arXiv subjects

Yuhan Wang

Publications and source records attributed to Yuhan Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

MM-Spectrum: Multimodal Multi-spectral Molecular Structural Elucidation with a Stable MoE Framework

Inferring molecular structures from multimodal spectroscopic measurements requires integrating complementary yet highly heterogeneous signals. However, the common paradigm of directly concatenating multispectral sequences can exhibit anomalous performance degradation, primarily due to pronounced heterogeneity and the resulting multimodal imbalance across modalities. As a remedy, we propose MM-Spectrum, a sparse Mixture-of-Experts framework tailored for multimodal multispectral spectra-to-structure elucidation. To better match the information characteristics under multispectral imbalance, MM-Spectrum introduces an explicit modality-aware routing mechanism that exposes spectral identity to the router in addition to token content representations. Moreover, it incorporates shared and interaction experts, together with heterogeneous expert capacities, to extract multispectral modality-unique and cross-modal synergistic information while suppressing noise-induced interference. Across full-modality, bimodal, and missing-modality settings on molecular structural elucidation, MM-Spectrum achieves consistent and substantial improvements, supported by ablation studies and interpretability analyses.

cs.LG↗

arXiv2026

One Policy, Many Embodiments: Unified Camera-Centric Action Geometry Pre-training for Heterogeneous Embodied Manipulation

Scaling generalist vision-language-action (VLA) policies is severely bottlenecked by the inherent heterogeneity of embodied data, which spans diverse robot morphologies, camera configurations, and low-level action spaces. Existing paradigms typically address this mismatch through explicit action retargeting, human-to-robot video synthesis, or dataset-specific adaptation branches, fundamentally hindering the joint learning of a unified policy. We introduce UCAG-P, a camera-centric unified action formulation that structurally aligns heterogeneous embodied datasets into a shared geometric action space. Rather than treating robot-specific commands as the shared policy target, UCAG-P represents manipulation through camera-observable anchor motion in image and camera-frame coordinates, treating robot arms, humanoids, and human hands as different embodiments of a common action schema. A geometry-conditioned action translator combines predicted motion with target-embodiment kinematics to produce executable controls. The resulting decoupled architecture allows a shared VLA policy to learn transferable manipulation geometry while retaining embodiment-specific controllability. UCAG-P is trained on 4.03K hours of robot and simulation data and 2.34K hours of human demonstrations. A single checkpoint reaches 98.3% on LIBERO, 88.7% and 89.2% on RoboTwin Easy and Hard, 82.0% zero-shot on LIBERO-Plus, and 62.0% on RoboCasa GR-1, without benchmark-specific fine-tuning.

cs.RO↗

arXiv2026

CCAT: The Prime-Cam Instrument for the Fred Young Submillimeter Telescope -- Overview and Status

Prime-Cam is a first-generation science instrument for the CCAT Observatory's six-meter aperture Fred Young Submillimeter Telescope (FYST), under construction at an elevation of 5600 m on Cerro Chajnantor in Chile's Atacama Desert. Prime-Cam will deliver over ten times greater mapping speed at submillimeter wavelengths than current facilities for unprecedented broadband and spectroscopic measurements in windows between 1.4 -- 0.3 mm (220 -- 850 GHz). When fully populated, Prime-Cam will field over 100,000 kinetic inductance detectors across seven independently optimized instrument modules. With Prime-Cam, the CCAT Collaboration will address a suite of science goals, from Big Bang cosmology, to galaxy evolution and star formation over cosmic time. Prime-Cam is scheduled for integration in FYST in late 2026, followed by a year of early science observations with the 280 and 350 GHz instrument modules. We discuss the design and in-lab testing of the 1.8-m diameter Prime-Cam receiver and 280 GHz instrument module, and give an update on deployment status and early science plans.

astro-ph.IM↗

arXiv2026

The Atacama Cosmology Telescope: Passband Measurements with an Analysis of Systematic Errors

