Large Language Models are increasingly used in software engineering, but both code generation and its evaluation remain predominantly English-centric. This leaves a major gap in our understanding of how well current tools support multilingual development, where code contains non-English natural language. In this paper, we investigate non-English code comment generation and the reliability of current methods for evaluating such outputs. We evaluate five code LLMs (CodeGemma, CodeLlama, CodeQwen1.5, GraniteCode, and StarCoder2) across five natural languages: Dutch, English, Greek, Polish and Chinese. We further conduct an open-coding study of 12,500 generated comments, from which we derive a publicly released human-annotated dataset and a taxonomy of 26 error types. We use these human annotations, to evaluate the performance of neural metrics, and LLM-as-a-judge pipelines. Our findings show that generative performance deteriorates substantially outside English, with linguistic errors increasing by up to 15.1$\times$, alongside frequent incoherent generations and a rise in semantic errors. More critically, we show that detecting errors in non-English comments underperforms. Across classical overlap-based metrics, off-the-shelf neural metrics, extended neural metrics using newer multilingual, language-specific, and code-specific models, and LLM-as-a-judge pipelines, no automatic approach provides reliable and consistent assessment. Neural metrics fail to distinguish correct comments from incorrect outputs or even random noise, and tend to overestimate quality in non-English settings. LLM-as-a-judge methods achieve the highest agreement with human annotations but fail to reliably capture important language-related and semantic errors. Overall, our results show that evaluation and generation are key barriers for multilingual tooling, and that human judgment remains indispensable.

The primary objective of this study is to remove duplicated monomial contributions that proliferate in Carleman linearization as state dimension and truncation order increase. To do so, we adopt a shift-and-lift architecture, since it exposes repeated exponent targets and allows duplicate-aware coefficient coalescing during lifted-operator assembly. This architecture also makes high-order truncation practical, but that regime intensifies local convergence and closure sensitivity for higher-order nonlinearities. We therefore pair shift-and-lift with a moving-center expansion so that shift and lift are updated jointly around evolving local centers, improving validity of the truncated model along the trajectory. The resulting workflow combines symmetry-reduced monomial bases, packed exponent-key indexing, and sparse triplet coalescing to preserve truncated affine dynamics while reducing index-resolution overhead and write-path irregularity. We analyze variable growth, preprocessing complexity, and truncation-induced error mechanisms, and we compare against Jacobian linearization through fixed-step error, admissible step size, and cost-at-target-accuracy criteria. Two benchmarks (bilinear driver and logistic interaction) show convergence under refinement for both approaches, with regime-dependent accuracy gains for the proposed method rather than universal superiority.

Migration is reshaping demographic landscapes across Europe, raising urgent questions about adapting to rapid population changes. This study examines the canton of Fribourg, Switzerland, which experienced a 30% population increase over the past 15 years, driven by international and internal migration. As local governments face mounting pressures from demographic shifts in housing, education, and social services, understanding the causal effects of migration is essential for evidence-based policymaking. We study how migration reshapes local demographic, educational, and housing outcomes across 112 Fribourg municipalities (2010-2021). Using the intertemporal difference-in-differences estimator of De Chaisemartin and D'Haultfoeuille (2024), which accommodates staggered timing and cumulative, non-binary treatment, we identify the effect of a one-percentage-point increase in cumulative migration balance (relative to baseline population). Migration exposure generates modest but persistent adjustments across demographic, educational, and housing dimensions. Both migration types reduce the share of elderly residents, and international inflows are associated with higher birth counts. Internal migration increases resident students and alters compulsory and secondary-school cohorts, while international migration slightly reduces the tertiary-education share. Housing adjustments are gradual and concentrated in household composition and selected dwelling types, with international migration increasing mid-sized households and internal migration reducing mixed-use dwellings. Though yearly effects are small, their persistence yields meaningful cumulative changes. Overall, migration acts as a counterweight to population aging and generates incremental adjustments in service demand, underscoring the need to incorporate migration exposure into cantonal and municipal planning.

