Search arXiv⌕ Search

arXiv subjects

Lei Wang

Publications and source records attributed to Lei Wang.

At least 73 records · Page 4Linked to original sources
arXiv2026

Policy Targeting with Binary Classification Trees: an Application to Rural Hospital Closures

Empirical researchers often use binary classification trees to identify subgroups at risk of adverse outcomes. We compare two classification tree algorithms in this policy-targeting context: classical classification and regression trees (CART) and the maximizing-distance final-split approach (MDFS). We establish a theoretical setting in which MDFS identifies a high-risk subgroup while CART identifies none. Applied to rural hospital closure data, MDFS targets a high-closure-probability subgroup: for-profit hospitals with more than 63 days in accounts receivable. CART misses this subgroup. To facilitate further application of these classification tree algorithms, we provide \texttt{targetree} package in Python, R, and Stata.

econ.GN↗
arXiv2026

DeepStack: Facilitating Co-Design Exploration of 3D DRAM-Stacked Accelerators for Distributed LLM Inference

Advances in hybrid bonding and packaging have driven growing interest in 3D DRAM-stacked AI accelerators. As large language models (LLMs) scale to hundreds of billions or trillions of parameters, distributed inference across multiple 3D chips has become essential for AI serving. This trend makes cross-stack co-design critical because system-level parallelization and scheduling choices are tightly coupled with hardware characteristics such as memory organization, interconnects, and thermal constraints. We present DeepStack, an accurate performance model and efficient design space exploration (DSE) framework for distributed 3D-stacked LLM inference. At the hardware level, DeepStack captures transaction-aware memory bandwidth, bank activation constraints, buffering limitations, and thermal and power behavior. At the system level, it incorporates comprehensive parallelization strategies and execution scheduling. Through a dual-stage network abstraction and tile-level compute-communication overlap modeling, DeepStack achieves up to 100,000x faster evaluation than state-of-the-art simulators at comparable accuracy. We cross-validate DeepStack against our in-house 3D designs, an NS-3 backend with 2.12% error, and vLLM serving on eight B200 GPUs with 12.92% error. Combined with hierarchical search, DeepStack efficiently explores about 2.5 x 10^14 design points spanning the number of stacked DRAM layers, DRAM vertical connectivity, interconnects, compute-memory allocation, and distributed scheduling under thermal and area constraints. A search-space ablation shows that restricted DSE baselines can miss up to 9.5x modeled throughput. Beyond modeling and DSE, DeepStack derives design implications for distributed 3D AI systems and guides performance optimization across the stack. Source code and artifacts are available at https://github.com/tile-ai/DeepStack/tree/ae.

cs.AR↗
arXiv2026

Toward Interpretable Multimodal Fusion: Heat Conduction Modeling for Hyperspectral and LiDAR Joint Classification

The fusion of hyperspectral (HS) and Light Detection and Ranging (LiDAR) data plays a crucial role in enhancing land-cover classification by jointly exploiting spectral, spatial, and structural cues. However, existing multimodal fusion methods still struggle to model long-range dependencies and complex anisotropic interactions while maintaining computational efficiency. This paper introduces M2Heat, a physics-inspired framework that investigates multimodal fusion through the lens of heat conduction. At its core, a physics-driven visual heat conduction module (vHeat) and enhanced Frequency Value Embeddings (FVEs) simulate anisotropic information flow, enabling the capture of global dependencies with sub-quadratic complexity and physical interpretability. This mechanism, combined with a hybrid spatial-frequency fusion strategy named Cross-Frequency Fusion (CFF) module, produces highly discriminative and robust feature representations. M2Heat achieves competitive overall performance on three benchmarks, i.e., Trento, Houston2013, and Augsburg, while providing an interpretable heat-conduction-guided perspective for multimodal feature fusion. These results indicate the potential of heat-conduction-guided neural operators for efficient and interpretable RS multimodal fusion. The source code is publicly available at https: /github.com/Weikan0425/M2Heat_HSI_LiDAR.

cs.CV↗
arXiv2026

A unified quantum electrical platform for synchronous metrological realization of volt, ohm and ampere

A co-located integration quantum electrical standard is essential to reduce reliance on distributed traceability in high-accuracy metrology, especially for portable and on-site use. Metrologically, realizing any two of voltage, resistance, and current is sufficient, as the third follows from Ohm's law. The combination of Josephson voltage and quantum Hall resistance offers better uncertainty, but conflicts with the tesla-level field for quantum Hall and near-zero field for Josephson operation. Here we report a compact unified platform enabling co-realization of quantum voltage and resistance in a single cryostat near 4 K, with quantum current derived via Ohm's law. A hierarchical magnetic shielding with staged attenuation and spatial confinement allows 6 T and below 50 nT to coexist within 270 mm axial separation with negligible cross-coupling. In integrated operation, the Josephson and quantum Hall subsystems agree with expected quantized values within relative standard uncertainties of 2.6E-9 and 1.4E-8, respectively. Linking them via an improved cryogenic current comparator realizes a 50 μA quantum current with relative uncertainty of 6.6E-8. These results demonstrate that three basic electrical units can be synchronously realized with superior metrological consistency on a single integrated platform, offering a viable transition from distributed calibration chains toward compact-integrated quantum-based realization.

