arXiv Science⌕ Search

arXiv subjects

Xuan Liu

Publications and source records attributed to Xuan Liu.

At least 19 recordsLinked to original sources
arXiv2026

Classification of threshold dynamics for the focusing mass-critical inhomogeneous NLS

We study the focusing mass-critical inhomogeneous nonlinear Schrödinger equation $i\partial_{t}u+Δu+|x|^{-b}|u|^{\frac{4-2b}{d}}u=0$ in dimensions $d\ge3$. For $\frac12<b<\min\{2,d/2\}$, we prove that every forward-global $H^1$ solution with mass equal to that of the ground state $Q$ either scatters forward or coincides with $e^{it}Q$, up to phase rotation and scaling. As a consequence, every solution with nonradial $H^1$ initial data at the ground-state mass must be global and scatter in both time directions. For $0<b\le\frac12$, analogous rigidity and scattering results hold under an additional finite-variance assumption. The key new ingredients in our proof are a nonradial incoming/outgoing decomposition and exterior frequency-decay estimates, which yield the kinetic-energy localization needed for a localized virial argument.

math.AP↗
arXiv2026

Almost global existence for quadratic Schrödinger equations in two dimensions

We study the two-dimensional quadratic Schrödinger equation $(i\partial_t+Δ)u=λ_1u^2+λ_2\overline u^{\,2}$ with initial data of size $\varepsilon$ in a weighted $H^2$ space. We prove almost global existence, with a lifespan of at least $\exp(c/\varepsilon)$ and decay at the optimal linear dispersive rate $t^{-1}$. Our proof is based on the space--time resonance method introduced by Germain, Masmoudi, and Shatah. The main new ingredient is a Besov-type profile norm $\mathcal Z$, which allows us to close the bootstrap argument without controlling the global second spatial momentof the free profile, thereby avoiding the substantial time-dependent losses arising from direct second-moment estimates.

math.AP↗
arXiv2026

$R^2E:$ A Role-driven Reward Evolutionary Framework for Automated Reward Function Design

Designing effective reward functions is a critical challenge in Reinforcement Learning (RL), which traditionally requires costly manual trial-and-error. While Large Language Model (LLM)-based methods have shown promise in reward design, undifferentiated sampling and limited use of historical feedback often lead to insufficient exploration of the reward space and optimization instability. To address these challenges, we propose $R^2E$, a Role-driven Reward Evolutionary framework for automated reward function design that structures the process through explicit role specialization. $R^2E$ employs multiple LLM roles with complementary objectives: an Explorer that promotes novelty to expand the search space, a Guardian that performs conservative refinements to improve training stability, and an Evolver that recombines high-performing rewards via crossover and mutation to integrate effective structures. A dynamic scheduling strategy coordinates these roles, progressively shifting the search from broad exploration to focused exploitation. To mitigate optimization instability and provide a medium for cross-role collaboration, $R^2E$ incorporates a Global Elite Pool that retains the best-performing rewards to guide subsequent generations, thereby enabling coordinated interaction among roles and ensuring a reliable and consistent refinement process. Extensive experiments on multiple robotic tasks demonstrate the effectiveness of our method.

cs.RO↗
arXiv2026

Ontology Refinement with Embeddings and LLMs: A Case Study on CASTON

Ontologies are widely used to represent and organize complex knowledge in domains such as biomedicine and healthcare. However, constructing reliable ontologies remains challenging, as class hierarchies, whether manually or automatically constructed, may contain subtle subsumption errors that require careful refinement. We study this problem in CASTON, a casting-domain ontology, and present a lightweight but scalable refinement workflow that combines a fine-tuned ontology embedding model with Large Language Models (LLMs). The embedding model is used to extract a limited number of candidate axioms for refinement together with possible replacement candidates, and an LLM then returns repair suggestions. In a small expert evaluation, the default configuration using a CASTON-fine-tuned OnT embedding model and Qwen3-8B produced reliable high-confidence repairs: all six high-confidence replacements were accepted by both domain experts. The implementation is available at https://github.com/HuiYang1997/OntRefine.

cs.AI↗
arXiv2026

Photoacoustic-guided reflection-mode wavefront shaping through dynamic biological tissue

