Science & Mathematics

We use the differential 3ω method to directly measure the temperature dependent thermal conductivity for about 90 nm Mo0.78Si0.22 thin films. The measurements are performed in a closed-cycle cryostat to access the temperature range between 50 K and 300 K. The MoSi thin films are capped with Al2O3 and probed with a Pt wire. The results show that the thermal conductivity does not satisfy the Wiedemann-Franz law. This implies that thermal transport in MoSi is not dominated by electrons but by phonon-like excitations, contrary to what is assumed in many electro-thermal models of superconducting devices. We argue that this property might be related to the weak localization visible in the resistivity as a function of temperature curves of the superconductor.

Scenario development, the design of the waveforms that take a tokamak discharge from breakdown to flat-top, is still an iterative, labour-intensive process. Every candidate scenario has to be simulated with a controller in the loop, and the classical shape and current controllers need low-level actuator references. So the process is iterative and includes a number of trial-and-error steps. We propose to replace the per-scenario controller tuning by a single goal-conditioned reinforcement-learning controller trained once, in advance, against a free-boundary equilibrium and transport simulator. A scenario is then specified only by the target waveforms of the physical quantities of interest (plasma current, line-averaged electron density, LCFS shape, x-point positions), and the controller produces the coil-current and fueling commands that realize them. We demonstrate the approach on an EAST plasma-current ramp-up: the controller takes over a small limited plasma, ramps the current and the density, grows the cross-section, forms two x-points and holds a double-null diverted configuration against sampled shape targets for the rest of a 1.2 s episode. The neural network controller is trained using TQC algorithm. On 1024 test scenarios sampled from a low-discrepancy scrambled Sobol sequence in 11-dimensional target space, the controller forms the divertor in 86% of the cases, holds it for more than 0.5 s in 67% and reaches the end of the episode in a diverted state in 58%, with flat-top tracking errors of a few kA in plasma current, $\sim 0.15\times10^{19}$ m$^{-3}$ in density, $\sim 2$ cm in position and $\sim 8$ cm in x-point location; the failures concentrate at the high-current density ($J$) region which is believed to be physically or from engineering perspective not feasible within ohmic mode.

Generating an articulated object from images and text requires more than plausible parts and movable joints: the assembled mechanism must execute the intended motion. This is particularly challenging for closed-chain and coupled mechanisms, where an error in one connection can block motion or separate connected parts. We introduce MARC, Mechanism-Aware Reflective Correction, which shares an executable kinematic graph across program construction, motion solving, and correction. The graph binds physical connections to local attachment frames and represents loop closures and joint-coordinate couplings. Given prescribed driver inputs, a numerical solver computes follower motion and associates errors with specific connections and configurations. The agent uses these diagnostics to probe possible causes, revise connections or geometry, and retest the mechanism along the original motion paths. On a 20-object test set, MARC with Kimi-K3 achieves 67.8% valid motion coverage and 60.0% overall success, outperforming six baselines. Starting from the same initial candidates, correction improves these scores by 13.7 and 20.0 percentage points, respectively. Ablations further show the benefits of graph-based mechanism support and graph-guided repair. Project homepage: https://demianmusk.github.io/MARC-Home/.

This work develops a physics-driven self-supervised graph learning framework that directly optimizes the distortion-aware precoder under nonlinear correlated power amplifier (PA) distortion, without relying on expensive optimizer-generated labels for a user-centric satellite system. We minimize nonlinear PA consumption subject to user quality-of-service (QoS) constraints under PA nonlinearity for a downlink multi-input-single-output (MISO) satellite system. Our pipeline jointly designs the precoder and admission control that minimize nonlinear PA consumed power. In particular, a graph neural network (GNN) for the design of a distortion-aware precoder is optimized end-to-end in a self-supervised manner using a differentiable Itô-Hermite model of the true modified Rapp amplifier model. The precoder output by the GNN is used to initialize a low-complexity distortion-aware signal-to-interference-distortion (SINDR) tracking algorithm that jointly optimizes for power and admission control. Our proposed physics-based pipeline closely tracks a successive convex approximation (SCA) based solution in consumed power performance with the advantage of faster runtime and a wide operating range covered during training.