We present measurements of the spectral response of the four Advanced ACTPol (AdvACT) multichroic detector arrays, which observed in frequency bands centered near 30, 40, 90, 150, and 220 GHz. The passbands were measured with a Fourier transform spectrometer (FTS) combined with coupling optics that direct and match the output of the FTS to the AdvACT receiver. We perform optical simulations of the FTS and coupling optics in order to model frequency-dependent systematic effects in this system. We use these simulations to apply corrections to the measurements and evaluate the remaining systematic uncertainty. We combine these systematic estimates with statistical errors to determine the measured passbands and their uncertainties from our FTS measurement, which are typically 0.75 to 1.5% per detector array. We present the passband characteristics for all measurements made by ACT in a common framework. Additionally, we discuss the performance achieved as well as ways to improve passband measurements in the future.

astro-ph.IM↗

arXiv2026

Secrecy Rate Maximization for UAV-Mounted Six-Dimensional Movable IRS-Assisted ISAC Systems

Integrated sensing and communication (ISAC) is a key enabling technology for 6G wireless networks, but its broadcast nature raises a physical-layer security concern when the sensing target can act as a potential eavesdropper. Although intelligent reflecting surfaces (IRSs) can enhance wireless propagation and improve secrecy, existing secure IRS-assisted ISAC designs are mostly limited to fixed deployments and passive phase control, which offer limited spatial adaptability in line-of-sight-dominated low-altitude scenarios. To address this limitation, we investigate an unmanned aerial vehicle (UAV)-mounted six-dimensional movable IRS-assisted secure ISAC system, where the IRS location, orientation, and reflection coefficients are jointly optimized with the BS beamformer to maximize the secrecy rate under communication quality-of-service (QoS), power, unit-modulus, and visibility constraints. The resulting problem is highly non-convex due to the coupled active/passive beamforming variables and the location-and-orientation-dependent (pose-dependent) channel responses. To solve it efficiently, we develop a three-block alternating optimization (AO) framework, in which the active beamformer, IRS pose, and passive reflection vector are updated via linearized ADMM, warm-started particle swarm optimization, and Riemannian gradient descent, respectively. Simulation results show that the proposed design significantly outperforms fixed-location and orientation-only baselines, highlighting the importance of joint translation, rotation, and phase control for secure ISAC.

eess.SP↗

arXiv2026

PhysMaster: Building an Autonomous AI Physicist for Theoretical and Computational Physics Research

Advances in LLM reasoning and tool use have enabled agentic science, yet frontier theoretical and computational physics remains challenging because research requires deep domain expertise, long-horizon reasoning, and reliable numerical computation. We introduce PRL-Bench, a research-reproduction benchmark adapted from 100 Physical Review Letters papers across major areas of modern physics. PRL-Bench distills realistic research workflows into traceable tasks with explicit intermediate artifacts and diverse evaluation rubrics; each task is estimated by domain experts to require more than six hours for a specialized PhD student to reproduce independently. Evaluations show that existing agents remain unreliable on extended research workflows. We therefore present PhysMaster, a scientific agent combining adaptive MCTS-based multi-trajectory exploration with hierarchical memory to improve long-horizon robustness and knowledge accumulation. PhysMaster achieves the highest overall PRL-Bench score of 51.08, outperforming Codex, OpenHands, OpenClaw, and ReAct, and yields relative improvements of 14.13 percent to 93.38 percent across backbone models. Error analysis shows that PhysMaster substantially reduces failures from incomplete long-horizon execution, while remaining bottlenecks lie in physics knowledge and analytical reasoning. Together, PRL-Bench and PhysMaster provide a rigorous benchmark and effective system for advancing autonomous AI research in frontier physics.

