Search arXiv⌕ Search

arXiv subjects

Liang Wu

Publications and source records attributed to Liang Wu.

At least 55 records · Page 3Linked to original sources
arXiv2026

PhoneBuddy: Training Open Models for Agentic Phone Use

Phones are becoming an important execution surface for general-purpose agents, but training open models for reliable phone use remains difficult because the environment that matters at deployment, real devices running real apps, is slow, stateful, side-effectful, and hard to reset or verify, while scalable mock environments only approximate real behavior. We present PhoneBuddy, a training recipe and open-model line for agentic phone use that combines a real-app environment with a mock-app environment, PhoneWorld, which reconstructs runnable mock apps from real GUI usage structure. PhoneBuddy first builds a shared supervised fine-tuning stage from trajectories collected in both environments, then compares real-app RL against mixed RL across both environments. Across a 150-task human evaluation on real phones spanning apps, mini-apps, and cross-app workflows, task success rate improves from 36.67\% after supervised fine-tuning to 40.67\% after real-app RL and 45.33\% after mixed RL. On AndroidWorld, the same progression rises from 60.3\% to 77.2\% to 83.2\%. These results show that mock-app training is not a replacement for real-app RL, but a complementary source of scalable, resettable, and automatically checked interaction. The gains are strongest on app and mini-app tasks, while long-horizontal cross-app workflows remain an important open challenge.

cs.CL↗
arXiv2026

E-VAds: An E-commerce Short Videos Understanding Benchmark for MLLMs

E-commerce short videos represent a high-revenue segment of the online video industry characterized by a goal-driven format and dense multi-modal signals. Current models often struggle with these videos because existing benchmarks focus primarily on general-purpose tasks and neglect the reasoning of commercial intent. In this work, we first propose a multi-modal information density assessment framework to quantify the complexity of this domain. Our evaluation reveals that e-commerce content exhibits substantially higher density across visual, audio, and textual modalities compared to mainstream datasets, establishing a more challenging frontier for video understanding. To address this gap, we introduce E-commerce Video Ads Benchmark, which is the first benchmark specifically designed for e-commerce short video understanding. We curated 3,961 high-quality videos from Taobao covering a wide range of product categories and used a multi-agent system to generate 19,785 open-ended Q&A pairs, which consist of five distinct tasks. Finally, we develop E-VAds-R1, an RL-based reasoning model featuring a multi-grained reward design called MG-GRPO. This strategy provides smooth guidance for early exploration while creating a non-linear incentive for expert-level precision. Experimental results demonstrate that E-VAds-R1 achieves a 109.2% performance gain in commercial intent reasoning with only a few hundred training samples. Data is available at https://github.com/TaobaoTmall-AlgorithmProducts/E-VAds_Benchmark.

cs.CV↗
arXiv2026

PhoneHarness: Harnessing Phone-Use Agents through Mixed GUI, CLI, and Tool Actions

Phone agents are increasingly expected to complete real mobile workflows rather than merely predict the next screen action. However, much of the current mobile-agent literature still evaluates agents primarily as GUI controllers that observe a screen, emit taps and swipes, and are scored by target app state. Real phone-use tasks are broader: they require deciding when to use app GUIs, device-side commands, or structured tools, while leaving evidence that the intended side effect actually occurred. We introduce PhoneHarness, a mixed-action benchmark and execution harness for studying phone-use agents on verifiable mobile workflows. PhoneHarness runs a device-side agent loop over GUI, CLI, and host-side tool actions, combining deterministic action routing with bounded GUI delegation and auditable execution traces. Its benchmark, PhoneHarness Bench, evaluates whether agents complete tasks with observable side effects, not only whether they produce plausible final answers. On the annotated evaluation split, PhoneHarness reaches a 75.0% pass rate, outperforming the strongest non-PhoneHarness settings by 12.9 percentage points. PhoneHarness and PhoneHarness Bench therefore play distinct but mutually dependent roles: the harness makes mixed phone workflows executable, while the benchmark measures whether agents can use that harness reliably and safely. Our findings suggest that reliable phone automation depends on action-surface routing and verifiable execution, not only visual GUI control.

