Searcharxiv⌕ Search

arXiv subjects

Xin Xie

Publications and source records attributed to Xin Xie.

At least 19 recordsLinked to original sources
arXiv2026

Reversibility of Nonequilibrium Dynamics in a Unitary Fermi Gas

We theoretically describe and experimentally observe the reversibility of a nonequilibrium state in a unitary Fermi gas confined in a time-dependent trap. Unlike the equilibrium case, reversibility out of equilibrium imposes a stringent condition: the accumulated dynamical phase must be an integer multiple of $2π$, i.e., $Δϕ= 2ω_{0} s(T) = 2nπ$, where $ω_{0}$ is the initial trapping frequency and $s(T)$ is the rescaled time after a total evolution time $T$. We demonstrate the reversibility of an initial nonequilibrium state in an isotropic unitary Fermi gas by driving the optical trap through a round trip. We find two types of reversible pathways, the space-time scaling-invariant and polynomial shortcut-to-adiabaticity (STA) protocols. The phase condition restricts the round-trip time $T$ to discrete values, a distinguishing feature of nonequilibrium dynamics. Our work establishes a clear criterion for the quantum control of nonequilibrium many-body systems beyond linear response and provides practical pathways to reversibility.

cond-mat.quant-gas↗
arXiv2026

WeGR: Towards Unified Generative Retrieval via Behavior-aware Tokenization and Preference Alignment

Generative retrieval (GR) offers a promising paradigm for industrial search by unifying candidate generation within an autoregressive framework. However, effectively leveraging user behaviors in industrial GR remains challenging: existing SID tokenization methods struggle to preserve collaborative signals while maintaining balanced codebook utilization, whereas reinforcement learning (RL) for prioritizing high-value candidates suffers from sparse rewards when generated SIDs fail to match positive items. To address these challenges, we propose WeGR, a unified GR framework with two complementary components. Specifically, the Collaborative and Collapse-free Tokenizer (CoCo-VAE) injects query-to-item and item-to-item collaborative signals into continuous representations before quantization, preserving behavioral information without disrupting codebook optimization. It further couples first-layer codebook refresh to reactivate underutilized codewords with fidelity anti-collapse regularization, which broadens gradient coverage over relevant codewords and penalizes collisions in deeper layers to yield behavior-aware SIDs with balanced utilization. For preference alignment, Dynamic Horizon Policy Optimization (DHPO) progressively expands reward evaluation and policy optimization horizons to alleviate reward sparsity, enabling coarse-to-fine preference learning from SID prefixes to complete items without auxiliary reward models. Extensive offline experiments validate that WeGR improves retrieval and ranking performance. Moreover, a six-week online A/B test on a large-scale e-commerce search platform achieves relative gains of 2.539% in orders, 1.508% in CTR, and 1.381% in CVR, while reducing CPU and GPU usage by 25.5% and 26.1%. These results highlight the effectiveness and efficiency of WeGR in industrial deployment, supporting multiple search scenarios and tens of millions of daily page views.

cs.IR↗
arXiv2026

CoWear: Towards Collaborative Wearable Inertial Localization with Multiple Smart Devices

Consumer smart devices observe the same human motion from different body locations, yet existing inertial-localization benchmarks rarely provide synchronized heterogeneous wearables with device-specific motion ground truth. We present CoWear, a dataset and reference framework for collaborative inertial localization using a smartphone, smartwatch, and smart glasses. The dataset contains over 30 participants, over 300 walking sessions, and more than 30 hours and 100 km of aggregate device trajectories, with each device individually tracked by motion capture. To enable collaboration across distinct body motions, we train role-specific estimators to predict the displacement of a common reference device over synchronized temporal windows. Their predictions are transformed to the world frame using gyroscope integration and fused according to learned uncertainty. Experiments using phone, glasses, and watch as reference targets show that multi-device fusion consistently improves both displacement and trajectory rollout accuracy over individual-device estimation. CoWear provides a dataset and reproducible baseline for studying collaborative inertial localization across heterogeneous consumer wearables.