physics.ins-det↗
arXiv2026

Search for neutrinoless quadruple beta decay of $^{136}$Xe in PandaX-4T detector

The observation of neutrinoless quadruple beta decay (0$ν$4$β$) in the absence of neutrinoless double beta decay (0$ν$2$β$) has been argued to provide a strong indication that neutrinos are Dirac particles. We report a search for 0$ν$4$β$ decay of $^{136}\text{Xe}$ using a total $^{136}\text{Xe}$ exposure of 148.4 kg$\cdot$yr, collected during the commissioning and the first science runs of the PandaX-4T experiment. No significant excess of events over the background is observed. A lower limit on the 0$ν$4$β$ decay half-life of $^{136}\text{Xe}$ is set at 6.01 x $10^{24}$ yr at the 90% confidence level. This result establishes the most stringent constraint on this process in xenon, demonstrating the unique capability of the PandaX-4T detector in probing lepton number violation and shedding light on the fundamental nature of neutrinos.

nucl-ex↗
arXiv2026

MotionScape: A Motion-Stratified UAV Video Benchmark for World Modeling and Future Video Generation

Unmanned aerial vehicles (UAVs) are increasingly crucial for low-altitude autonomy and complex environment understanding. World models enable UAVs to anticipate how future states may evolve under potential actions, providing predictive support for autonomous decision-making. For video world models, future video generation serves as a means of simulating future visual states, providing a direct basis for evaluating their predictive capability. However, existing UAV video resources typically focus on specific control signals or simulated environments. Standardized benchmarks for multi-condition UAV-view future video generation are still limited. To bridge this gap, we introduce MotionScape, a real-world UAV-view benchmark comprising 228 high-resolution video clips totaling 62,700 frames, with semantic annotations of weather and illumination conditions, scene environment, and camera-viewpoint motion. The clips are stratified into low-, medium-, and high-motion strata using optical-flow-based motion intensity. MotionScape supports Text2World, Image2World, and Video2World evaluation under a unified future-generation protocol. Technical validation with representative baseline models assesses performance across conditioning settings and motion strata, providing a unified platform for evaluating the future visual simulation capabilities of UAV world models.

cs.CV↗
arXiv2026

SmartANN: Object Causal Modeling Boosts Approximate Nearest Neighbor Diagnosis and Auto-Design

Approximate Nearest Neighbor (ANN) algorithms achieve high efficiency through interdependent phases across index construction and query execution. This coupling allows upstream performance loss to propagate downstream, affecting execution behavior and measurable outputs. Existing component-level analyses mainly compare isolated design choices, while end-to-end benchmarks report aggregate metrics; neither traces loss propagation across dependent phases, hindering root-cause attribution and automated redesign. We present SmartANN, a framework based on the object causal model (OCM) for ANN bottleneck attribution and automated redesign. SmartANN represents an ANN workflow as eight ordered, replaceable objects and diagnoses them with a sequential diagnose-and-replace loop. At each iteration, it identifies the first object deviating from expected behavior or output as a bottleneck. Because an upstream bottleneck can obscure downstream ones, SmartANN replaces it with a test oracle when available, or with an implementation producing a better outcome, then continues downstream diagnosis. From the diagnosed bottlenecks and failure causes, SmartANN composes compatible actions from a pluggable action library to generate an optimized end-to-end ANN design. We instantiate SmartANN for IVF-PQ and HNSW, covering partition-and-quantization and graph-based ANN families. Experiments on eight real-world datasets show that SmartANN improves Recall by 0.24--74.20%, and increases QPS by 28.8--256.5% at comparable Recall, with low diagnosis and auto-design overhead. The code is available at https://github.com/zhouyutong20/SmartANN.

cs.DB↗
arXiv2026

Novel Transition Mechanisms of Vector Localized Waves Induced by the Fourth-Order Effect

We investigate novel vector localized wave solutions in the coupled Lakshmanan-Porsezian-Daniel equa?tions, which describe the dynamics of the Heisenberg ferromagnetic spin chain. We present Tajiri-Watanabe breathers, rogue waves and resonant modes in both the degenerate and non-degenerate regions, together with the degenerate beating solitons. The fourth-order effect induces state transitions in both regions. In particular, the degenerate breathers can be transformed into solitons, whereas such transitions are absent in the coupled Hirota equations. Moreover, beating solitons can be converted into stable solitons only in the degenerate region, the phenomena not found in the Manakov system. We further uncover the state transitions of the resonant modes and derive the corresponding transition conditions for each branch. We derive the physical spectra and subsequently identify the state transition conditions in the spectral domain for both the degenerate and non-degenerate cases. These spectra provide an additional characterization of the transition dynamics. Finally, direct numerical simulations are performed to verify the validity of the exact solutions.