cs.AI↗

arXiv2026

ECAD: Expanding Class-Agnostic Detection Beyond Thing-Centric Objectness

Object detection is a fundamental task in visual perception, providing structured region representations for recognition, grounding, reasoning, and interaction. However, existing detection paradigms largely inherit a thing-centric notion of objectness, where detectors are mainly trained to localize discrete and countable object instances. Consequently, many semantically meaningful visual elements, such as sky, road, grassland, water, and sports courts, are often absorbed into the background despite their importance for scene understanding and spatial reasoning. In this paper, we formulate Expanded Class-Agnostic Detection (ECAD), a new setting that aims to discover category-agnostic visual candidates beyond conventional thing-centric objects. To support this setting, we construct BTCO-Bench, a Beyond Thing-Centric Objectness benchmark with category-agnostic box annotations covering both real-world and cross-domain scenarios. We further propose ECADet, a lightweight DETR-based detector built upon a frozen DINOv3 encoder, and introduce Geometry-Aware Expert Regression (GAER) and Prototype-Guided Query Modulation (PGQM) to improve localization and objectness estimation for diverse visual elements, respectively. Extensive experiments show that ECADet consistently outperforms representative class-agnostic and proposal-based detectors on BTCO-Bench, demonstrating the effectiveness of expanded objectness discovery. Code and benchmark will be released.

cs.CV↗

arXiv2026

RegionDet: A Benchmark for Region Detection Beyond Object Instances

Object detection is a fundamental task in computer vision and has achieved remarkable progress on standard benchmarks by localizing discrete and well-bounded object instances. However, many visual targets in real-world scenarios are not individual objects, but regions defined by visual states, scene context, object relations, and human activities, such as construction areas, damaged road regions, queues, group conversations, and vendor regions. Existing detection benchmarks are mainly built around object instances, providing limited support for systematically evaluating such region targets. To address this gap, we introduce Region Detection, a task that extends conventional object detection beyond object instances, and construct RegionDet, a benchmark for region target localization. RegionDet contains eight region categories, including Construction, Crossing, Damage, Queuing, Talking, Vendor, Waiting, and Walking, with COCO-style bounding-box annotations and evaluation protocols. We systematically evaluate representative closed-set and zero-shot/open-vocabulary detectors on RegionDet. Results show that closed-set detectors can partially learn region-level patterns under supervision, while zero-shot/open-vocabulary detectors struggle severely, revealing the strong object-centric bias of current vision-language detectors. Further analyses highlight key challenges in Region Detection, including weak boundary cues, strong context dependency, and insufficient relation-level region understanding. The RegionDet will be released.

cs.CV↗

arXiv2026

CCAT: Design, Implementation, and Testing of a System to Read Out over 10,000 280 GHz KIDs using RFSoC Electronics

Over the past decade, kinetic inductance detectors (KIDs) have emerged as a viable superconducting technology for astrophysics at millimeter and submillimeter wavelengths. KIDs spanning 210 - 850 GHz across seven instrument modules will be deployed in the Prime-Cam instrument of CCAT Observatory's Fred Young Submillimeter Telescope at an elevation of 5600 m on Cerro Chajnantor in Chile's Atacama Desert. The natural frequency-division multiplexed readout of KIDs allows hundreds of detectors to be coupled to a single radio frequency (RF) transmission line, but requires sophisticated warm readout electronics. The FPGA-based Xilinx ZCU111 radio frequency system on chip (RFSoC) offers a promising and flexible solution to the challenge of warm readout. CCAT uses custom packaged RFSoCs to read out KIDs in the Prime-Cam instrument. Each RFSoC can simultaneously read out four RF channels with up to 1,000 detectors spanning a 512 MHz bandwidth per channel using the current firmware. We use five RFSoCs to read out the >10,000 KIDs in the broadband 280 GHz instrument module. Here, we describe and demonstrate the readout hardware, software and pipeline for the RFSoC system. We present a detector position map of the 280 GHz module focal plane and preliminary averaged spectral responses of a small subset of detectors from the TiN and first Al arrays. These measurements demonstrate our ability to simultaneously readout thousands of detectors, validate the end-to-end performance of the readout and optical systems, and represent a critical step toward reading out the ~100,000 KIDs in Prime-Cam in its future full capacity configuration.