cs.RO↗
arXiv2026

Polarforming for MIMO Covert Communications

Covert communication conceals wireless transmission activity, but conventional multi-antenna designs rely mainly on spatial beamforming and can be constrained by limited spatial degrees of freedom. This paper investigates polarization-reconfigurable antenna (PRA)-aided multiple-input multiple-output (MIMO) covert communication, where Alice and Bob perform transmit and receive polarforming, respectively, and Willie uses a vertically polarized antenna for radiometric detection. We jointly optimize the transmit covariance matrix and the transmit and receive phase shift vectors to maximize Bob's achievable rate under the covertness and transmit power constraints. Under Gaussian signaling, perfect polarized channel state information, and known noise power at Willie, we derive his optimal radiometric threshold and exact minimum detection error probability (DEP) for a finite observation length, and convert the DEP requirement into a deterministic constraint on his received signal power. We further prove that, for a fixed transmission design, positive signal leakage becomes detectable as the observation length tends to infinity. To solve the design problem for a finite observation length, we develop an alternating optimization (AO) algorithm with a generalized water-filling covariance update and closed-form phase updates. Numerical results validate the exact DEP analysis and the rapid convergence of the proposed algorithm. Compared with the sufficient divergence-based condition, the exact covertness constraint permits a $0.99$--$1.12$~dB higher received signal-to-noise ratio (SNR) at Willie, while joint transmit and receive polarforming provides a rate gain of up to $53.8\%$ over the considered benchmark schemes.

cs.IT↗
arXiv2026

Exploiting UAV Attitude for Covert Communications: Dynamics-Consistent Trajectory-Attitude Co-Design

Existing trajectory designs for covert communications with uncrewed aerial vehicles (UAVs) typically rely on point-mass kinematics, which overlook the inherent coupling between flight maneuvers and antenna orientation. Motivated by recent studies linking UAV acceleration to attitude, we investigate dynamics-consistent trajectory and attitude co-design for rotary-wing UAV covert communications. We consider a UAV equipped with a directional antenna transmitting confidential information to a legitimate ground receiver in the presence of a warden. The acceleration-induced reduced attitude of the UAV is explicitly modeled together with its impact on the attitude-dependent directional channel gain. To ensure covertness, we derive a per-slot constraint based on the Kullback--Leibler (KL) divergence at the warden. We then formulate a joint trajectory and attitude optimization problem to maximize the average achievable covert rate subject to covertness, motion, thrust-magnitude, attitude-smoothness, and acceleration-domain safety constraints induced by roll and pitch limits. The resulting problem is non-convex because the channel gains depend jointly on position, acceleration, and attitude. To solve it, we develop a minorization--maximization (MM)-based algorithm that constructs a concave lower bound on the transmission rate and a convex upper bound on the covert constraint at each iteration. We further extend the framework to imperfect attitude control and develop a robust design against attitude tracking errors caused by actuator limitations, sensor noise, and control delays. Numerical results demonstrate the effectiveness of the proposed dynamics-consistent design and highlight the benefit of jointly optimizing UAV trajectory and attitude for covert communications.

cs.IT↗
arXiv2026

Topological Fermi-polaron Polaritons with Electrically Invertible Topology

Topological polaritons provide a route to engineer optical Chern bands, but realizing large gaps or efficient electrical tunability remains challenging because time-reversal-symmetry breaking in semiconductor polariton systems is typically limited by weak Zeeman splittings. Here, we demonstrate a topological Bose-Fermi mixture --- Fermi-polaron polaritons ---where large gaps and electrically invertible topology emerge from the interplay of Fermi-polaron many-body interactions, photonic symmetry, and strong light--matter coupling. Using a gate-controlled monolayer transition-metal dichalcogenide integrated with a symmetry-engineered photonic crystal, we create Fermi-polaron polaritons with topological band structures. The direct coupling between band topology and gate-tunable many-body interactions provides a mechanism for amplified time-reversal-symmetry breaking beyond bare Zeeman splitting and enables electrical inversion of topological gaps, Berry curvature, and Chern numbers. These results establish an electrically reconfigurable platform for interaction-driven topological states and programmable topological photonics.