The Mixture-of-Experts (MoE) architecture is crucial for scaling large language models, but its scalability is severely limited by inter-GPU communication bottlenecks in multi-GPU systems. Although overlapping communication with computation is a widely recognized optimization, its effective deployment still remains challenging, both in terms of performance and programmability. In this work, we identify the root cause as a fundamental abstraction mismatch between MoE's dynamic, irregular token-to-expert mapping and the static, address-centric communication model of modern GPUs, which necessitates a complex software mediation phase to resolve addresses before data transfers, limiting performance and software flexibility. To resolve this, we propose MoE-Hub, a hardware-software co-design that introduces a destination-agnostic communication paradigm. MoE-Hub decouples data transmission from address management, allowing producers to send data immediately after routing using only a logical destination, while address allocation and data-flow orchestration are handled transparently by lightweight hardware in the GPU hub. By hardware-accelerating the entire communication control plane, MoE-Hub enables seamless and transparent overlap. Our evaluation shows that MoE-Hub achieves 1.40x-3.08x per-layer and 1.21x-1.98x end-to-end speedup over state-of-the-art systems.

Low-latency anonymity networks such as Tor remain vulnerable to infrastructure-level traffic analysis that exploits side-channel information observable from encrypted communications. We introduce NATA, a non-invasive active traffic-correlation analysis algorithm that injects distinguishable throughput patterns into traffic flows through controlled bandwidth perturbations. Unlike passive correlation methods, NATA does not require endpoint compromise, Tor-browser modification, or packet-payload decryption or modification. It can be carried out by an adversary that controls an upstream network gateway and observes traffic at adversary-controlled exit relays. To identify perturbed flows under substantial network variability, we develop BM-Net (Bandwidth Modulation Network), a selective state-space learning framework adapted for bandwidth-modulation detection. Given the limited availability of high-fidelity ground truth on real-world cross-continental Tor paths, BM-Net adopts a data-efficient learning strategy that separates self-supervised representation learning from supervised task-specific classification. It first learns reusable traffic representations through masked pre-training on serialized traffic traces, and then adapts these representations to binary perturbation detection and fine-grained modulation classification using task-specific labeled data. Through real Tor traffic measurements, BM-Net achieves a 99.65% binary detection F1 score and a 97.5% macro-F1 score for fine-grained modulation classification under our evaluated settings. In addition, tornettools-based scaled simulations are used to estimate exit-observation probability under bandwidth-weighted relay selection. These results suggest that active bandwidth perturbation can serve as an infrastructure-level side channel for traffic correlation under a clearly defined adversary model.

The cubic Fe2-xMnxO3 is an intriguing material that has recently been investigated for various applications, including lithium-ion battery anodes, catalysts, energy storage media, humidity sensors, and photocatalysts. Despite its wide range of promising applications, the magnetic properties of Fe2-xMnxO3 remain controversial, with different sources reporting conflicting information regarding the type of magnetic ordering, phase transition temperature, and magnetic moment of this compound. This work presents a study of the magnetic state of three Fe2-xMnxO3:Ga solid solutions with varying Mn:Fe:Ga ratios, along with one gallium-free Fe2-xMnxO3 reference sample. We performed a detailed analysis of the actual chemical composition and crystal structure of the synthesized samples using energy-dispersive X-ray spectroscopy (EDX), powder X-ray diffraction (XRD), and X-ray absorption spectroscopy (XAS) to evaluate compositional differences. The magnetic states of the three Fe2-xMnxO3:Ga samples and the gallium-free Fe2-xMnxO3 were investigated using magnetometry and Mossbauer spectroscopy. The low-temperature magnetic anomalies were found to be more consistent with spin-glass-like freezing than with conventional long-range antiferromagnetic ordering. Although variations in magnetic behavior were observed and found to depend on composition and the cooling rate during synthesis, our results demonstrate that these factors do not account for the drastically different magnetic properties reported for similar bixbyite-type oxides. Instead, the apparent room-temperature ferrimagnetism observed in one sample is most likely extrinsic and can be attributed to a trace spinel-type impurity phase, as supported by magnetizations and ESR measurements. Thus, the origin of these discrepancies lies primarily in the chemical purity of the samples and, to a significant extent, in the synthesis technique employed.

This paper develops a multi-port S-parameter framework for the analysis and optimization of stacked intelligent metasurfaces (SIMs) with unilateral active interconnections. By modeling each unit cell as a non-reciprocal two-port network, the resulting SIM exhibits a feed-forward structure that enables a recursive, cascade-like representation of the end-to-end transfer function while preserving electromagnetic accuracy. Based on this model, we derive an efficient gradient-based optimization algorithm with reduced computational complexity compared to conventional reciprocal SIM architectures. Numerical results, obtained from full-wave simulations, illustrate the trade-offs among inter-layer spacing, active gain, and SIM size in terms of channel diagonalization and achievable spectral efficiency.

