable tree-based ensembles to integrate granular topological, environmental, and operational features. The overarching finding establishes that while feature-rich tree-based models improve simultaneous (within-month) prediction, predictive performance systematically degrades when evaluated across non-simultaneous (future) months. Furthermore, multi-horizon SHAP and dispersion analyses explicitly link this degradation to environmental feature volatility and instability within the statistical target definition. Ultimately, this study demonstrates that richer feature sets alone are insufficient to resolve long-term forecasting constraints, underscoring the necessity to transition toward dynamic, season-aware architectures anchored by absolute operational boundaries.

Automated detection of LLM-generated texts (LGTs) is critical, yet dedicated detectors often struggle to generalize across domains and models. While general-purpose LLMs offer flexible zero-shot authorship classification with explanatory rationale, their detection behavior, especially regarding self-detection versus cross-detection across model generations, remains poorly understood. We systematically evaluate 15 LLMs spanning three model generations as both generators and detectors. Using a benchmark of 1,000 human-written texts and 15,000 LGTs (1,000 per model), we collected over 233,000 binary classifications alongside natural-language explanations. Our results reveal that detection efficacy is primarily driven by detector capability rather than generator provenance, although outputs from newer generators remain notably harder to detect. Crucially, statistical comparisons show no systematic advantage or disadvantage for self-detection across models. Error analysis further exposes generational bias shifts: first-generation detectors under-detect LGTs (high false-negative rates), second-generation detectors over-flag human texts (high false-positive rates), and the latest models achieve balanced trade-offs. Finally, we highlight significant inconsistencies in how different LLMs apply textual cues to justify their decisions. Code: https://github.com/hyyuan/detect-llm-generated-texts.

Molecular crystal packing shapes properties from drug bioavailability to charge transport, yet generating realistic structures requires coordinating molecular flexibility, intermolecular interactions and symmetry. Here we introduce SALA, a flow-matching model built on state–context separation: it evolves only asymmetric-unit (ASU) atomic coordinates and symmetry-compatible lattice parameters while predicting their updates from the full periodic environment. Conditioned on molecular graphs and a space group, SALA generates molecular conformation, lattice geometry and packing end to end while preserving the specified symmetry. Across 799 held-out crystals spanning four chemical classes and all seven crystal systems, SALA achieves a 72.0% hit rate from 50 candidates per target, compared with 8.1% for an atomistic full-cell generative baseline. Mean best-candidate packing root-mean-square deviation decreases from 4.48 to 1.70~Å, with the advantage widening as system size and symmetry multiplicity increase. SALA thus enables end-to-end generation of complex, symmetry-constrained molecular crystals, with potential applications in CSP, structure-based pharmaceutical polymorph design and organic molecular materials discovery.

We investigate two natural formulations of equal-length Littlewood–Paley–Rubio de Francia (LPR) inequalities on arbitrary Vilenkin systems \(\Gm\). The canonical formulation uses canonical intervals in \(\mathbb N\) of equal cardinality, while Vilenkin's formulation uses translates of a common initial interval under the group operation on the dual of \(\Gm\). In this paper, our main results show that the definitions of these two types of intervals exhibit dramatically different properties. More precisely, for the canonical formulation, we construct a counterexample showing that, for every \(2<p<\infty\), the Schatten class \(S^p\) fails the equal-length LPR property, in sharp contrast to the corresponding result on the torus; for Vilenkin's formulation, we prove that, for every \(2\le p<\infty\), a Banach space \(X\) has the equal-length LPR property if and only if it is UMD and has type~\(2\). The sufficiency proof combines a transference argument from the torus to finite cyclic groups with an embedding of Vilenkin intervals into digital rectangular envelopes, yielding bounds independent of the generating sequence.

Reward-maximizing reinforcement learning (RL) is widely used to post-train stochastic diffusion and flow policies for text-to-image (T2I) generation. However, reward-maximizing RL causes policy mode collapse even under reference KL or entropy regularization, reducing the policy to a single high-reward mode. In T2I, this produces similar images and reward hacking. When extended to vision-language-action (VLA) models, the same collapse removes alternative successful strategies and weakens task and scene generalization. To address this limitation, we introduce Unified Trajectory Matching Policy Optimization (Uni-TMPO), a unified RL post-training framework for diffusion and flow policies. First, Uni-TMPO converts standardized rewards into a target distribution within each trajectory group and derives the policy distribution from trajectory log probabilities. Then, forward Kullback-Leibler optimization matches the two distributions instead of maximizing expected reward. A progress-conditioned coarse-to-fine scheduler efficiently constructs T2I trajectories. Within the unified framework, feedback-conditioned sampling uses updated observations to construct VLA trajectories. Extensive experiments show that Uni-TMPO achieves higher T2I rewards and VLA ID success rates than the strongest baselines. More importantly, it achieves the best T2I reward-diversity-efficiency trade-off and VLA generalization to held-out tasks and scenes, while real-robot evaluation demonstrates the value of multiple action strategies when the higher-reward target is blocked.

Spatial reasoning is fundamental to embodied agents, yet it remains unclear whether spatial understanding can be carried forward to guide sequential interactions. Existing spatial-reasoning benchmarks typically terminate at offline predictions, while navigation benchmarks evaluate spatial reasoning as part of instruction following and exploration. We introduce VCN-Bench, a Video-Contextualized Navigation benchmark for probing closed-loop spatial reasoning over prior visual experience in MLLMs. Given a prior video covering both the initial location and destination, the agent is tasked with reasoning out the instruction-specified target and navigating toward it with the inferred spatial context. Built on Matterport3D, VCN-Bench contains five instruction types, 100k training episodes, and 1,250 evaluation episodes. Navigation serves as the primary evaluation, while diagnostic goal identification helps distinguish destination-resolution errors from subsequent navigation failures. We further propose MV-DualVLN, a planning-oriented baseline that jointly leverages prior video and in-episode observations. Experiments reveal limited navigation performance, a substantial destination-resolution-to-navigation gap, and frequent navigation failures even after correct destination identification.