cs.CL↗
arXiv2026

HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling

Multimodal Large Language Models (MLLMs) have made significant strides in visual understanding tasks. However, their performance on high-resolution images remains suboptimal. While existing approaches often attribute this limitation to perceptual constraints and argue that MLLMs struggle to recognize small objects, leading them to use "zoom in" strategies for better detail, our analysis reveals a different cause: the main issue is not object size, but rather caused by complex background interference. We systematically analyze this "zoom in" operation through a series of decoupling experiments and propose the Hierarchical Decoupling Framework (HiDe), a training-free framework that uses Token-wise Attention Decoupling (TAD) to decouple the question tokens and identify the key information tokens, then leverages their attention weights to achieve precise alignment with the target visual regions. Subsequently, it employs Layout-Preserving Decoupling (LPD) to decouple these regions from the background and reconstructs a compact representation that preserves essential spatial layouts while eliminating background interference. HiDe sets a new SOTA on V*Bench, HRBench4K, and HRBench8K, boosting Qwen2.5-VL 7B and InternVL3 8B to SOTA (92.1% and 91.6% on V*Bench), even surpassing RL methods. After optimization, HiDe uses 75% less memory than the previous training-free approach. Code is provided in https://tennine2077.github.io/HiDe.github.io/.

cs.CV↗
arXiv2026

Accelerating and Scaling MPC-Guided Reinforcement Learning for Humanoid Locomotion and Manipulation

In humanoid motion control, model predictive control (MPC) offers physically grounded prediction and constraint handling, while reinforcement learning (RL) enables robust whole-body skills through large-scale simulation. However, using MPC inside RL often requires time-consuming problem construction or excessive training overhead, making such frameworks difficult to justify in practice. This work studies efficient training-time MPC guidance for humanoid locomotion and manipulation, termed MPC-RL. We introduce a centroidal-dynamics MPC reward formulation that leverages guidance from MPC trajectories in training time. To make this practical in massively parallel RL, we develop $π^n$MPC, a parallel-in-horizon and construction-free batched GPU MPC solver that operates directly on time-varying dynamics to avoid high memory usage and pre-compilation. Through a variety of comparative studies and hardware validations, we have found that MPC-RL achieves superior performance in locomotion and manipulation skills. The code base is available at https://github.com/junhengl/mpc-rl.

cs.RO↗
arXiv2026

PageLLM: A Multi-Grained Reward Framework for Whole-Page Optimization with Large Language Models

Whole-page optimization (WPO) decides how search and recommendation results are surfaced to users, and large language models (LLMs) open a new route to it by treating page generation as sequence generation. Adapting LLMs to web-scale WPO, however, remains bottlenecked by the need for costly human annotations and by the mismatched granularity between page-level coherence and item-level placement. In this work we show that these two challenges are coupled: implicit user feedback alone suffices for alignment, provided the reward signal is decoupled into two complementary granularities. We propose PageLLM, a reward-based fine-tuning framework that (i) turns implicit feedback into four contrastive preference-pair families covering relevance, ranking, diversity, and redundancy, (ii) learns a coarse page-level reward and a fine item-level reward that captures engagement-sensitive position swaps, and (iii) combines both rewards in PPO-based RLHF over a pre-trained LLM. Extensive experiments on seven Amazon categories against eleven baselines show that neither reward alone is sufficient -- dropping the page-level or item-level signal reduces NDCG@100 by 17.8% and 15.2% respectively, whereas the joint reward improves NDCG@100 by up to 46.8%. Deployed in a 10M-user online A/B test, PageLLM raises GMV by 0.44% and click-through rate by 0.14%, confirming that multi-grained rewards from implicit feedback scale to production WPO. Code and data are available at an anonymized repository.

