Searcharxiv⌕ Search

arXiv subjects

Hanwen Wang

Publications and source records attributed to Hanwen Wang.

At least 19 recordsLinked to original sources

MetaAct: Self-Evolving Executable Action Constraints for Embodied Language Agents

Self-evolving language agents have emerged as a promising paradigm for continuously improving agent behavior through experience. However, most existing approaches evolve textual artifacts, which can influence action preferences but cannot reliably enforce action admissibility in embodied environments with discrete and often irreversible actions. We propose MetaAct, which shifts self-evolution from textual guidance to executable action constraints by synthesizing GEAR (Grounded Executable Action Rules) from execution feedback. Each rule is represented as a typed specification, compiled into a deterministic guard, and validated offline through trajectory replay without additional model calls or environment interactions. Across four benchmarks, MetaAct improves task performance; on ScienceWorld and WebShop, violations drop from 70% to 0% and 40% to 1%, with score gains of +10.1 and +8.0. Our code is available at https://github.com/BaiCubed/MetaAct.

cs.CL↗

Magnetism modulation in Co3Sn2S2 by current-assisted domain wall motion

The efficiency of spintronic devices can be improved by generating higher effective magnetic fields with lower working currents. Spin-transfer torques can drive magnetic domain wall motion in a device composed of a single material, but a high threshold current density is typically required to move the domain wall and improving the effective magnetic field in common itinerant ferromagnets is difficult. Here we report magnetism modulation in Co$_3$Sn$_2$S$_2$, a magnetic Weyl semimetal, via spin-transfer-torque-driven domain wall motion. We examine the effect of d.c. current on magnetic reversal using anomalous Hall resistance measurements and domain wall motion using time-of-flight measurements. At 160 K, the threshold current density for driving domain wall motion is less than $5.1\times10^5$ A cm$^{-2}$ at zero external field and less than $1.5\times10^5$ A cm$^{-2}$ at a moderate external field (0.2 kOe). The spin-transfer-torque effective field can reach as high as 2.4 to 5.6 kOe MA$^{-1}$ cm$^2$ at 150 K.

cond-mat.mtrl-sci↗

Dual Exam: Controlled Evaluation of Multimodal Agents and World Models

World Models (WMs) can support Multimodal Agents (MAs) through simulated experience, while Agent trajectories can inform World Model refinement. Evaluating this interaction requires separate evidence about action selection and state evolution. We introduce Dual Exam, which tests Agents under fixed environment response mechanisms and World Models under fixed action interventions across physical, digital, social, and scientific domains. The Agent track compares 10 configurations on 1,088 items in 15 reporting categories. The World Model track contains 400 cases across eight scenarios; 15 configurations have full coverage and one covers six scenarios. GPT 6 Astra leads the case-weighted Agent Index at 31.3, and GPT-Image-2 leads the full-coverage WM Overall at 21.7. These scores use different tasks and metrics and are not directly comparable. A 150-case pilot of one coupled pairing records zero task successes and motivates component-level diagnosis. Results reveal uneven task performance and failures to preserve task-relevant state. Their value for predicting coupled success or sustained reciprocal learning remains to be established.

cs.AI↗

In-context Robot Learning Made Simple: A Democratized Recipe for Manipulation Tasks

We study robotic in-context learning (ICL), an emerging paradigm that enables robots to infer and execute tasks from visual demonstrations. Despite its growing promise, the problem itself remains under-defined: a visual demonstration simultaneously conveys action trajectories, object semantics, manipulation affordances, spatial relations, and task goals, making it unclear what information the robot is actually expected to follow. In this work, we first provide a clear problem definition of robot ICL that explicitly defines its learning target and resolves this fundamental prompt ambiguity. Building on this definition, we develop a minimalist and reproducible ICL framework (SimpleICL) with a visual prompt encoder and a low-cost data collection protocol. Without massive pre-training or specialized data infrastructure, our framework achieves strong performance in both simulation and real-world environments. Extensive experiments further reveal several key properties of robot ICL, including semantic discrimination and task-relevant disentanglement. We will fully open-source our data and training pipeline to facilitate systematic and reproducible research on robot ICL. The project page can be found at https://simpleicl.github.io/simpleicl.