Simulating coronal mass ejections (CMEs) from their origin in active regions (ARs) to their propagation to Earth remains challenging, particularly when aiming to resolve AR scales and employ realistic magnetic field strengths without compromising computational efficiency. Here we present a methodology for end-to-end CME modeling that addresses these challenges. Three nested magnetohydrodynamic simulations are coupled to jointly cover the heliosphere from solar surface to beyond $1.5$ au. A block-structured adaptive mesh refinement scheme is employed to achieve $\sim 700$ km resolution in the low corona, allowing AR scales to be resolved while maintaining the total grid count below $10^8$ across the entire computational domain. A semi-relativistic Boris correction combined with a relativistic mass-density factor is used to handle magnetic field strengths up to $10^3$ G without prohibitively small time steps. Using this model, we simulate the emergence of a bipolar AR into the corona, the initiation of a CME by shearing of the AR core field and the subsequent evolution. Our simulation captures its pre-eruption energy buildup, triggering by magnetic reconnection, rapid acceleration, and propagation to 1 au and beyond. The simulated CME exhibits a three-part structure in synthetic coronagraph images and a torus-shaped flux rope in the heliosphere, with synthetic in-situ observations showing shock formation, density compression, and a prolonged southward $B_z$ component at 1 au. The entire simulation requires about one day on a moderately sized cluster (e.g., $600$ processors), while the simulated CME takes three days to arrive at $1$ au, offering a lead time of two days if used for forecasting.

We present a fault-tolerant mapping of rotated surface codes onto a $2\times N$ silicon spin-qubit railway architecture, utilizing electron shuttling to resolve the wiring fan-out bottleneck. Employing circuit-level noise modeling, we evaluate threshold performances across various noise biases. We demonstrate that shuttling check qubits instead of data qubits fundamentally improves system thresholds. Crucially, under a noise model biased towards dephasing for spin-qubit shuttling, the non-CSS XZZX surface code outperforms standard CSS variants. By tailoring the topological code to this specific inherent bias, we show that the Megaquop footprint is achievable with a distance 7 code requiring a $p = 10^{-3}$ physical error rate, highlighting a pathway for substantial hardware reductions in early fault-tolerant quantum processors.

Key aspects of the quantum oscillations and magnetoresistance in Weyl semimetals $AM$Te$_4$ ($A$=Nb,Ta, $M$=Rh, Ir) persist as open questions, obscuring the link between their topological electronic structure and practical implementations. Employing a generalised search procedure, we carry out a comprehensive scan of WPs accounting for all the subbands close to the Fermi energy, and show that this dramatically alters the WP landscape in these compounds. In particular, we predict these compounds to feature WPs within a few meV of the Fermi energy which significantly influence their properties. Remarkably, most of the considered compounds host WPs of more than one type, including NbRhTe$_4$ which hosts type-I, II and III Weyl points. Our comparative analysis of structure and fidelity of computational parameters/models not only provides a detailed mapping of the complex electronic structure in these compounds, but also clarifies quantum oscillations and magnetoresistance observations in this family, bridging the gap between theory and experiments and offering a framework for precise tunability of WPs.

We develop a discrete optimal transport framework for analyzing simulated annealing algorithms on finite state spaces. Building on the discrete Wasserstein metric introduced by Maas (J. Funct. Anal., 2011), we define a generalized discrete Wasserstein-2 distance and the associated notion of \emph{discrete action} for paths of probability measures on graphs. Using these tools, we establish non-asymptotic convergence guarantees for simulated annealing: the KL divergence between the algorithm's output and the target distribution is controlled by the discrete action of the annealing path. This can be viewed as the discrete counterpart of the action-based analysis of annealed Langevin dynamics in continuous spaces by Guo, Tao, and Chen (ICLR 2025). As applications, we analyze simulated annealing for two fundamental models in statistical physics. For the \emph{mean-field Ising model}, we show that annealed single-site Glauber dynamics achieves $\varepsilon$ error in KL divergence in $O(n^5β^2/\varepsilon)$ steps at \emph{any} inverse temperature $β\ge 0$. For the \emph{mean-field $q$-state Potts model}, we show that annealed $(q-1)$-block Glauber dynamics achieves $\varepsilon$ error in $\mathrm{poly}(n, β, 1/\varepsilon)$ steps for all $β\ge β_{\mathsf{s}}=q/2$, the regime where the disordered phase has completely lost stability. In both cases, the key technical contribution is a polynomial upper bound on the discrete action, obtained by exploiting the symmetry of the model to reduce the analysis to a low-dimensional projected chain.