cs.LG↗
arXiv2026

Observation of magnetically switchable quantum geometric photocurrents

In non-centrosymmetric materials, light can be rectified into two types of DC photocurrents, known as injection and shift currents, through the bulk photovoltaic effect. Recent theory has uncovered their deep relation with the two-state quantum geometry of resonant transitions: In non-magnetic crystals, where these currents have been routinely observed, the injection current responds to circular light and probes the Berry curvature, while the shift current responds to linear light and probes the geometric connection. Magnetic crystals have been predicted to show a new set of hitherto unobserved magnetically switchable photocurrents, with the roles of linear and circular light interchanged: A linear injection current, which probes the quantum metric, and a circular shift current, which probes the geometric torsion. In this work, we demonstrate the existence of such currents for the first time, demonstrating the switching of the current by flipping the Néel vector in a van der Waals antiferromagnet. Furthermore, their specific frequency and temperature dependence confirm the assignment of circular shift and linear injection currents. Our work demonstrates a new way to control photocurrents in magnets that are directly tied to geometry and have promising applications in antiferromagnetic spintronics and light harvesting.

cond-mat.mtrl-sci↗
arXiv2026

$π$MPC: A Parallel-in-horizon and Construction-free NMPC Solver

The alternating direction method of multipliers (ADMM) has gained increasing popularity in embedded model predictive control (MPC) due to its code simplicity and pain-free parameter selection. However, existing ADMM solvers either target general quadratic programming (QP) problems or exploit sparse MPC formulations via Riccati recursions, which are inherently sequential and therefore difficult to parallelize for long prediction horizons. This technical note proposes a novel \textit{parallel-in-horizon} and \textit{construction-free} nonlinear MPC algorithm, termed $π$MPC, which combines a new variable-splitting scheme with a velocity-based system representation in the ADMM framework, enabling horizon-wise parallel execution while operating directly on system matrices without explicit MPC-to-QP construction. Numerical experiments and accompanying code are provided to validate the effectiveness of the proposed method.

math.OC↗
arXiv2026

Towards Resilient and Autonomous Networks: A BlueSky Vision on AI-Native 6G

The proliferation of emerging applications, such as autonomous driving and immersive experiences, demands cellular networks that are not only faster, but fundamentally more resilient and autonomous. This paper presents a BlueSky vision on how Artificial Intelligence will be natively integrated into 6G, shifting the paradigm from \underline{Network for AI} to \underline{AI for Network}. We envision that, unlike 5G's reliance on scattered, ad-hoc models each trained for a single task, native AI in the 6G era will be anchored by a foundation model and and orchestrated via collaborative multi-agent systems, framing network management as a unified, multi-modal, multi-task optimization problem. Built on this vision, we outline two transformative directions. The first focuses on developing a 6G foundation model as a unified backbone, with task-specific knowledge distilled into compact models suited for diverse edge deployments. The second advances multi-agent systems designed to autonomously diagnose, maintain, and recover networks with minimal human intervention. These directions chart a roadmap for 6G to evolve into an intelligent, self-sustaining communication infrastructure.

cs.AI↗
arXiv2026

Accelerated and data-efficient flow prediction in stirred tanks via physics-informed learning

The simulation of fluid flows is computationally expensive due to the complexity of its governing partial differential equations. Machine learning models offer a potential surrogate, enabling learning from simulations and significantly faster predictions of flow fields. However, these models require large training datasets, which introduces a trade-off between dataset generation cost and predictive accuracy. In this work, we investigate the relationship between the size of the training-set and accuracy of the prediction when learning steady flow fields in an industrial-scale stirred vessel. A data set of steady flows is generated using Reynolds Averaged Navier Stokes (RANS) simulations in a range of realistic operating conditions, including impeller speeds and liquid heights. We train implicit neural representations of flow fields and compare purely data-driven and constrained variants. Model performance is evaluated using global mean squared error (MSE), qualitative spatial comparisons of predicted and reference flow fields, and tracer transport simulations. We find that the prediction error decreases monotonically with increasing training data, but also that it exhibits clear diminishing returns beyond moderate dataset sizes. Physics-based constraints significantly improve accuracy and reduce variability across training runs in low-data regimes, and they lead to more stable tracer-transport behavior. Furthermore, reasonable interpolation can be achieved over different impeller speeds and liquid heights. However, these benefits come with an increase in the complexity of training, and their relative advantage diminishes as the training set grows.

cs.CE↗
arXiv2026

Weak localization and universal conductance fluctuations in large area twisted bilayer graphene