astro-ph.IM↗

arXiv2026

RIS-Enabled Energy-Efficient ISAC for Vehicular Applications

By incorporating integrated sensing and communication (ISAC) into vehicle-to-infrastructure (V2I) networks, roadside units (RSUs) can support data transmission while providing additional sensing capabilities, thereby enabling intelligent transportation services. By deploying large-scale antenna arrays at RSU, the V2I network can realize more reliable connectivity and more accurate vehicle tracking by harnessing the significant beamforming gains provided by the enlarged antenna aperture. However, it is not energy efficient to realize such arrays using conventional phased arrays, which rely on numerous power-hungry phase shifters. To address this, this demo presents the first reconfigurable intelligent surface (RIS)-based ISAC-empowered vehicular network prototype, which operates at sub-6 GHz band to be compatible with existing vehicular systems. The RIS has a low power consumption of 6.8W. Experimental results show that compared with the system without RIS, the proposed RIS-based prototype enables more accurate vehicle trajectory tracking with an average localization error of 0.11m and supports more robust data transmission, as evidenced by a 41.9% reduction in error vector magnitude (EVM). These results validate the effectiveness of the RIS-based ISAC system for supporting vehicular networks.

eess.SP↗

arXiv2026

CCAT: Design and Characterization of the 350 GHz Instrument Module

The CCAT Collaboration's Prime-Cam instrument will soon be deployed to the Fred Young Submillimeter Telescope (FYST) in Chile's Atacama Desert. Featuring prominently in Prime-Cam's calibration and early science observations will be the 350 GHz instrument module, a broadband camera that will field more than 10,000 microwave kinetic inductance detectors (KIDs) across three detector arrays. Forecasts show this module will be capable of making the most sensitive to-date measurements of polarized dust emission over a large fraction of the sky at this frequency, enabling new galactic polarization science and improved understanding of cosmological foregrounds. In this work we discuss the design of the 350 GHz instrument module, covering aspects of the optics, readout, and detector arrays. We then report on the results of in-lab testing of the fully-integrated module, achieving stable cryogenic performance with a 100 mK focal plane, high detector yield, and a passband comparable to designed specifications. Upon completion of these tests, this module was shipped to the telescope site in Chile for integration in Prime-Cam.

astro-ph.IM↗

arXiv2026

PAGE-4D: Disentangled pose and geometry estimation for vggt-4d perception

Recent 3D feed-forward models, such as the Visual Geometry Grounded Transformer (VGGT), have shown strong capability in inferring 3D attributes of static scenes. However, since they are typically trained on static datasets, these models often struggle in real-world scenarios involving complex dynamic elements, such as moving humans or deformable objects like umbrellas. To address this limitation, we introduce PAGE-4D, a feedforward model that extends VGGT to dynamic scenes, enabling camera pose estimation, depth prediction and point cloud reconstruction - all without post-processing. A central challenge in multitask 4D reconstruction is the inherent conflict between tasks: accurate camera pose estimation requires suppressing dynamic regions, while geometry reconstruction requires modeling them. To resolve this tension, we propose a dynamics aware aggregator that disentangles static and dynamic information by predicting a dynamics-aware mask - suppressing motion cues for pose estimation while amplifying them for geometry reconstruction. Extensive experiments show that PAGE-4D consistently outperforms the original VGGT in dynamic scenarios, achieving superior results in camera pose estimation, monocular and video depth estimation, and dense point map reconstruction. Necessary code and additional demos are available at Link: https://page4d.github.io/, including both the training-and-inference masking variant and the training-only masking variant (= VGGT architecture at inference). Keywords: VGGT-4D, 4D Perception, Dynamic Scene Reconstruction.