cs.RO↗

Generative World Simulation with DAO and JING

We present a Generative World Simulation framework that couples persistent shared world state with continuous first-person audiovisual generation. The framework integrates DAO, a computable world engine, and JING, an Egocentric Interactive Experience Model, to support sustained interaction for humans and agents. DAO maintains the states of characters, objects, and spaces and their relationships, advancing the world according to interaction rules and world time while preserving the consequences of past actions. From this evolving state, DAO derives observer-specific conditions for JING, which combines these conditions with character, object, and scene references, observation history, and current actions to generate each observer's audiovisual experience. Users can explore environments, manipulate objects, and communicate with characters through viewpoint control and text instructions, with their actions shaping both the shared state and subsequent experiences. By connecting world-state evolution with experience generation in a continuous interaction loop, the framework offers an approach to building interactive worlds that retain their history, evolve through action, and persist beyond any individual observer.

cs.CV↗

Pomeranchuk-like electronic localization above 100 K in twisted MoS$_{2}$

Twisted transition-metal dichalcogenides (TMDs) have manifested a rich variety of emerging physical phenomena, yet experimental studies have so far been largely limited in their valence bands (p-doped). Here, we show correlated electronic states in the conduction bands (n-doped) of near-AA-twisted bilayer MoS$_2$ with twist angles ranging from $\sim2.3^\circ$ to $\sim3.5^\circ$, down to the mK temperature regime. A strongly reconstructed correlated phase diagram as a function of twist-angle has been observed - correlated gaps persist to temperatures approaching $160$ K at small twist angles, but collapse to only $\sim20$ K at intermediate angles, where a richer landscape of interaction-driven states emerges. At the largest twist-angle $\sim 3.5\,^{\circ}$, correlated resistance at 1 electron per moiré unit cell is enhanced upon heating, consistent with thermally assisted localization, or, a Pomeranchuk-like behaviour. Its magnetic-field response, however, is highly anisotropic, which differs markedly from that of canonical isospin moiré Pomeranchuk effect in graphene systems. Strikingly, such signature can persist even above 100 K around a filling of 2 electrons per moiré, despite of its weak resistive nature. Our results establish the twisted MoS$_2$ as a platform for studying the complexity of charge localization, internal flavour degrees of freedom, and band topology in conduction bands of semiconducting moiré systems.

cond-mat.mes-hall↗

DexJoCo: A Benchmark and Toolkit for Task-Oriented Dexterous Manipulation on MuJoCo

Achieving human-level manipulation requires dexterous robotic hands capable of complex object interactions. Advancing such capabilities further demands standardized benchmarks for systematic evaluation. However, prior work does not focus on task-oriented dexterous manipulation. Existing benchmarks also often lack high quality human demonstrations and an easy-to-use evaluation pipeline. In this paper, we present DexJoCo, a benchmark and toolkit for task-oriented dexterous manipulation, comprising 11 functionally grounded tasks that evaluate tool-use, bimanual coordination, long-horizon execution, and reasoning. We develop a low-cost data collection system and collect 1.1K trajectories across these tasks, with support for domain randomization to assess robustness. We benchmark modern models under diverse settings, including visual and dynamics randomization, multi-task training, and action-head adaptation. Through extensive empirical analysis, we identify several important insights and common limitations of current policies in dexterous manipulation, highlighting key challenges for future research in dexterous hand robot learning. Code is available at: https://dexjoco.github.io

cs.RO↗

FRIST: FMRI Representation Informed Shared-space Training Improves EEG-only Individual-Finger BCI Decoding

Finger-level motor decoding is important for naturalistic brain-computer interface (BCI) control, yet individual-finger decoding from scalp electroencephalography (EEG) remains challenging because finger representations are spatially close in the sensorimotor cortex and blurred by volume conduction. Leveraging the high spatial resolution of functional MRI (fMRI), we introduce fMRI Representation-Informed Shared-Space Training (FRIST), a two-stage EEG decoding framework that first learns fMRI-informed spectral projections from simultaneous EEG-fMRI recordings and then uses fMRI-derived class geometry to guide residual refinement of EEG predictions. FRIST transfers information across recordings through shared finger labels without requiring paired trials and uses only EEG at inference. We evaluated 12 able-bodied participants during movement execution (ME) and motor imagery (MI) under two-class and three-class chronological session-held-out decoding simulating the online scenario. Using EEGNet as the EEG feature extractor, FRIST increased group average accuracy from 66.93% to 74.53% for two-class ME, from 44.83% to 56.58% for three-class ME, from 80.78% to 85.63% for two-class MI, and from 60.93% to 69.90% for three-class MI compared with the EEG-only EEGNet baseline. FRIST is also shown to improve EEG-only decoding when the target participant's own fMRI data were unavailable. FRIST also generalized across multiple EEG decoding backbones, reaching 87.40% in two-class MI and 72.54% in three-class MI with EEG Conformer as the EEG feature extractor. These findings indicate that fMRI provide useful spatial constraints for EEG representation learning. FRIST improves noninvasive EEG-based finger-level BCI decoding, offering a multimodal strategy for integrating the spatial specificity of fMRI with real-time applicability of EEG.