physics.optics↗
arXiv2026

Label-Efficient Generative Inverse Design and Design-Rule Discovery in Freeform Topological Photonics

Topological photonic crystals support robust, disorder-resilient light transport protected by their band-structure topology, yet their design remains confined to a small set of symmetry-defined templates. The surrounding freeform design space, where optimal performance and new functionality may reside, is difficult to access because of its high dimensionality and costly full-wave simulation needed for evaluating each candidate. % Here we report generative inverse design of freeform valley photonic crystals. A label-free diffusion prior learns the distribution of three-fold rotationally symmetric geometries from 7,689 unlabeled designs that cost 0.15~s each to produce, while a convolutional surrogate is trained on 1,666 full-wave band-gap simulations costing 4.28~min each, exploiting a 1,700-fold cost asymmetry. Across nine target band gaps up to $225~$meV, we generate and verify by full-wave simulation 270 valley photonic crystal designs, reaching a mean absolute error of $1.4$--$6.7$~meV within the labeled range and retaining below $5.1\%$ fractional error at targets $25\%$ beyond its upper bound, while a label-conditioned diffusion model trained on the same label data performs substantially worse. Beyond inverse design, the trained model functions as an instrument for discovering structure--property relations. At a target band gap where the labeled set contains only five designs, it generates hundreds, resolving geometric trends that are statistically inaccessible in the training data and yielding interpretable rules for large-gap valley photonic crystals. Our results establish surrogate-guided diffusion as a label-efficient route to freeform topological-photonic design, and as a means of extracting design principles where physical intuition and simulation labels are both scarce.

physics.optics↗
arXiv2026

Test-Time Weak-to-Strong Alignment: Transferring Implicit Rewards from Weak to Strong Flow Models

Aligning a text-to-image generation flow model with a reward makes it follow objectives that the training data alone does not provide. Alignment fine-tuning delivers this by reinforcement learning (RL) or preference optimization, but it must be repeated for every checkpoint and returns a model fixed at the reward and strength it was trained with. Test-time alignment instead steers a frozen model during sampling, allowing task-specific and sample-specific guidance. Existing methods obtain this only by drawing the per-step signal from the reward function itself, through its gradient, or through a separately trained value function. We propose changing the supervision source: let a pair of weak models, not a reward function, supply the supervision. A source aligned model, kept together with its base as a source alignment pair, stores its training reward as an implicit, step-wise, KL-anchored signal expressed in the sampler's own coordinates. We explore whether this model-form supervision can cross scale, and show that it does: our method, AlignGraft, aligns a larger, frozen, never-tuned model by adding the pair's velocity difference during sampling. The transport is exact under a shared noising kernel and needs neither the reward nor its gradient at test time. The method has no schedules, only a single scalar that controls the alignment strength and can extrapolate it beyond that of the source alignment pair. Across image and video flow models (Stable Diffusion 3.5, FLUX, and Wan), the transfer lifts the frozen large model on preference, compositional, and text-rendering rewards, can exceed the source aligned model itself, and preserves the large model's fidelity at a small constant sampling overhead. Extensive experiments show that one alignment run on a weak model produces supervision that the whole model family can reuse at test time.