cs.CV↗

arXiv2026

Geometric Generalization of Neural Operators from a Kernel Integral Perspective

Neural operators are neural network-based surrogate models for approximating partial differential equation solution operators, enabling efficient many-query computations in science and engineering when low-to-moderate accuracy is sufficient. Many applications, including engineering design, involve variable and often nonparametric geometries, for which generalization to unseen shapes remains a central practical challenge. In this work, we adopt a kernel-integral perspective motivated by classical boundary integral formulations and study operator learning on variable geometries through the approximation of geometry-dependent kernel operators, including singular kernels. This perspective clarifies a mechanism for geometric generalization for fixed linear operators and reveals a direct connection between operator learning and fast kernel summation methods. Leveraging this connection, we propose a multiscale point cloud neural operator inspired by Ewald summation, combining Fourier long-range interactions with local geometry-aware corrections. We further establish approximation guarantees for the resulting multiscale representation of linear operators defined by singular kernels. Numerical experiments demonstrate robust generalization across diverse geometries for several commonly used kernels, Laplace-type boundary integral maps, and a large-scale three-dimensional nonlinear vehicle flow example.

math.NA↗

arXiv2026

Practice Makes Policies: Bootstrapping and Consolidating Robotic Capabilities from Zero Human Demonstrations

General-purpose robotic manipulation requires robots to perform diverse tasks in open-world environments while improving their skills over time. Despite recent progress in robotic manipulation, existing systems still primarily acquire manipulation skills in a static manner, where capabilities are learned for specific tasks or settings rather than adaptively evolving through physical interaction. Resembling how repeated practice enables humans to develop muscle memory, advanced manipulation proficiency requires an autonomous capability evolution mechanism that allows robots to progressively transform interaction experiences into increasingly effective manipulation abilities. To this end, we propose HERO, a self-improving hierarchical embodied agent that enables autonomous capability evolution from zero human demonstrations. HERO organizes heuristic reasoning, exemplar reuse, and reflexive execution into a unified orchestration framework, allowing robots to autonomously bootstrap manipulation experience, rapidly accumulate reusable behaviors through experience transfer, and progressively consolidate recurring interactions into efficient closed-loop visuomotor policies. By tightly coupling autonomous data collection with task execution, HERO continuously expands and dynamically schedules manipulation capabilities according to different stages of experience accumulation and execution requirements. Extensive experiments demonstrate that HERO substantially reduces human intervention during robotic data collection while achieving robust manipulation across diverse tasks, providing a promising path toward self-improving robotic systems.

cs.RO↗

arXiv2026

Analyzing and Correcting Benevolence Bias in Large Language Models

Large language models (LLMs) are increasingly used as stand-ins for human respondents, from opinion polls and simulated survey participants to agent-based social simulations. These uses rest on one assumption: that conditioning a model on who a person is yields answers resembling those of real people from that group. Here we identify and measure benevolence bias, a small but consistent tendency for aligned LLMs to lean toward the kinder, safer, more socially approved answer on value-laden survey questions. Across 18 widely used models, four social-science datasets (ANES, GSS, WVS, and a cross-cultural prospect-theory replication) and six psychological categories, we find that the bias is a stable model property, not a quirk of any one system: it points the same way across models, grows with model size, and traces to the post-training stage. Prompt language and framing change its size but never its direction, and a "malicious persona" stress test shows a one-sided limit: aligned models struggle to play people who are less kind, less prosocial or more harm-tolerant than average. The issue is thus not only a shifted average, but a narrowed range of people the model can imitate. The bias sits in the middle of the answer distribution rather than its tails, and survives changes in sampling temperature and simple prompted reflection. The encouraging news is that it is easy to diagnose and straightforward to fix: a light-touch contrastive calibration, which needs no retraining and works on black-box APIs, brings all six categories back to the human baseline. Our results give researchers a clear map of where aligned LLMs can already be trusted as human stand-ins, where they need care, and a ready-to-use method for closing the gap.

cs.HC↗

arXiv2026

CCAT: Optical Responsivity, Noise, and Readout Optimization of KIDs for Prime-Cam