While language models have demonstrated remarkable capabilities, their performance remains disproportionately skewed towards high-resource languages. This disparity is particularly acute for the 700 million inhabitants of Southeast Asia, where vast linguistic diversity remains under-represented in existing benchmarks, often suffering from limited language coverage or narrow task scopes. To address this, we introduce DeepSEA, a comprehensive benchmark constructed from established datasets through professional human translation into 12 Southeast Asian languages, spanning both high-resource (e.g., Indonesian, Thai, Vietnamese) and low-resource (e.g., Burmese, Khmer, Lao) tiers across five distinct tasks: mathematical reasoning, scientific reasoning, question answering, abstractive summarization, and machine translation. Extensive evaluation of frontier proprietary and open-weights models reveals that while frontier models like Gemini-3.7-Flash maintain robustness across resource tiers, smaller models face severe performance degradation on low-resource languages. Crucially, models perform better in reasoning and summarization when pivoted through English rather than target languages, exposing a critical dependency on English representations, especially for low-resource languages. Beyond these observations, we provide a cross-lingual error analysis and a cross-script tokenizer fertility analysis. We hope these findings shed light on the development of next-generation inclusive multilingual models.

We investigate whether cyclically varying the class composition of training batches can modify both the performance and the internal organization of neural networks. Motivated by nonequilibrium systems, in which cycling between related sub-goals can make switching between their solutions increasingly efficient, we introduce Dynamical Batching (DB), a sampling protocol that periodically emphasizes each class while leaving the loss and the optimizer unchanged, and compare it with Uniform Batching (UB), in which every mini-batch has equal class proportions. In the regime where standard training fails, DB changes the outcome: on a three-arm spiral, a capacity-limited MLP that plateaus far below perfect accuracy under UB fits the training data under DB. Tracking individual weights, we find that under DB the change between class-specialized states is carried by a smaller set of weights than under UB, and that this set is markedly more persistent from cycle to cycle, on every synthetic dataset we test and in a wider network that fits the data under both protocols. This organization is a signature of the protocol rather than a predictor of its benefit: it appears equally where DB leaves accuracy unchanged. On CIFAR-10, the same protocol modestly improves the test accuracy of a channel-narrowed ResNet-10, a standard residual architecture, and of a second capacity-limited convolutional network. These results show that how class emphasis is distributed in time can shape both the solution a network reaches and the internal structure of the network.

GaN-on-diamond heterostructures are promising for high-power and high-frequency electronics, however, direct epi- taxial growth on diamond (100) remains challenging due to severe lattice and thermal mismatches, which induce high threading dislocation densities. To overcome this challenge, we have used Plasma Assisted Molecular Beam Epitaxy (PAMBE) to grow wurtzite GaN layers on diamond(100) substrates using nanowire-assisted epitaxial lateral overgrowth (ELOG). Two nanowire templates (NW100, NW600) were benchmarked against a nanowire-free reference (NW000). High-resolution XRD, φ -scans, (0002) pole figures, and reciprocal space mapping show that NW100 attains the lowest dislocation density (5.51 x 10-4 nm-2) and microstrain (2.736 x 10-3), the narrowest (0002) reflection, and a c-axis lattice parameter (5.1821 Å) within 0.06% of the bulk value; NW000, in contrast, exhibits a contracted c-axis (5.1633 Å) and elevated dislocation density (6.26 x 10-4 nm-2), consistent with residual tensile stress. Cross-sectional TEM con- firms the intended diamond/AlN/nanowire/ELOG/planar-GaN layer sequence and shows threading dislocations bending into near-lateral trajectories, directly evidencing defect filtering by the nanowire template. Together, these results estab- lish nanowire-assisted ELOG as an effective strategy to suppress dislocation propagation and recover near-bulk lattice parameters in GaN-on-diamond heteroepitaxy, providing a structurally well-defined platform for device integration.

Efficient resource management is a challenge in multi-tenant cloud environments as workloads compete for shared resources such as last-level cache and memory bandwidth, whose congestion can change dynamically as workload composition and resource demands evolve. Existing resource management approaches often rely on static resource requirements or prior workload profiling, limiting their ability to respond to changing resource contention as required in the cloud. In this paper, we leverage workload buoyancy, a runtime framework for capturing workload sensitivity to individual resources and application-level performance, to enable adaptive scheduling and resource management. We propose a holistic approach that incorporates buoyancy into workload scheduling, node-local resource allocation, and workload rescheduling throughout the workload lifecycle. We implement our approach as a Kubernetes-native prototype and evaluate it on a multi-node cluster using diverse cloud workloads under varying levels of resource contention. In a high-congestion scenario, our approach increases the percentage of workloads satisfying their SLOs from 7.74% to 87.8% on average compared with the Kubernetes baseline, while reducing resource congestion, the severity of SLO violations, and cluster fragmentation using limited rescheduling.