We study diffusive magnetotransport in highly p-doped large area twisted bilayer graphene in 1°, 7°, 9° and 20° samples. We report weak localization in twisted bilayer graphene for the first time. All samples exhibit weak localization, from which we extract the phase coherence length and intervalley scattering lengths, and from that determine that dephasing is caused by electron-electron scattering and intervalley scattering is caused by point defects. We observe signatures of universal conductance fluctuations in the 9° sample, which has high mobility and is near the van Hove singularity. Further improvements in sample quality and applications to large area moire materials will open new avenues to observe quantum interference effects.

cond-mat.mes-hall↗
arXiv2026

Towards Real-World Document Parsing via Realistic Scene Synthesis and Document-Aware Training

Document parsing has recently advanced with multimodal large language models (MLLMs) that directly map document images to structured outputs. Traditional cascaded pipelines depend on precise layout analysis and often fail under casually captured or non-standard conditions. Although end-to-end approaches mitigate this dependency, they still exhibit repetitive, hallucinated, and structurally inconsistent predictions - primarily due to the scarcity of large-scale, high-quality full-page (document-level) end-to-end parsing data and the lack of structure-aware training strategies. To address these challenges, we propose a data-training co-design framework for robust end-to-end document parsing. A Realistic Scene Synthesis strategy constructs large-scale, structurally diverse full-page end-to-end supervision by composing layout templates with rich document elements, while a Document-Aware Training Recipe introduces progressive learning and structure-token optimization to enhance structural fidelity and decoding stability. We further build Wild-OmniDocBench, a benchmark derived from real-world captured documents for robustness evaluation. Integrated into a 1B-parameter MLLM, our method achieves superior accuracy and robustness across both scanned/digital and real-world captured scenarios. All models, data synthesis pipelines, and benchmarks will be publicly released to advance future research in document understanding.

cs.CV↗
arXiv2026

Channel Knowledge Map-Enabled NLoS ISAC Localization

Accurate localization in non-line-of-sight (NLoS) environments remains challenging even with both angle-of-arrival (AoA) and time-of-arrival (ToA) measurements. In complex urban scenarios, the absence of line-of-sight (LoS) paths and the lack of environment prior knowledge make geometric based localization methods inapplicable, while prior-based approach such as fingerprinting is sensitive to environmental perturbations. This paper proposes a novel environment-aware localization framework enabled by the emerging concept called channel knowledge map (CKM). In the offline stage, AoA-ToA path signatures are learned by the CKM, with each path mapped to one candidate scatterer, thereby forming geometric priors within the environment. In the online stage, observed paths are matched to the CKM to extract high-confidence scatterers. Nonlinear least squares (NLS) method is then applied to jointly estimate the user and dominant scatterer locations. Even with imperfect CSI matching, geometric feasibility consistent with CKM scatterer priors provides corrective information and suppresses ambiguity. Simulations demonstrate that the proposed scheme outperforms fingerprinting and offers a robust and scalable solution to address the challenging NLoS localization for integrated sensing and communication (ISAC) systems.

eess.SP↗
arXiv2026

Fixed-time-stable ODE Representation of Lasso

Lasso problems arise in many areas, including signal processing, machine learning, and control, and are closely connected to sparse coding mechanisms observed in neuroscience. A continuous-time ordinary differential equation (ODE) representation of the Lasso problem not only enables its solution on analog computers but also provides a framework for interpreting neurophysiological phenomena. This article proposes a fixed-time-stable ODE representation of the Lasso problem by first transforming it into a smooth nonnegative quadratic program (QP) and then designing a projection-free Newton-based ODE representation of the Lasso problem by first transforming it into a smooth nonnegative quadratic program (QP) and then designing a projection-free Newton-based fixed-time-stable ODE system for solving the corresponding Karush-Kuhn-Tucker (KKT) conditions. Moreover, the settling time of the ODE is independent of the problem data and can be arbitrarily prescribed. Numerical experiments verify that the trajectory reaches the optimal solution within the prescribed time.