Thermal protection remains a critical challenge for oblique detonation engines (ODEs) operating under hypersonic conditions due to the extreme heat release and compact combustor geometry associated with oblique detonation waves (ODWs). In the present study, the effectiveness of film cooling for a kerosene-air ODE combustor is numerically investigated under a flight Mach number of 10 and an altitude of 15 km. Three active cooling strategies are considered, including air film cooling, gaseous-kerosene film cooling, and liquid-kerosene mist cooling. The results show that all cooling strategies preserve stable oblique-detonation propagation and maintain the canonical wave-system structure within the investigated operating range. Air cooling produces stronger disturbances near the initiation region and triple point, resulting in enhanced downstream wave interactions and larger propulsion penalties. In contrast, fuel-based cooling induces milder disturbances and better preserves the global detonation structure. All cooling methods substantially reduce the near-wall thermal load, although their cooling characteristics differ significantly. Gaseous-kerosene film cooling exhibits a spatially periodic near-wall thermal response associated with the discrete cooling hole arrangement, while liquid-kerosene mist cooling produces a smoother near-wall temperature distribution due to enhanced two-phase mixing and phase-change heat absorption. Among the investigated strategies, mist cooling provides the best overall balance between thermal protection and propulsion performance at coolant mass ratios of 1%-3%, whereas gaseous-kerosene film cooling becomes advantageous at higher injection levels due to improved wall coverage continuity. The present results demonstrate the feasibility and potential of fuel-based film cooling for thermal management in hypersonic ODE combustors.

Active control of dark long-lived excitonic states in molecular aggregates using local electric fields is a pivotal challenge for advancing nanoscale optoelectronics and quantum device engineering. This experimental study investigates the collective excitonic states in aggregates composed of radical chromophores. With the strong optical enhancement provided by tip-enhanced photoluminescence (TEPL) spectroscopy, bright and dark excitonic modes are observed emerging due to interexciton coupling and induce changes in their spectra with the electric field locally applied within the nanocavity gap. Proportionally scaling Stark shifts are revealed as well as the emission peak sharpening of the dark states and a divergent behavior of the bright states in asymmetric measurement positions of the nanocavity above the aggregates. The observed complex behavior is discussed in terms of influence of the field, molecule arrangement, nanocavity coupling, dark mode lifetimes and electrostatic charge inhomogeneities in the clusters. This sensitivity to the external parameters demonstrates an effective means of control over radical excitonic aggregates.

We consider a shallow water model in a homogenization framework. For periodic topographies, Craig, Lannes and Sulem have established a consistency result under some non-resonance conditions. In the present contribution, we significantly relax the periodicity condition and treat general topographies under minimal assumptions.

Agent Skills have become a practical way to extend LLM agents by packaging metadata, natural-language instructions, and executable resources into reusable capability bundles. However, this growing Skill ecosystem introduces a new compliance risk: a Skill may perform high-impact actions that exceed the minimum necessary scope of the user's current task, thereby violating least-privilege. Existing skill detection approaches are insufficient for this problem because it is inherently task-conditioned: the same action may be necessary under one user prompt but over-privileged under another. In this paper, we present SkillScope, a framework for fine-grained least-privilege enforcement in Agent Skills. SkillScope adopts a graph-based analysis approach that models instruction-level procedures and code-level operations as fine-grained action nodes. It extracts potential over-privilege candidates, validates them under graph-instantiated user tasks through replay-based analysis, and constrains validated over-privileged actions via control-flow privilege constraining. We evaluate SkillScope through effectiveness experiments and large-scale real-world measurement. SkillScope achieves 94.53% F1 for skill over-privilege detection. In the wild, SkillScope validates 7,039 Skills with over-privileged behaviors, showing that least-privilege violations are prevalent in current Skill ecosystems. In the privilege-constraining evaluation, SkillScope reduces triggered over-privileged action-in-task instances by 88.56% while preserving legitimate task completion.