Autoregressive (AR) video diffusion has emerged as a leading paradigm for efficient video generation, yet its practical deployment remains bottlenecked by the trade-off between generation quality and inference efficiency. We attribute this bottleneck to three error sources in AR rollouts: the base model error, the train-inference distribution mismatch, and the upstream error propagation. Building on this decomposition, we derive closed-form upper bounds on the per-step and cumulative errors, providing a principled basis for jointly optimizing generation quality and inference efficiency. Guided by this analysis, we propose TriTD, a Tri-Partite Trajectory-Distribution co-optimized distillation framework with three complementary modules. First, for ODE initialization, we adopt a Next-Node Trajectory Distillation (NTD) objective, which shortens the supervision horizon and admits a lower regression-error floor. Second, we introduce Chunk-Graded Noising (CGN), a monotonically non-decreasing per-chunk noising schedule that helps narrow the expected train-inference noise mismatch and removes redundant KV-cache updates, yielding substantial speedups at no quality cost. Third, we propose Orthogonal-Parallel DMD (OP-DMD), a distribution-matching loss that retains the original DMD supervision while introducing an additional regression constraint in the orthogonal subspace to suppress off-axis drift, improving both training stability and final generation quality. Extensive experiments show that TriTD consistently surpasses existing baselines in visual quality while reaching a real-time throughput of 26.8 FPS on a single H100 GPU, delivering a balanced quality-efficiency solution for interactive streaming video generation.

Let $S\subseteq \mathbb R^+$ be a subset of positive reals. We say that a metric space is an $S$-metric space if the range of the metric is a subset of $S\cup \{0\}$. In 2007, Delhommé, Laflamme, Pouzet, and Sauer characterized those sets $S$ for which the class of all finite $S$-metric spaces has the amalgamation property. Given an amalgamation class, a natural question is what its Ramsey expansion is. Ramsey expansions are known for classes of $S$-metric spaces for which $S\cup\{0\}$ carries a natural distance monoid structure, for example for $S=\mathbb Q^+$ or when $S$ is topologically closed. In this paper, we extend this result to all amalgamation classes of finite $S$-metric spaces, thereby completing the Ramsey-theoretic classification of amalgamation classes of finite $S$-metric spaces.

Humanoid manipulation requires reusable arm skills, yet demonstrating every possible skill combination is impractical. We present Humanoid Semantic Mixture-of-Experts Diffusion Policy (HS-MoDP), which composes separately demonstrated arm skills through independent semantic routing over a shared expert pool. This design preserves arm-specific skill control while sharing action-generation parameters across arms. A semantic scheduler integrates the learned policy with an existing locomotion controller through alternating manipulation and repositioning stages, extending the effective workspace and enabling skill reuse across location changes. To acquire demonstrations of whole-body contact manipulation, we develop G1Exo, an isomorphic exoskeleton that provides direct joint-level control of arm configurations. Experiments on a Unitree G1 show stronger skill composition than the baselines. G1Exo enables more reliable contact teleoperation than end-effector VR-IK, and its demonstrations improve contact-skill learning across four policy architectures. For HS-MoDP, switching from VR-IK to G1Exo demonstrations raises full cabinet workflow success from 35% to 65%. All project hardware designs and source code will be released as open source after the review process.

Learning from Demonstration (LfD) allows robots to learn new skills from human examples; however, many vision-based methods require large datasets and resource-intensive end-to-end training. This paper introduces a data-efficient LfD framework for humanoid robots that learns manipulation skills from compact, task-specific features instead of raw images. The proposed approach integrates layered visual preprocessing, Kinect-based human joint estimation, prompt-based object detection, and depth information to extract human motion features and three-dimensional object locations from demonstration videos. These features are used to train a temporal deep learning model that predicts robot joint-angle trajectories for execution. The framework is implemented on a NAO humanoid robot, evaluated quantitatively in simulation, and demonstrated on physical hardware. In simulation, the robot achieves a 78% success rate on a pick-and-place task using only 16 demonstrations, which increases to 86% when trained with 100 demonstrations. The framework also achieves a 71% success rate on a pushing task using eight demonstrations, highlighting its effectiveness for data-efficient humanoid skill acquisition.