cs.CV↗
arXiv2026

SkillNet: Create, Evaluate, and Connect AI Skills

Current AI agents can flexibly invoke tools and execute complex tasks, yet their long-term advancement is hindered by the lack of systematic accumulation and transfer of skills. Without a unified mechanism for skill consolidation, agents frequently ``reinvent the wheel'', rediscovering solutions in isolated contexts without leveraging prior strategies. To address this challenge, we introduce SkillNet, an open infrastructure for creating, evaluating, and organizing AI skills at scale. SkillNet structures skills within a unified ontology that supports creating skills from heterogeneous sources, establishing rich relational connections, and performing multi-dimensional evaluation across Safety, Completeness, Executability, Maintainability, and Cost-awareness. Our infrastructure integrates a repository of over 600,000 skills, an interactive platform, and a versatile Python toolkit. Experiments on ALFWorld, WebShop, and ScienceWorld show 40% higher average rewards and 30% fewer execution steps across multiple backbone models. Furthermore, SkillNet-Gym benchmarks skill retrieval, utilization, and composition, while SkillNet-Fabric enables task-specific skill routing through lightweight Wikis. By formalizing skills as evolving, composable assets, SkillNet provides a robust foundation for agents to move from transient experience to durable mastery.

cs.AI↗
arXiv2026

When Distillation Breaks Motion Control: Restoring Generative Trajectories for Fast Video Generators

Training-free motion customization imposes motion patterns from reference videos onto video generators through test-time computation. Most existing methods target full diffusion models, requiring many denoising steps and high computational cost. With the rise of efficient distilled models, a natural question arises: can test-time motion customization be applied directly to distilled generators with their accelerated sampling and efficiency gains? However, our analysis reveals that existing training-free techniques fail on distilled models. Distillation fundamentally alters the denoising dynamics that prior test-time guidance relies on, and the large denoising steps of distilled generators discard the dense intermediate states that score guidance requires, rendering existing motion control strategies incompatible with fast generation. To address this limitation, we propose MotionEcho, a novel training-free test-time distillation framework that enables motion customization for distilled video generators. The key idea is to correct the student model's sampling trajectory with restricted usage of a high-quality diffusion teacher at inference time. Teacher supervises the student's denoising by re-noising the student's endpoint onto its dense trajectory to form a motion-aligned clean endpoint, then interpolating it with the student's, while an adaptive scheduling mechanism determines when and how much teacher guidance is needed. As a result, MotionEcho restores generative trajectories for distilled video generators via lightweight, adaptive test-time teacher guidance, enabling accurate motion control without compromising generation efficiency. Extensive experiments on multiple distilled video generation models demonstrate that our method significantly improves motion fidelity and visual quality while retaining the efficiency advantages of distilled generation.

cs.CV↗
arXiv2026

SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories

Large language model (LLM) agents increasingly rely on reusable external skills to solve long-horizon interactive tasks. Existing training-free skill adaptation pipelines usually update skills from full trajectories or session-level feedback, which makes failure attribution coarse and often produces unstable or overly broad revisions. We propose SkillAdaptor, a training-free step-level skill adaptation framework with explicit failure attribution, and it can plug into OpenClaw-class agent harnesses. Given a failed trajectory, SkillAdaptor identifies a first actionable fault step, links responsibility to candidate skills, and applies targeted updates under explicit acceptance checks while keeping the backbone frozen. We evaluate on WebShop, PinchBench, and Claw-Eval with Kimi-K2.5, GLM-5, and GPT-5.2. SkillAdaptor improves over no-skill and skill-adaptation baselines on all three suites, with the largest single-metric improvements of +1.5 points on PinchBench Avg Score%, +1.8 on Claw-Eval Avg Score, and +1.7 on WebShop success rate. These results indicate that step-level attribution supports more stable and auditable training-free skill maintenance\footnote{The code will be released at https://github.com/zjunlp/SkillAdaptor.}.