The security of AI-generated code remains a major obstacle to its widespread adoption. Although code generation models achieve strong performance on functional benchmarks, their outputs frequently contain bugs and security weaknesses that undermine their trustworthiness. Prior work has explored a range of approaches to mitigate security issues in AI-generated code, e.g., using static analysis-guided generation and prompt engineering. However, their effectiveness varies widely across models and settings. This paper presents a systematic investigation of strategies for hardening model-generated code against a list of Common Weakness Enumeration (CWE). We assess the extent to which these strategies improve security across models and programming languages, using fine-tuning and prompting approaches for model output refinement. Beyond the prevalence of security weaknesses, we analyse the severity of identified CWEs, their co-occurrence, and the unintended consequences of remediation (i.e., whether fixing certain weaknesses introduces new weaknesses elsewhere in the same code). Our results show that security improvements are highly strategy- and model-dependent. Although some approaches reduce specific classes of weaknesses, they often introduce new weaknesses as side effects of the fixes. Moreover, no strategy consistently eliminates weaknesses across all models and scenarios, highlighting the absence of a universally effective "bulletproof" solution for secure AI-generated code.

We classify the several classes of the set of smooth measures from the perspective of the denseness and the locality, and consider their relationships, in particular, that of the Kato class and Radon measures of finite energy integrals. We also introduce the Miyadera metric on the Dynkin class, and obtain the continuity of the Revuz correspondence.

A Russell graph measure (RGM) is one of the standard DEA models, but its efficiency measure is not well-defined--or has unacceptable properties--at the boundary of the nonnegative orthant. This is known as a boundary problem. Existing studies have tackled this issue; however, their models may fail to identify an efficient target or fail to satisfy some desirable properties of efficiency measures. In this paper, we incorporate a closer target setting approach into the RGM model with production trade-offs to overcome such issues. We demonstrate that the efficiency measure of the proposed model overcomes the boundary problem and has stronger properties than existing models. We also demonstrate that the efficiency scores of the proposed model can be computed by solving a series of LPs. We conduct a numerical experiment with a real-world dataset to illustrate how targets provided by our model are realistic compared with the existing model, which also suggests the validity of our model in applications.

In placebo-controlled randomized trials, the post-randomization use of concomitant medications may be higher in the placebo arm than in the treatment arm. This may dilute the full benefits of the randomized drug as estimated by the intention-to-treat analysis. We focus on cardiovascular outcomes trials in type-2 diabetes patients of glucose-lowering treatments where patients in the placebo arm are more likely to add other glucose-lowering agents with established cardio-protective properties. As a supplement to the intention-to-treat analysis, we propose a class of estimands within a causal framework that isolates the specific impact of the treatment being studied from that of concomitant treatment use. These estimands are defined under time-dependent treatment interventions to balance exposure to additional medications across intervention arms. We advocate for specific stochastic interventions to achieve this balance while minimizing positivity violations, which arise when certain treatment combinations or characteristics are not sufficiently represented in the data. We employ targeted minimum loss-based estimation (TMLE) to optimize the estimation procedure for our estimands while allowing for flexible adjustments for time-dependent covariates from follow-up visits. Finally, we demonstrate the application of the methods through a simulation study and a real-world example from the LEADER cardiovascular outcomes trial, which assessed cardiovascular risk for liraglutide versus placebo.

In micromagnetic simulations, the constant magnitude of the magnetization can be derived from the continuity equation. Since the time evolution of the magnetization in the continuity equation is perpendicular to the plane determined by the magnetization and the effective field, taking the inner product of both sides of the model with the magnetization shows that the evolution rate of the magnitude of the magnetization is zero, thus keeping the magnitude constant. From this perspective, the equation itself can maintain the constraint of constant magnetization magnitude. We discretized the continuity equation and compared two first-order semi-implicit strategies in time: one is the implicit Gauss-Seidel method, and the other is the semi-implicit Backward Differentiation Formula (BDF) method. We considered the comparison between these two schemes with and without the projection step. The results of micromagnetic simulations show that when the dissipation coefficient is large, the implicit Gauss-Seidel method without the projection step has significant differences from the method with the projection step in both the achieved steady state and domain wall motion. When an appropriate dissipation coefficient is selected, the difference between the two narrows, and both the steady state and domain wall motion can be simulated. For the other method, BDF1, whether the dissipation coefficient is large or small, the results with and without the projection step are quite consistent, and it can effectively simulate the domain wall motion.