cs.LG↗

A substrate booster for P-type 2D ferromagnetic semiconductor

Spin transistors with its both charge and spin properties tuned via electrostatic gating are believed capable for widespread use, which however have proven challenging due to the extreme rareness of their physical base -- magnetic semiconductors. The latter are limited within very few systems including diluted magnetic semiconductors (DMS) and two-dimensional ferromagnetic semiconductors (2D-FMS), and known to suffer from inadequate gate-tunability of their electric and/or magnetic properties. Here, we show a substrate engineering paradigm by interfacing few-layered Cr$_{2}$Ge$_{2}$Te$_{6}$ (FL-CGT) with an antiferromagnetic insulator CrOCl. Owing to the subtle interfacial charge transfer couplings, CGT can be drastically turned from an ambipolar semiconductor into a high performance P-type semiconductor. When cooled below the Curie temperature, the ON-OFF ratio in such substrate-boosted FMS field-effect transistor (FET) reaches 10$^{5}$ with its coercive field $H_{c}$ of magnetic hysteresis loop tunable by a factor of more than 200$\%$, enabling {gate-assisted magnetic switching in the prototype semiconducting spin transistor architecture}. A crossover from critical power-law scaling to a dual power-law behaviour under heavy hole doping was further observed. Our findings {signify} an efficient interfacial charge transfer and electrically modulated magnetic anisotropy energy supported by calculations. This high performance P-type FMS-FET system suggests that active substrate-boosting paradigm might be a powerful path for the investigation of future gate-tunable spintronic devices.

cond-mat.mes-hall↗

Hilbert-space selected switch of helical edges in an artificial quantum Hall insulator

Quantum Hall effects (QHE) host one-dimensional topologically-protected edge channels, which can serve as an essential ingredient in exotic quantum electronic systems. Yet the manual reconstruction of Landau-level topology, by electrostatic confinement or symmetry breaking, remains experimentally challenging. Here, we show that interfacial charge transfer in between CrOCl and large-angle twisted bilayer graphene offsets the two otherwise decoupled Dirac Landau-level ladders in each graphene layer, creating a new sequence of composite filling configurations. At charge neutrality, the composited $(+2,-2)$ state involves only the zeroth Landau levels and becomes fully insulating, with longitudinal resistance reaching the G$Ω$ regime. By contrast, higher composite zero-filling quantum Hall states, including $(+6,-6)$ and $(+10,-10)$, retain counter-propagating helical edge channels and exhibit pronounced non-local transport, reaching up to $50\%$ of the local response. We attribute such switching-behavior to the Landau-spinor Hilbert space -- as the filling is reduced from $(+6,-6)$ to $(+2,-2)$, the orthogonal $N=\pm1$ orbital components are removed, eliminating the edge-compatible channel and gapping both bulk and boundary transport. The interaction nature of the observed gapped sates was further examined both experimentally and theoretically. Our results suggest that charge transfer provides a direct route to engineer artificial quantum Hall insulators, opening possibilities for wavefunction-selective control of helical edge modes.

cond-mat.mes-hall↗

Latent-MoE: Domain-Aware Mixture-of-Experts for PDEs with Multi-Regime Physics

Physics-informed neural networks (PINNs) struggle on PDEs whose governing physics varies across the domain. We trace this to a structural property of standard coordinate networks: their neural tangent kernel (NTK) is translation-variant and lets training points of large coordinate magnitude disproportionately influence predictions elsewhere, producing long-range coupling and gradient conflict during training. We show analytically and empirically that mixture-of-experts (MoE) architectures with centered, compact-support routers yield a uniformly banded NTK whose kernel-regression weights decay exponentially with distance, localizing the learning. Building on this, we propose \emph{Latent-MoE}, which interleaves domain-aware MoE blocks within a shared backbone. Unlike FB-PINNs or X-PINNs, which rigidly partition both the domain and the parameters so that the parameters on different subdomains are updated independently, Latent-MoE is designed to preserve the localization benefit of domain-aware routing while allowing capacity to flow across regions through the shared backbone. On standard homogeneous-physics benchmarks Latent-MoE is competitive with established baselines; on benchmarks with multi-stage time-variable physics, where global models and rigid domain decompositions both fall into spurious solutions, it improves over them by more than an order of magnitude, with markedly reduced gradient conflict during training.