cs.CL↗
arXiv2026

4DThinker: Thinking with 4D Imagery for Dynamic Spatial Understanding

Dynamic spatial reasoning from monocular video is essential for bridging visual intelligence and the physical world, yet remains challenging for vision-language models (VLMs). Prior approaches either verbalize spatial-temporal reasoning entirely as text, which is inherently verbose and imprecise for complex dynamics, or rely on external geometric modules that increase inference complexity without fostering intrinsic model capability. In this paper, we present 4DThinker, the first framework that enables VLMs to "think with 4D" through dynamic latent mental imagery, i.e., internally simulating how scenes evolve within the continuous hidden space. Specifically, we first introduce a scalable, annotation-free data generation pipeline that synthesizes 4D reasoning data from raw videos. We then propose Dynamic-Imagery Fine-Tuning (DIFT), which jointly supervises textual tokens and 4D latents to ground the model in dynamic visual semantics. Building on this, 4D Reinforcement Learning (4DRL) further tackles complex reasoning tasks via outcome-based rewards, restricting policy gradients to text tokens to ensure stable optimization. Extensive experiments across multiple dynamic spatial reasoning benchmarks demonstrate that 4DThinker consistently outperforms strong baselines and offers a new perspective toward 4D reasoning in VLMs. Our code is available at https://github.com/zhangquanchen/4DThinker.

cs.CV↗
arXiv2026

SkillX: Automatically Constructing Skill Knowledge Bases for Agents

Learning from experience is critical for building capable large language model (LLM) agents, yet prevailing self-evolving paradigms remain inefficient: agents learn in isolation, repeatedly rediscover similar behaviors from limited experience, resulting in redundant exploration and poor generalization. To address this problem, we propose SkillX, a fully automated framework for constructing a \textbf{plug-and-play skill knowledge base} that can be reused across agents and environments. SkillX operates through a fully automated pipeline built on three synergistic innovations: \textit{(i) Multi-Level Skills Design}, which distills raw trajectories into three-tiered hierarchy of strategic plans, functional skills, and atomic skills; \textit{(ii) Iterative Skills Refinement}, which automatically revises skills based on execution feedback to continuously improve library quality; and \textit{(iii) Exploratory Skills Expansion}, which proactively generates and validates novel skills to expand coverage beyond seed training data. Using a strong backbone agent (GLM-4.6), we automatically build a reusable skill library and evaluate its transferability on challenging long-horizon, user-interactive benchmarks, including AppWorld, BFCL-v3, and $τ^2$-Bench. Experiments show that SkillKB consistently improves task success and execution efficiency when plugged into weaker base agents, highlighting the importance of structured, hierarchical experience representations for generalizable agent learning. Our code will be publicly available soon at https://github.com/zjunlp/SkillX.

cs.CL↗
arXiv2026

SGA-MCTS: Decoupling Planning from Execution via Training-Free Atomic Experience Retrieval

LLM-powered systems require complex multi-step decision-making abilities to solve real-world tasks, yet current planning approaches face a trade-off between the high latency of inference-time search and the limited generalization of supervised fine-tuning. To address this limitation, we introduce \textbf{SGA-MCTS}, a framework that casts LLM planning as non-parametric retrieval. Offline, we leverage Monte Carlo Tree Search (MCTS) to explore the solution space and distill high-fidelity trajectories into State-Goal-Action (SGA) atoms. These atoms are de-lexicalized primitives that abstract concrete entities into symbolic slots, preserving reusable causal logic while discarding domain-specific noise. Online, a retrieval-augmented agent employs a hybrid symbolic-semantic mechanism to fetch relevant SGAs and re-ground them into the current context as soft reasoning hints. Empirical results on complex benchmarks demonstrate that this paradigm enables frozen, open-weights models to match the performance of SOTA systems (e.g., GPT-5) without task-specific fine-tuning. By effectively amortizing the heavy computational cost of search, SGA-MCTS achieves System 2 reasoning depth at System 1 inference speeds, rendering autonomous planning both scalable and real-time feasible.