The Prime-Cam instrument on the Fred Young Submillimeter Telescope (FYST) at the CCAT Observatory will conduct sensitive millimeter to submillimeter surveys for a range of astrophysical and cosmological sciences. Prime-Cam will use kinetic inductance detectors (KIDs) sensitive to multiple frequency bands spanning 280-850 GHz. With over 100,000 sensors under development, these KID arrays will soon form the largest submillimeter focal plane ever built. With fixed microwave tones probing amplitude and phase modulations in the KIDs due to incoming radiation, challenges arise in determining the optimal readout settings, especially under varying atmospheric loading. Realizing the science goals of FYST requires operating the detectors at optimal performance and determining accurate responsivities, which depend on readout tone placement and power. To address these challenges, we present laboratory measurements of sample pixels from the 280 GHz TiN and Al arrays using a blackbody cold load to simulate observing conditions. These measurements probe detector responsivity and noise across varying optical loading, tone power, and tone placement, providing the foundation to guide in situ calibration and operation of the >100,000 KIDs. We characterize detector sensitivity via the Noise Equivalent Power (NEP) as a function of readout tone power and placement, and measure the impact of detuning due to varying optical power on the NEP. Our test setup and methodology will inform the commissioning of Prime-Cam, in situ detector calibration procedures, the cadence of probe tone resetting, and potential design refinements for future arrays, supporting FYST's planned first light in 2026.

astro-ph.IM↗

arXiv2026

Better Starts, Better Ends: Bootstrapped Iterative Self-Reasoning Distillation for Compressed Reasoning

Large reasoning models often solve problems through long chain-of-thought (CoT) traces, yet much of this computation is spent on redundant derivations, repeated self-verification, and detours that do not improve the final answer. Existing on-policy self-distillation methods reduce this cost by matching a student model to a concise copy of itself on prefixes sampled from the student's own rollouts. We show that this objective has an initialization bottleneck. Since supervision is applied only to visited prefixes, training from a verbose base model places the KL loss on contexts that are often noisy, redundant, or already off track. In such regions, a concise teacher can provide only local corrections, while the student continues to explore trajectories that an efficient reasoner should avoid. In this paper, we propose BIRD(Bootstrapped Iterative Self-Reasoning Distillation), a two-stage self-reasoning distillation method that improves the rollout distribution before on-policy training. BIRD first samples concise solutions from the base model under a brevity instruction, keeps only answer-correct traces, and performs a lightweight prompt-switch SFT step. The traces are generated with the brevity instruction but learned under the original task prompt, turning instruction-induced conciseness into a default reasoning behavior. Starting from this warm model, BIRD then applies on-policy reverse-KL distillation with a concise self-teacher, now on cleaner and more informative prefixes. Across Qwen3 series models, BIRD achieves a stronger accuracy-efficiency trade-off than prompting and cold-start on-policy distillation on MATH-500 and AIME benchmarks. On Qwen3-8B, it improves MATH-500 accuracy from 86.2% to 92.0% while reducing the average response length from 3,099 to 1,115 tokens. These results highlight prefix support as a central factor in efficient reasoning distillation.

cs.CL↗

arXiv2026

How to Guide LLM Generation: Dual-Surrogate Guided Search for Automated Heuristic Design

Large language models (LLMs) have made automated heuristic design (AHD) increasingly practical by generating executable heuristic code from task descriptions and evaluator feedback. Yet under a limited query and evaluation budget, search efficiency depends critically on a pre-generation decision. Before each LLM query and black-box evaluation, the system must choose which archived heuristics to reuse as parents and which generation operator should transform them. Existing methods typically choose such actions with predefined rules, leaving the expected outcome of each concrete operator-parent action only indirectly modeled. Therefore, we propose \emph{\fullmethod{}} (\method{}), a surrogate-guided action-selection module for operator-parent selection in LLM-based AHD. \method{} guides the LLM code-generation process by scoring pre-generation actions with two complementary surrogates. Specifically, a transition surrogate is proposed to predict the latent distribution of the child representation induced by an operator-parent action, while an instance-conditioned utility surrogate is proposed to estimate the expected performance of sampled child latents. Moreover, we propose an uncertainty-aware acquisition rule that combines predicted utility, utility uncertainty, and transition uncertainty to select the next LLM generation action. Across a diverse heuristic-design suite, \method{} is competitive with strong LLM-AHD baselines, and ablation and action-selection analyses suggest that its behavior goes beyond simple archive ranking or fixed operator preferences.

cs.NE↗