math.OC↗
arXiv2026

Polynomial Parametric Koopman Operators for Stochastic MPC

This paper develops a parametric Koopman operator framework for Stochastic Model Predictive Control (SMPC), where the Koopman operator is parametrized by Polynomial Chaos Expansions (PCEs). The model is learned from data using the Extended Dynamic Mode Decomposition -- Dictionary Learning (EDMD-DL) method, which preserves the convex least-squares structure for the PCE coefficients of the EDMD matrix. Unlike conventional stochastic Galerkin projection approaches, we derive a condensed deterministic reformulation of the SMPC problem whose dimension scales only with the control horizon and input dimension, and is independent of both the lifted state dimension and the number of retained PCE terms. Our framework, therefore, enables efficient nonlinear SMPC problems with expectation and second-order moment constraints with standard convex optimization solvers. Numerical examples demonstrate the efficacy of our framework for uncertainty-aware SMPC of nonlinear systems.

eess.SY↗
arXiv2026

Tracing Star Formation in Quasar Hosts via [O II] $λ$3727: A Kinematically Consistent Approach

Measuring star formation in quasar host galaxies is crucial for understanding the coevolution of supermassive black holes (SMBHs) and galaxies, yet remains observationally challenging due to severe contamination from active galactic nucleus (AGN) emission. In this work, we present a new method to robustly isolate the AGN contribution to the [O II] $λ$3727 emission line in quasars, based on a kinematically consistent decomposition of [O II] and the high-ionization [Ne V] $λ$3426 line. We find that the [O II] emission in quasars is primarily dominated by star formation, with only a weak AGN contribution, and thus can be reliably used as a tracer of star formation in quasar hosts. Applying this technique to a large sample of Sloan Digital Sky Survey quasars, we derive mean SFRs as a function of bolometric luminosity. We find a tight correlation between mean SFR and luminosity. Further analysis, assuming a constant dust extinction correction to [O II] emission, shows that luminosity is the primary parameter most strongly associated with star formation, rather than SMBH mass or Eddington ratio. This supports the scheme in which star formation and black hole accretion are closely linked through their common dependence on the cold gas supply.

astro-ph.GA↗
arXiv2026

Intrinsic Even-Odd Thickness-Driven Anomalous Hall in Epitaxial MnBi2Te4 Thin Films

We demonstrate precise control of magnetism in MnBi2Te4 thin films through careful synthesis by molecular beam epitaxy, achieving minimal defects and accurate layer thickness control. By optimizing Mn-Bi-Te ratios and growth temperatures, we minimize detrimental self-doping effects and accurately target integer-layer films. X-ray diffraction and reflectivity provide quantitative measures of film quality and thickness. When these macroscale probes of structure and thickness are integrated with magnetotransport measurements, a striking even-odd layer dependence of the anomalous Hall effect is revealed. Odd-layer films exhibit a large hysteresis up to the Néel temperature (~25K), consistent with non-compensated antiferromagnetism, while even-layer films show minimal response, as expected for an antiferromagnet. The sign of the anomalous Hall effect exhibits a sign reversal for intrinsic magnetism versus magnetism associated with defects. This work identifies critical factors for inducing pure, non-compensated ferromagnetism and reveals the character of the intrinsic anomalous Hall effect in MnBi2Te4, which together is a step towards realizing the zero-field quantum anomalous Hall effect in topological materials.

cond-mat.mtrl-sci↗
arXiv2026

Koopman-BoxQP: Solving Large-Scale NMPC at kHz Rates

Solving large-scale nonlinear model predictive control (NMPC) problems at kilohertz (kHz) rates on standard processors remains a formidable challenge. This paper proposes a Koopman-BoxQP framework that i) learns a linear Koopman high-dimensional model, ii) eliminates the high-dimensional observables to construct a multi-step prediction model of the states and control inputs, iii) penalizes the multi-step prediction model into the objective, which results in a structured box-constrained quadratic program (BoxQP) whose decision variables include both the system states and control inputs, iv) develops a structure-exploited and warm-starting-supported variant of the feasible Mehrotra's interior-point algorithm for BoxQP. Numerical results demonstrate that Koopman-BoxQP can solve a large-scale NMPC problem with $1040$ variables and $2080$ inequalities at a kHz rate.

eess.SY↗