Strong optical scattering limits the depth at which light can be focused and used for imaging in biological tissue. Photoacoustic (PA)-guided wavefront shaping enables non-invasive light focusing in scattering media, but most existing implementations require optical and acoustic access from opposite sides of the sample and a rigidly aligned free-space optical path, which have confined these demonstrations to thin or static samples. Here, we combine a real-valued intensity transmission matrix (RVITM) algorithm with a photoacoustic guidestar (PA-RVITM) in a reflection-mode configuration. Modulated light is delivered through a flexible multimode fiber passing through the central aperture of a focused ultrasound transducer, enabling optical illumination and acoustic detection from the same side of the sample without a free-space combiner. Fiber integration preserved the wavefront shaping performance, and the reflection-mode PA-RVITM system achieved sub-second wavefront optimization with a ~14-fold PA signal enhancement through a diffuser. In ex vivo chicken breast tissue, the enhancement decreased with tissue thickness, reaching ~3.5 through a 2-mm chicken breast layer, with a speckle decorrelation time of ~450 ms. By performing an independent RVITM optimization at each scan position, our system enabled photoacoustic microscopy of a black nylon wire through the same layer, reducing the imaged line width from ~141 μm to ~102 μm, close to the nominal diameter of ~100 μm. This work advances PA-guided wavefront shaping toward a compact, same-side implementation for future biomedical therapeutic and imaging applications.

physics.optics↗
arXiv2026

ControBench: An Interaction-Aware Benchmark for Controversial Discourse Analysis on Social Networks

Understanding how people argue across ideological divides online is important for studying political polarization, misinformation, and content moderation. Existing datasets capture only part of this problem: some preserve text but ignore interaction structure, some model structure without rich semantics, and others represent conversations without stable user-level ideological identity. We introduce ControBench, a benchmark for controversial discourse analysis that combines heterogeneous social interaction graphs with rich textual semantics. Built from Reddit discussions on three topics, Trump, abortion, and religion, ControBench contains 7,370 users, 1,783 posts, and 26,525 interactions. The graph contains user and post nodes connected by semantically enriched edges; in particular, user-comment-user edges encode both a reply and the parent comment that it responds to, preserving local argumentative context. User labels are derived from self-declared Reddit flairs, providing a scalable proxy for ideological identity without manual annotation. The resulting datasets exhibit low or negative adjusted homophily (Trump: -0.77, Abortion: 0.06, Religion: 0.04), reflecting the cross-cutting structure of real-world debate. We evaluate graph neural networks, pretrained language models, and large language models on ControBench and observe distinct performance patterns across topics and model families, especially when ideological boundaries are ambiguous. These results position ControBench as a challenging and realistic benchmark for controversial discourse analysis.

cs.CL↗
arXiv2026

When LLM Reward Design Fails: Diagnostic-Driven Refinement for Sparse Structured RL

For sparse, structured reinforcement-learning tasks with semantic reward-function interfaces, LLM-generated reward shaping is better framed as debugging than one-shot generation. We study PPO-trained agents using MiniGrid as core evaluation and MuJoCo reaching/locomotion as boundary stress tests. Our qualitative audit finds two recurring one-shot failure modes -- reward flooding and semantic/API misunderstanding -- and a rarer, less reliably distinguished weak-shaping case. We propose diagnostic-driven iterative refinement, using training diagnostics and a failure-mode taxonomy to guide reward-function revision. On sparse structured tasks with diagnosable reward failures, refinement improves DoorKey-8x8 from 2.3% success without shaping to 97.6%, and KeyCorridor from 31.2% one-shot to 86.7% with high seed-to-seed variance. Controls show these gains are not from generic retrying or extra training alone: metrics-only re-prompting causes large drops, while a static-vocabulary control recovers much of the gap. Thus taxonomy prompting is a major mechanism, with only partially isolated incremental evidence for dynamic labels. In prompt-matched comparisons across Haiku 4.5, Opus 5, and DeepSeek V4 Pro, effects depend on task and model; none of the six contrasts survives family-wise correction. A four-call, episode-budget-matched population control is also competitive. Component-removal and sensitivity tests reveal calibration limits. In continuous control, success-based diagnostics can misfire in dense-reward locomotion; return-trend feedback removes one false-positive mechanism without robust locomotion gains. The method is bounded to sparse structured tasks with reliable structured interfaces under PPO training.

cs.LG↗
arXiv2026

DCVD: Dual-Channel Cross-Modal Fusion for Joint Vulnerability Detection and Localization