cs.LG↗

MVWeaver: A Hierarchical Music Video Generation Agent with a Learned Song-to-Visual Bridge

Music videos are an important form of audiovisual expression in contemporary culture. They translate and extend the expressive content of songs through deliberate visual design. Existing automatic music video (MV) generation systems can generate visually plausible shots, yet often struggle with long-form coherence and song-grounded visual development. We present MVWeaver, a music video generation agent that integrates hierarchical planning with a learned song-to-visual bridge that translates song understanding into executable shot plans. The MVWeaver architecture comprises a comprehensive song analysis module, a visual planner that constructs hierarchical plans, and downstream image and video generation models that render the planned content. To equip a general-purpose LLM with MV-specific song-to-visual knowledge, we learn a bridge between song analysis and visual planning from real-MV-derived supervision and curate 1,861 real-world song--MV pairs with structured song-side, MV-side, and teacher-inferred song-to-visual rationale annotations. Using these annotations, we perform LoRA-based supervised fine-tuning (SFT) of a large language model to predict song-to-visual bridges that guide hierarchical visual planning. Our experiments demonstrate stronger song-grounded visual translation, richer visual development, and greater conceptual and shot-to-shot coherence, while ablations support the benefits of learned bridge conditioning.

cs.MM↗

Caliber: Cross-Architecture Extraction-Cost Control for Score-Returning APIs

We present Caliber, an output-perturbation defense against model extraction that formulates noise selection as a calibration problem: how much the defense degrades the supervision signal used to train a surrogate, and the provable per-input query cost of recovering the clean logits. To defend against an attacker that uses returned scores for knowledge distillation, Caliber adds independent and identically distributed Gaussian noise to the internal logits. We establish two properties of the resulting perturbed predictions. Monotone agreement degradation: When the clean logits have a unique maximizer, agreement with the clean prediction decreases strictly with the noise scale, so every target in $(1/K,1)$ corresponds to a unique positive scale; task accuracy is bounded by computable lower and upper envelopes. Per-input recovery cost: We derive a closed-form minimax lower bound on the repeated queries needed to recover the clean logits for a fixed input. Caliber normalizes noise variance by the squared median top-two logit margin and fits the resulting noise-utility relationship with a logistic curve, either per model or shared within a task. Across more than thirty model-dataset combinations, per-model calibration achieves mean absolute relative errors of 0.6-1.4%. End-to-end experiments show that surrogate performance generally tracks the configured degradation, while fixed-input averaging follows the expected variance reduction.

cs.LG↗

MatPhaseBench: A Semantics-Guided Benchmark for Materials Phase Diagrams Understanding

Materials phase diagrams are a core knowledge representation in materials science, encoding temperature,composition, phase stability, and phase transformation pathways, with their full understanding requiring thermodynamic mechanism analysis and scientific reasoning. Although VLMs have shown promise in scientific image understanding, their systematic evaluation on such logically complex images demanding deep mechanistic interpretation remains limited, and phase diagrams provide a challenging testbed for this purpose. We introduce MatPhaseBench, a high-quality, high-reliability benchmark for complex scientific image understanding, focused on materials phase diagrams. MatPhaseBench is constructed from 3681 papers in classical materials science journals, from which 200 high-quality diagram-text pairs were selected, covering 189 material systems and 70 elements. The benchmark has three key features: (1)targeting complex scientific image understanding-it moves beyond simple objective tests to open-ended tasks requiring deep comprehension; (2)comprehensive image-text alignment-semantic information associated with images is fully preserved during literature mining and matching; (3) high-quality human-supervised text acquisition-all descriptions undergo strict manual validation. Experimental results show that current VLMs remain substantially behind expert-level understanding: they are largely limited to surface visual perception, lack deep reasoning grounded in thermodynamic mechanisms, have limited domain awareness and expert analytical experience, and perform poorly in distinguishing fine-grained differences in composite or multi-diagram settings. Overall, MatPhaseBench constitutes a challenging research-grade benchmark, providing a foundational platform for complex scientific image understanding, phase diagram analysis, and trustworthy multi-modal AI in science.