Large language models (LLM) are increasingly integrated into pervasive applications spanning heterogeneous cloud-edge resources. These heterogeneous resources differ in computational capacity, energy efficiency, network conditions, and deployed model quality, creating trade-offs among response quality, monetary cost, energy consumption, and batch completion time when routing inference requests. Existing LLM routers typically make per-query routing decisions based primarily on model quality and token cost, whereas cloud-edge schedulers often overlook quality differences among deployed models. Consequently, these objectives remain insufficiently coordinated. We present CEMA-Router, an infrastructure-aware batch router for LLM inference in pervasive cloud-edge systems. CEMA-Router formulates batch routing as a multi-objective search problem that jointly minimizes model-quality loss, monetary cost, energy consumption, and straggler-dominated batch completion time. To efficiently explore the search space, it employs a routing-specific memetic algorithm, combining largest-request-first, load-aware seeding with probabilistic, evaluator-guided Lamarckian repair. Routing decisions are informed by deployment-specific node profiles, dynamic backlog estimates, and model-quality signals. We further implement CEMA-Router as an end-to-end system comprising a request manager, routing service, and profiler, supporting both real cloud-edge deployment and trace-driven simulation. We evaluate the system using RouterBench traces and GSM8K and MMLU workloads on a heterogeneous cloud-edge testbed. Compared with state-of-the-art methods, CEMA-Router reduces the weighted objective by up to 86% for GSM8K and 80% for MMLU, while reducing energy consumption by up to 33% and 19% and batch completion time by up to 27% and 30%, respectively.

Identifying conceptually equivalent questions across different surveys is a challenging information retrieval task where surface-level textual similarity is insufficient. Two questions may be lexically similar yet assess entirely different sociological sub-concepts, or lexically different yet conceptually equivalent. We introduce a conditional information retrieval framework that addresses this by decomposing the task into two steps: concept extraction, which identifies the underlying sociological concept of a survey question using a fine-tuned generative model, and concept-aware retrieval, which augments question representations with the extracted concept as an explicit conditioning signal for finer-grained alignment. To reflect the realistic data scarcity conditions of specialised social science annotation, we use CLOSER as a case study. Our framework consistently outperforms baselines by 2-3 points in both MAP and Recall. We further analyse fine-tuning strategies including SFT, DPO, and dynamic DPO, finding that SFT delivers the most stable gains while dynamic DPO interacts differently with instruction-tuned versus distilled reasoning backbones. Finally, we demonstrate that a model fine-tuned on a general social science thesaurus (ELSST) transfers robustly to task-specific survey data, offering a practical solution under data scarcity.

Tactile perception is crucial for precise robotic manipulation, providing an important foundation for dexterous, closed-loop force control. Among various tactile sensing techniques, magnetic tactile sensors stand out thanks to their fine-grained, three-dimensional force perception capability and strong robustness against visual occlusion. In this study, we discover a physical loophole called PhantomTac in magnetic tactile sensors. The attacker can alter the magnetic field, thus spoofing the force interpreted by magnetic tactile sensors. The attacker can stealthily deploy such an attack by simply attaching a thin (the thickness is around 0.5 mm), battery-free magnetic tag to the surface of an object. Hence, the attack tag can be easily concealed by a barcode sticker. Such an attack scheme is "deploy-and-forget" -- the attacker can deploy the tag and leave. When a robotic hand is picking up an object, it may severely damage the object. Specifically, spoofing the normal force can crush the object, whereas spoofing the shear force can lead to a loose grip, thus dropping the object. Compared with directly sabotaging the robotic hand, PhantomTac is much more concealed as the attacker does not need access to the device. We performed extensive reverse-engineering efforts to characterize the sensing mechanism of commercial-off-the-shelf tactile sensors and proposed a practical tag design. End-to-end evaluations conducted on real robots and robotic arms demonstrate that our attack can bypass existing anti-interference mechanisms and effectively hijack their closed-loop control processes, achieving an attack success rate exceeding 95% when the grasping-position offset is bounded within 8 mm. Finally, based on the limitations of existing anti-magnetic interference mechanisms, we propose a multimodal-consistency defense mechanism to mitigate such physical-layer security threats.