Software vulnerability detection plays a critical role in ensuring system security, where real-world auditing requires not only determining whether a function is vulnerable but also pinpointing the specific lines responsible. However, existing approaches either rely on a single information source -- sequential, structural, or semantic -- failing to jointly exploit the complementary strengths across modalities, or treat statement-level localization merely as a byproduct of function-level detection without explicit line-level supervision. To address these limitations, we propose DCVD (Dual-Channel Cross-Modal Vulnerability Detection), a unified framework that performs joint function-level detection and statement-level localization. DCVD extracts control-dependency and semantic features through two parallel branches and integrates them via contrastive alignment coupled with bidirectional cross-attention, effectively bridging the cross-modal representation gap. It further introduces explicit supervision signals at both the function and statement levels, enabling collaborative optimization across the two granularities. Extensive experiments on a large-scale real-world vulnerability benchmark demonstrate that DCVD consistently outperforms state-of-the-art methods on both function-level detection and statement-level localization. Our code is available at https://github.com/vinsontang1/DCVD.

cs.CR↗
arXiv2026

TAP3D: Thermal-Assisted 3D Human Point Clouds

Human body point clouds are a versatile representation for AI-enabled human sensing. However, existing methods using LiDAR, radar, and depth cameras suffer from inherent drawbacks in high cost, sparse reconstruction, and privacy concerns, etc. In this paper, we exploit low-cost thermal arrays and present TAP3D, the first system to reconstruct 3D human point clouds from body heat signatures, offering significant advantages in cost, density, human sensitivity, and privacy. To overcome major challenges in depth estimation, thermal interference, and multi-person separation, we propose a novel physics-informed design, which integrates a forward thermal physics model with two distinct modules: multi-primitive estimation for self-supervised joint recovery of depth and other thermal properties, and geometric perspective fusion for suppressing interference and disentangling multiple people. We implement TAP3D using a single commodity thermal array sensor and build a large-scale dataset (160K samples, 8 environments, 11 users) for evaluation. TAP3D achieves remarkable accuracy for dense point cloud generation, enabling downstream tasks like fall detection (91.46%), indoor tracking (21.86 cm MAE), and human mesh recovery (4.87 cm error). By transforming body heat into point clouds for the first time, TAP3D pioneers a new paradigm for privacy-first, fully passive human sensing for many applications. TAP3D is open-sourced at https://github.com/aiot-lab/TAP3D.

cs.CV↗
arXiv2026

Where Do Test-Time Scaling and Training Fall Short in Individual Stance Prediction?

Test-time scaling and post-training have improved LLM performance in coding and mathematical reasoning, but their effectiveness for individual stance prediction remains unclear. We study this question by predicting a person's stance in a new discussion from their history. We evaluate widely used test-time scaling strategies and post-training methods, such as supervised fine-tuning and reinforcement learning, and identify four failure modes across generation, selection, and learning: (1) incorrect consensus, where repeated samples agree on the wrong stance; (2) selection failure, where generation covers the observed stance but selection misses it; (3) response overfitting, where supervised fine-tuning improves imitation but harms prediction; and (4) early plateau, where reinforcement learning shows modest initial gains followed by limited further improvement. We expose these failures using STANCE-BENCH, which contains 2499 prediction tasks from 500 Hacker News users. Guided by this analysis, we explore a simple approach that combines direct scores for all candidate stances with explicit assessments of support from the individual's history. On the 781-task test set, this approach achieves 21.83 discussion-specific Macro F1 with Qwen3-8B, compared with 19.27 for direct scoring. Our results motivate evaluating candidate generation, final selection, and person-specific evidence use separately. Our data is available at https://github.com/stance-bench/Stance-Bench.

cs.CL↗
arXiv2026

Reach or Solve? Deep Diving into Agentic RL Gains with Checkpoint Handoffs

Reinforcement learning (RL) is widely used to improve language-model agents, and its gains are usually measured by final task success. However, an agent's earlier actions shape the states in which its later decisions are made, so final task success conflates the ability to reach useful states with the ability to complete the task once there. Comparing agents only on the states each one reaches does not separate the two, since each agent is then scored on states selected by its own actions. To address this conflation, we introduce checkpoint handoff, an evaluation protocol that decouples reaching from completing without retraining. One checkpoint acts as a reacher up to a handoff point, and another continues as the solver from the same replayed history. In detail, (1) Reach measures how often a reacher arrives at states that a replayable environment verifies to be a fixed number of actions from success, and (2) Solve measures how often a solver completes the task from identical cloned copies of those states. Crossing supervised fine-tuning (SFT) and RL checkpoints in both roles across two benchmarks and two independently released training pipelines, we find that the gain from switching the solver from SFT to RL is consistently larger when RL is the reacher, at all three model scales on TravelPlanner and on both ALFWorld splits. Further analyses on ALFWorld show that RL improves both Reach and Solve. The solver gain is larger under an RL reacher because RL reaches solvable states more often, and separately measured Reach and Solve gaps recover most of this difference. Because handoff only requires replaying one checkpoint's history under another, agentic RL evaluations can report arrival and completion alongside final success.