cs.CV↗

In-situ operation of amorphous circuits under heavy-ion irradiation

Radiation-hardened electronics using semiconductors beyond silicon are essential for computation and control in extreme environments. Yet complex digital circuits based on such material platforms operating in situ under heavy-ion irradiation remain largely unexplored. Here, we show a timing circuit based on amorphous thin-film semiconductors at the 100-transistor scale, and demonstrate its robust operation through a functional "Hello World" ASCII output sequence. Beyond static device characterization, we evaluate the circuit under powered heavy-ion irradiation using tantalum ions, providing an operationally relevant assessment of radiation tolerance at the system level. Under a high particle flux of 2.5 x 10^3 ions cm^-2 s^-1, the circuit maintains stable operation during the irradiation test, achieving a total fluence of 1 x 10^6 ions cm^-2, establishing a milestone of prolonged powered digital operation under extreme conditions. Our work expands the design space of radiation-tolerant electronics, highlighting amorphous semiconductors as a promising foundation for digital circuits deployed in harsh environments.

cond-mat.mtrl-sci↗

Small Models, Strong Priors: Architectural Inductive Bias for Parameter-Efficient Neural PDE Solvers

Neural PDE solvers have followed the scaling trajectory of vision and language, with recent foundation models reaching billions of parameters. We argue that scale is a poor substitute for architectural inductive bias in this domain: structured priors deliver outsized parameter efficiency, and the pattern of where they succeed and fail is itself informative about what they capture. We instantiate this argument in WaveLiT, an architecture combining a discrete wavelet transform for lossless multi-resolution tokenization, an augmented linear attention block, a shared-weight multiscale feature pyramid, and a wavelet-domain auxiliary loss. Bespoke 1-10M-parameter WaveLiT models compete with foundation models of 100-1000$\times$ their size across eight TheWell benchmarks, with the largest gains on wave and acoustic-dominated benchmarks where the wavelet-multiscale prior fits the dominant dynamical structure and small per-step errors do not compound geometrically under rollout. Trained jointly across all eight benchmarks, a 10M-parameter foundation variant exhibits a structured, physically interpretable transfer pattern -- strongest where the wavelet-multiscale prior matches the dynamics, weakest on chaotic advection-dominated flows. The entire pipeline trains on a single GPU. The results suggest that small-model PDE performance is shaped by architectural inductive bias rather than scale, and that the structure of a prior's failures is a useful empirical signal about its content.

cs.LG↗

GeoBrowse: A Geolocation Benchmark for Agentic Tool Use with Expert-Annotated Reasoning Traces

Deep research agents integrate fragmented evidence through multi-step tool use. BrowseComp offers a text-only testbed for such agents, but existing multimodal benchmarks rarely require both weak visual cues composition and BrowseComp-style multi-hop verification. Geolocation is a natural testbed because answers depend on combining multiple ambiguous visual cues and validating them with open-web evidence. Thus, we introduce GeoBrowse, a geolocation benchmark that combines visual reasoning with knowledge-intensive multi-hop queries. Level 1 tests extracting and composing fragmented visual cues, and Level 2 increases query difficulty by injecting long-tail knowledge and obfuscating key entities. To support evaluation, we provide an agentic workflow GATE with five think-with-image tools and four knowledge-intensive tools, and release expert-annotated stepwise traces grounded in verifiable evidence for trajectory-level analysis. Experiments show that GATE outperforms direct inference and open-source agents, indicating that no-tool, search-only or image-only setups are insufficient. Gains come from coherent, level-specific tool-use plans rather than more tool calls, as they more reliably reach annotated key evidence steps and make fewer errors when integrating into the final decision. The GeoBrowse bernchmark and codes are provided in https://github.com/ornamentt/GeoBrowse

cs.CL↗

Efficient and Versatile Quadrupedal Skating: Optimal Co-design via Reinforcement Learning and Bayesian Optimization

In this paper, we present a hardware-control co-design approach that enables efficient and versatile roller skating on quadrupedal robots equipped with passive wheels. Passive-wheel skating reduces leg inertia and improves energy efficiency, particularly at high speeds. However, the absence of direct wheel actuation tightly couples mechanical design and control. To unlock the full potential of this modality, we formulate a bilevel optimization framework: an upper-level Bayesian Optimization searches the mechanical design space, while a lower-level Reinforcement Learning trains a motor control policy for each candidate design. The resulting design-policy pairs not only outperform human-engineered baselines, but also exhibit versatile behaviors such as hockey stop (rapid braking by turning sideways to maximize friction) and self-aligning motion (automatic reorientation to improve energy efficiency in the direction of travel), offering the first system-level study of dynamic skating motion on quadrupedal robots.

cs.RO↗