This paper investigates the application of soft magnetic composites (SMCs) in the stators of wound field synchronous machines for automotive traction. While SMCs are traditionally employed in axial flux topologies, this study examines their use in radial-flux electrically excited synchronous machines (EESMs). Multiple SMC materials and lamination thicknesses are evaluated, with the optimal configuration combining a SMC material in the stator and 0.35 mm NO35 laminated steel in the rotor. This combination delivers improved torque and efficiency compared to conventional designs. When integrated into a full electric drive unit (EDU), this motor achieves 89.7% efficiency over the WLTP drive cycle, representing a 1.4 percentage point improvement over a reference permanent magnet synchronous machine-based EDU. The proposed solution eliminates rare-earth materials, reduces cost through thicker laminations, and offers environmental benefits through SMC utilization. This novel material combination, previously unexplored for radial EESMs, presents a promising direction for affordable, high-efficiency, rare-earth-free automotive traction machines.

Non-terrestrial networks (NTN) have been standardized by the 3rd generation partnership project (3GPP) as a key component of future 6G systems to enhance coverage and resilience. In particular, NTN technologies such as low-earth orbit (LEO) satellites, high-altitude platform stations (HAPS), and unmanned aerial vehicles (UAVs) are expected to support terrestrial networks (TN) during extreme events and disasters. In this paper, we present a lightweight system-level simulator for evaluating post-failure fallback behavior in integrated TN-NTN wireless networks under a partial-failure disaster model. The simulator follows 3GPP Rel-17/18 modeling principles, supports probabilistic terrestrial next-generation node B (gNB) failures, and service migration to NTN. The simulator supports comparative analysis of throughput, packet reception ratio (PRR), and latency under different user loads, disaster severities, and NTN provisioning levels. Results show the expected capacity-delay tradeoff of terrestrial operation, the reliability and stability of non-terrestrial service, and the balanced resilience behavior of hybrid TN-NTN operation. The proposed framework provides a tractable tool for studying wireless network resilience and traffic management in future integrated 6G mobile systems.

Let $F$ be a field, and $\mathcal{M}$ be a linear subspace of $n$-by-$n$ matrices with entries in $F$ that have at most two eigenvalues in $F$ (respectively, at most one non-zero eigenvalue in $F$). In a previous article, we have determined the greatest possible dimension for $\mathcal{M}$ when the characteristic of $F$ is not $2$. In this article and its sequel, we solve this problem for all fields with characteristic $2$.

In this paper, we design a bifocusing-based imaging strategy for the rapid identification of small penetrable dielectric inhomogeneities within a two-dimensional bistatic measurement setup. To address the applicability and limitation, we carefully explore the mathematical structure of the indicator function by establishing a relationship involving the infinite series of Bessel functions, the material characteristics, and the bistatic angle. Through this theoretical result, we rigorously verify that the imaging resolution degrades as the bistatic angle approaches $\SI{180}{\degree}$, and specifically, that target identification becomes impossible when the bistatic angle is $\SI{180}{\degree}$. Conversely, relatively high-resolution results are obtained when the bistatic angle is close to $\SI{0}{\degree}$. The theoretical findings are validated through numerical simulations using the Fresnel experimental dataset, which confirm the applicability and limitations of the proposed method for both dielectric and metallic objects.

Radio map (RM) reconstruction is essential for environment-aware wireless networks, but practical measurements are often collected along mobility trajectories rather than randomly scattered over the target region. Such trajectory-sampled observations induce spatially heterogeneous uncertainty: near-trajectory regions are directly constrained, whereas distant or occluded regions remain weakly observed, leading to degraded reconstruction accuracy in under-constrained areas. To address this problem, we propose Trajectory-Guided Plug-and-Play Priors (TGPP), a general guidance module for sparse RM reconstruction. TGPP learns an explicit guidance map as an interpretable input-space risk prior, and an implicit guide feature that is projected and fused with backbone hidden representations. TGPP can be attached to different reconstruction backbones without changing their original task formulation. We further introduce RadioFlow-LDM, a latent flow-based generative backbone, and apply TGPP to deterministic, adversarial, graph-based, and latent generative reconstruction models. Experiments on RadioMapSeer with five trajectory sampling rates show that trajectory-sampled reconstruction differs substantially from random sparse interpolation. TGPP improves most reconstruction metrics across backbones, achieving up to 43.1% NMSE reduction relative to the corresponding base backbone without trajectory-guided priors.