cs.AI↗
arXiv2026

Learning from the Gap Between Pass@K and Pass@1

Sampling many responses and keeping one that passes a verifier lets large language models solve problems beyond their single-response ability, but this search must be paid again for every query, while many deployments answer with a single response. Post-training on verified responses can transfer the benefit of search into the model. With a fixed budget, selecting by correctness alone spends slots on problems the model already answers correctly, leaving fewer to correct its failures. To address this imbalance, we propose GapFT, which trains on the gap between Pass@K and Pass@1: problems that the source model fails with one response but solves within K samples. GapFT keeps the objective and training budget fixed and changes only which verified responses enter training; an exact decomposition splits the resulting Pass@1 change into corrected failures and regressions on problems the source model already solved. On LogiQA 2.0 and ReClor with three model families, GapFT is above budget-matched uniform rejection-sampling fine-tuning (RFT) in every setting, with a positive pooled effect, and on Llama-3.1-8B and Mistral-7B it recovers about two thirds to four fifths of the gain of fine-tuning on the entire verified pool with 11-34% of its problems. Further analyses reveal that the gain comes from failures that the first few search samples recover, while failures found only by deeper search displace replay and add no net gain, that filling the same budget with gold-labeled failures search cannot reach lowers accuracy, and that the gain is bounded by how many transferable failures search exposes.

cs.LG↗
arXiv2026

Structural limit of volatility target under stochastic volatility and discrete corrections

The exact limiting distribution of a volatility target index has recently been established under a log-normal assumption on the risky asset. In this paper, we show that the structural limit remains valid under a broad class of stochastic volatility processes. In addition, we show that, when the limit of the joint distribution of the volatility target index and the driving Brownian motions is considered, an extra independent Brownian motion must be involved. Besides the structural limit theorem, we derive a correction formula for the expected quadratic variation of the discrete rebalancing volatility target index, under the assumption that the volatility of the underlying risky asset is a deterministic function of time. As a consequence of the discrete correction formula, it is shown that, as the rebalancing time step $Δt\to 0$ and the observation window parameter $λ\to 1$ simultaneously along a path, the expected quadratic variation converges to the target variance if and only if $(1-λ)^{-1}Δt \to 0$ along the same path. Numerical results are provided to support the structural limit, using the local volatility model and the Heston stochastic volatility model as examples, as well as the effectiveness of the discrete correction formula.

math.PR↗
arXiv2026

Validated Behavioral Hypotheses as a Lens for Evaluating Participant Simulation

We propose using validated behavioral hypotheses as a lens for evaluating LLM agents as simulated human participants. This approach makes behavioral agreement in participant simulation measurable and decomposable, revealing which human effects agents reproduce, where they diverge, and how agent design changes that agreement. To operationalize this idea, we build HumanStudy-Bench, an open benchmarking platform that reconstructs experimental protocols from published human studies and administers these protocols to agents serving as silicon participants. The benchmark compares population-level effects derived from agent responses with the corresponding published human findings using two metrics: the Probability Alignment Score (PAS) for inferential agreement and the Effect Consistency Score (ECS) for effect-magnitude agreement. We apply HumanStudy-Bench across 12 human studies, evaluating 10 models under four agent designs, with 6,588 simulated participants per agent configuration. By making published human studies and their validated behavioral hypotheses reusable for evaluating simulated participants, we hope to make the behavioral assumptions underlying LLM-based social simulations and their sensitivity to agent design explicit and empirically testable.

cs.CY↗
arXiv2026

GeoRefer-Bench: A Benchmark from Referring Pixels to Verifiable Geospatial Reasoning