cs.AI↗
arXiv2026

HyperAlign: Hypernetwork for Efficient Test-Time Alignment of Diffusion Models

Diffusion model alignment aims to bridge the gap between generated outputs and human preferences by enhancing both semantic consistency with textual prompts and overall visual quality. Existing alignment methods face a challenging trade-off: test-time approaches enable input-specific adaptability but introduce significant computational overhead and tend to under-optimize, while fine-tuning approaches risk reward over-optimization and loss of generation diversity. To bridge this gap, we propose HyperAlign, a framework that trains a hypernetwork for efficient and effective test-time alignment. Instead of modifying latent states directly, HyperAlign dynamically generates input-and-state-conditioned low-rank adaptation weights to modulate the denoising trajectory toward target rewards. We introduce multiple HyperAlign variants of varying granularity to balance alignment quality and computational efficiency. The hypernetwork is optimized with a reward objective regularized by preference data to mitigate reward hacking. We evaluate HyperAlign across multiple generative paradigms, including Stable Diffusion and FLUX, where it significantly outperforms existing alignment methods in semantic consistency and visual quality.

cs.CV↗
arXiv2026

CATNet: Collaborative Alignment and Transformation Network for Cooperative Perception

Cooperative perception significantly enhances scene understanding by integrating complementary information from diverse agents. However, existing research often overlooks critical challenges inherent in real-world multi-source data integration, specifically high temporal latency and multi-source noise. To address these practical limitations, we propose Collaborative Alignment and Transformation Network (CATNet), an adaptive compensation framework that resolves temporal latency and noise interference in multi-agent systems. Our key innovations can be summarized in three aspects. First, we introduce a Spatio-Temporal Recurrent Synchronization (STSync) that aligns asynchronous feature streams via adjacent-frame differential modeling, establishing a temporal-spatially unified representation space. Second, we design a Dual-Branch Wavelet Enhanced Denoiser (WTDen) that suppresses global noise and reconstructs localized feature distortions within aligned representations. Third, we construct an Adaptive Feature Selector (AdpSel) that dynamically focuses on critical perceptual features for robust fusion. Extensive experiments on multiple datasets demonstrate that CATNet consistently outperforms existing methods under complex traffic conditions, proving its superior robustness and adaptability.

cs.CV↗
arXiv2025

EPARA: Parallelizing Categorized AI Inference in Edge Clouds

With the increasing adoption of AI applications such as large language models and computer vision AI, the computational demands on AI inference systems are continuously rising, making the enhancement of task processing capacity using existing hardware a primary objective in edge clouds. We propose EPARA, an end-to-end AI parallel inference framework in edge, aimed at enhancing the edge AI serving capability. Our key idea is to categorize tasks based on their sensitivity to latency/frequency and requirement for GPU resources, thereby achieving both request-level and service-level task-resource allocation. EPARA consists of three core components: 1) a task-categorized parallelism allocator that decides the parallel mode of each task, 2) a distributed request handler that performs the calculation for the specific request, and 3) a state-aware scheduler that periodically updates service placement in edge clouds. We implement a EPARA prototype and conduct a case study on the EPARA operation for LLMs and segmentation tasks. Evaluation through testbed experiments involving edge servers, embedded devices, and microcomputers shows that EPARA achieves up to 2.1$\times$ higher goodput in production workloads compared to prior frameworks, while adapting to various edge AI inference tasks.

cs.DC↗
arXiv2025

Energy Dynamics of a Nonequilibrium Unitary Fermi Gas

We investigate the energy dynamics of a unitary Fermi gas driven away from equilibrium. The energy is injected into the system by periodically modulating the trapping potential of a spherical unitary Fermi gas, and due to the existence of SO(2,1) symmetry, the breathing mode is excited without dissipation. Through the long-lived breathing oscillation, we precisely measure the energy evolution of the nonequilibrium system during the trap modulation. We find the trapping potential and internal energies increase with modulation time and simultaneously oscillate nearly $\textrm{180}^{\textrm{o}}$ out of phase. At large modulation amplitudes, the energy-injection efficiency is strongly reduced due to the trap anharmonicity. Unlike the equilibrium system, the measured energy evolution agrees well with predictions of the dynamic virial theorem. Our work provides valuable insights into the energy injection and redistribution in a non-equilibrium system, paving a way for future investigations of nonequilibrium thermodynamics.

cond-mat.quant-gas↗