The quest for ubiquitous mobile coverage has catalyzed two fundamentally distinct architectural paradigms: Direct-to-Cell (D2C) and standardized 3GPP Non-Terrestrial Networks (NTN). D2C, pioneered by SpaceX Starlink and AST SpaceMobile, leverages existing terrestrial spectrum and unmodified consumer handsets to provide emergency connectivity as a market-driven overlay. In contrast, 3GPP NTN, standardized across Releases 17-19, offers a systematic satellite-native framework designed for long-term scalability, high-throughput broadband, and deep integration with terrestrial 5G/6G networks. This paper presents a comprehensive technical comparison of these approaches, analyzing their standardization trajectories, network architectures, physical-layer innovations, security postures, and operational trade-offs. We further examine their implications for emerging 6G use cases, particularly autonomous driving, where safety-critical redundancy motivates a hybrid tri-link architecture combining terrestrial 5G, NTN broadband, and D2C emergency fallback. Our analysis shows that, although D2C enables rapid market entry through legacy-device compatibility, NTN provides superior performance, security, and scalability, positioning it as the foundational framework for 6G satellite-terrestrial convergence. A hybrid model that combines the strengths of both paradigms is identified as the most practical path toward truly global connectivity.

Gauge theories form the foundation of the Standard Model of particle physics. These theories can exhibit confinement, where charged particles only occur in bound states, connected by flux strings whose energy grows linearly with separation. Simulating the real-time dynamics of such strings, including their breaking, remains a major challenge for classical computations and a promising target for quantum simulations. While recent quantum simulation experiments explored string-breaking dynamics in abelian lattice gauge theories, non-abelian theories are qualitatively distinct because gauge fields themselves carry charge. Here, we report the first quantum simulation of genuine non-abelian string-breaking dynamics in a pure SU($2$) lattice gauge theory, where gauge-field self-interactions drive string breaking even in the absence of dynamical matter. Our results are obtained on a trapped-ion quantum computer, using native qudit Hilbert spaces to encode truncated gauge fields on a ladder geometry and implement digital Trotter dynamics. We experimentally study unbreakable and breakable strings generated by fundamental and adjoint static charges, respectively. We locally resolve string oscillations and coherent string breaking through the creation of gluonic excitations driven by non-abelian plaquette interactions. Our work establishes hardware-efficient, problem-tailored qudit simulations as a promising route for accessing non-perturbative dynamics relevant to high-energy physics.

The paper concerns the problem for the ultrahyperbolic equation in the Euclidean space with data on a characteristic hyperplane. Smoothness and asymptotics of the solution along characteristic lines transversal to the initial hyperplane are investigated.

We study the problem of approximating a discrete probability distribution, such as the next-token distribution of a large language model, by a dyadic distribution induced by a binary tree under encoding rate constraints. The objective is to partition the support of the distribution and assign dyadic probabilities to minimize total variation distance while achieving a prescribed rate. We formulate this task as a tree-based partitioning problem and develop a polynomial-time additive approximation scheme for the rate-constrained setting in the constant-rate regime. Our results provide provable guarantees for near-optimal dyadic approximations and, as an application, yield a principled framework for LLM-based steganography, where the rate maps to bits of hidden information embedded per token and the total variation bound controls statistical detectability.

Pair programming is a widely used collaborative learning practice in computer science education yet its effectiveness varies substantially due to breakdowns in coordination attention and cognitive regulation between partners. This paper investigates whether AI supported feedback grounded in joint visual attention and joint mental effort can improve collaborative programming performance and how feedback timing shapes learner AI interaction. Two experimental studies using dual eye tracking capture real time indicators of collaborative regulation during debugging tasks. Study 1 examines reactive feedback that intervenes when observed joint visual attention or joint mental effort deviates beyond predefined thresholds while Study 2 evaluates proactive feedback that forecasts future regulatory breakdowns using machine learning models and intervenes pre emptively. Across both studies feedback effectiveness is assessed through debugging success time on task and feedback uptake reflected in code changes. Multimodal feedback significantly improves collaborative performance compared to no feedback conditions. Reactive feedback yields strong gains in debugging success and efficiency particularly when joint visual attention and joint mental effort based feedback are combined. Proactive forecast based feedback further enhances performance reduces time on task and increases constructive feedback uptake while relying less on intrusive interventions. Proactive feedback better preserves learner agency by maintaining optimal collaboration states, particularly for high-performing pairs. These findings demonstrate that gaze and mental effort synchrony can serve as reliable actionable triggers for AI supported collaborative learning highlighting the importance of feedback timing transparency and anticipatory regulation in supporting effective pair programming.