Referring segmentation in overhead imagery is inherently relational: a query may ask for the buildings north of the road or the pond closest to a residential area, so the correct referent can contain one object, several objects, or none. Existing benchmarks mainly score mask overlap, which cannot verify whether a model actually resolved the stated spatial relation. We introduce GeoRefer-Bench, a benchmark for verifiable geospatial referring segmentation. Each query is represented by an executable logical form over a metric scene graph, and predictions are evaluated with Exact Query Success (EQS), which is satisfied only when the returned instance set exactly matches the set denoted by the query. GeoRefer-Bench contains 700 whole 2048x2048 UAV scenes (2.94 Gpx) at 12.5 and 25 cm ground sampling distance, 26,217 instances, 142,796 spatial relations, and 20,916 executable queries spanning five reasoning levels. It further includes three paraphrases per query, 24.0% unanswerable queries, 2,477 counterfactual pairs, and five leakage-controlled evaluation splits. An independent audit re-derives object geometry, mask ownership, relation values, query execution, and split provenance, finding zero issues across all 700 scenes. Relation-blind strategies can retain non-trivial mIoU while achieving at most 22.7 EQS overall, showing that overlap alone does not certify relational grounding. Across fifteen current models, the strongest reaches 74.1 EQS but drops from 98.9 at level 1 to 60.5 at level 5, while ten models score below 5 EQS on two-hop queries. GeoRefer-Bench turns geospatial referring segmentation from mask matching into verifiable reference resolution.

cs.CV↗
arXiv2026

Beyond a Scalar: Distributional Serving Interfaces for Watch-Time Prediction

Watch time is the primary engagement signal in short video feeds, and its prediction directly affects ranking and exposure. Existing methods improve watch time prediction by correcting duration bias or modeling richer distributions, but most expose only an expected or debiased watch time at serving time. Even when video duration is available to later models, the interface gives only one estimate of watch time and no probabilities for completion, overplay, or other regions relevant to downstream tasks. To address this limitation, we propose the Distributional Serving Interface (DSI), which has a distribution provider, a compact, low-dimensional summary, and lightweight readouts tailored to each task. The provider learns a joint distribution over four watch states derived from watch ratio and their event times; rules based on video duration remove incompatible combinations, while a restoration loss preserves accuracy in seconds. The summary reduces this distribution to a small set of event probabilities, time scales relative to duration, and uncertainty statistics. After training the provider, we fix its parameters and train value and ranking readouts that combine the summary with raw context. Across KuaiRec, KuaiRand-1K, and WeChat21, the complete DSI system achieves the lowest MAE on all three datasets, beating the strongest result among nine baselines by 1.9% to 8.5%, and achieves the best XAUC on two. It also leads retrieval metrics that account for video duration when complete systems are compared. With matched readouts held constant, the summary retains information relevant to each task beyond a predicted mean paired with video duration. Using the same lightweight linear heads for each new target, it also performs best on two new watch-time targets and improves a separately logged engagement target, while a randomly initialized provider does not reproduce this gain.

cs.IR↗
arXiv2026

Exploring Causal Mechanisms with Generative Agent-Based Models

In this paper, we explore using generative agent-based models for a classical ABM application: testing how individual-level behavioral rules produce collective phenomena. We introduce RePair, a method that calibrates simulation worlds, operationalizes candidate mechanisms as natural-language rules, estimates their effects through matched interventions, and examines behavioral traces. We assess the method by testing it in four simulation worlds grounded in established social-science models and empirical studies. Our results reveal that (1) natural-language rules can produce measurable collective effects; (2) rule comparisons can converge as configurations accumulate; and (3) behavioral traces connect collective effects to agents' actions and interactions, helping researchers evaluate the proposed causal process. Together, these findings show the feasibility of using generative agent-based models to explore causal mechanisms and provide practical guidance for producing reliable, interpretable explanations.

cs.MA↗
arXiv2026

Depth-resolved magnetization dynamics in Fe thin films after ultrafast laser excitation

We performed time-resolved x-ray resonant magnetic reflectivity measurements on a laser-excited ferromagnetic Fe thin film to simultaneously probe the transient magnetic and structural depth profiles with nanometer spatial and femtosecond temporal resolution. Our results show that during the first picoseconds after optical excitation, the magnetization of the Fe layer is strongly inhomogeneous, especially in the vicinity of the buried interface. By comparing our experimental results to predictions based on the microscopic three-temperature model and simulations of laser-induced spin-currents, we demonstrate that local and non-local angular momentum transfer phenomena take place simultaneously. After a few picoseconds, the magnetization relaxes back to equilibrium while the total thin film thickness starts oscillating periodically, with a maximum dilation of approximately 1.3% of the entire thin film thickness due to laser-induced stresses.

cond-mat.mtrl-sci↗