nlin.PS↗
arXiv2026

The Inert Doublet Model of Dark Matter and the LUX-ZEPLIN High-Recoil Event

The recent LUX-ZEPLIN (LZ) search reported a nuclear-recoil event near $248~{\rm keV}$. Such a high-recoil event can be interpreted in terms of endothermic inelastic dark matter scattering, which typically requires a mass splitting of order a few hundred keV between the initial and final dark states. The inert doublet model (IDM) provides a natural realization of this scenario through the $Z$-mediated transition $H+N\to A+N$. In this work, we systematically examine whether this interpretation can be consistently realized in the IDM under the relevant theoretical constraints and existing experimental bounds. We perform a detailed profile-likelihood analysis of the LZ event, and find a viable high-mass IDM region with a profile best fit at $m_H=1080~{\rm GeV}$ and $m_A-m_H=369~{\rm keV}$.

hep-ph↗
arXiv2026

VTOS: Learning to Orchestrate Vision Tools by Co-Searching Solutions and Observers

Vision foundation tools such as open-vocabulary detectors, segmentation models, and post-processing operators are powerful building blocks for computer vision, but their effectiveness depends heavily on how they are orchestrated: which tools are used, in what order, with what parameters, and under what visual conditions. Existing visual-programming agents typically generate a fixed solution pipeline, making them brittle under dense objects, occlusion, small targets, and domain shift. We introduce VTOS (Vision Tools Orchestration Search), a framework for adaptive visual tool orchestration through joint solution-observer search. VTOS co-searches executable solution programs that compose vision tools such as Grounding DINO, SAM, NMS, and slice-and-detect, together with observer programs that diagnose candidate solutions, identify failure modes, and generate actionable feedback. These observations are accumulated in a shared VisionThoughts knowledge base to guide subsequent search. We evaluate VTOS through two case studies: dense object counting on LVIS-Count and zero-shot plant-disease segmentation on PlantSeg-OOD, which stress different orchestration challenges including threshold calibration, NMS, slicing, mask refinement, and domain generalization. Across both tasks, VTOS outperforms static tool pipelines and agentic visual-programming baselines, specifically in complex settings such as dense, occluded scenes and out-of-distribution segmentation where static pipelines leave measurable headroom, rather than in standard tasks where a single well-calibrated tool already approaches its ceiling.

cs.CV↗
arXiv2026

An Adaptive Projected-Gradient Algorithm for Sample-Average Approximations of Stochastic Multi-Objective Optimization

We consider stochastic multi-objective optimization over a nonempty closed convex set, where every objective is an expectation and only sample-gradient information is available. We develop a line-search-free and function-value-free adaptive projected-gradient algorithm for the sample-average approximation (SAA) problem. Each iteration computes a feasible regularized multi-gradient step and updates the regularization parameter from the projected step length. A normal-cone-based certificate yields descent estimates and an explicit complexity bound for the Pareto-stationarity residual of the SAA problem. The consistency of SAA gradients then transfers vanishing SAA residuals to Pareto stationarity for the population problem, while an additional concentration argument gives a finite-sample residual bound on compact sets. Experiments on synthetic problems, classification, portfolio selection, multi-task learning, and robot control illustrate the practical performance of our algorithm.

math.OC↗
arXiv2026

Probing pair production of long-lived scalars via an off-shell Standard-Model-like Higgs boson at the LHC

We study the collider phenomenology of a long-lived scalar particle $S$ that arises from Higgs mixing in a broad class of Standard-Model (SM) extensions. When the mixing angle is sufficiently small, $S$ becomes long-lived, while its pair production via the Higgs portal can remain sizable. We focus on the production channel $gg \to h^* \to SS$ at the LHC, mediated by an \textit{off-shell} SM-like Higgs boson. This mechanism provides a complementary probe of $S$ in the mass region above the kinematic threshold of the conventional on-shell decay $h \to SS$, thereby extending the accessible parameter space to heavier scalars. The long-lived $S$ particles can decay inside the inner detector, leading to displaced vertices (DVs) accompanied by jets. We perform a detailed Monte Carlo simulation and reinterpret an existing recast of an ATLAS search for DV-plus-jets signatures in this scenario. We also consider a modified analysis strategy based on the same search to assess potential improvements in sensitivity. We find that the current ATLAS search already excludes a significant region of the parameter space, reaching scalar masses up to $m_S \sim 230$~GeV for a benchmark $hSS$ coupling $λv$ of $246$~GeV. The modified analysis and projections to the high-luminosity LHC further extend the sensitivity to wider regions of the mass--lifetime parameter space.

hep-ph↗
arXiv2026

Breaking the Structural Identity: Personalized Federated LoRA Fine-tuning under Rank Heterogeneity

Large Language Models (LLMs) have achieved remarkable success across diverse domains, but their adaptation to privacy-sensitive, distributed datasets remains a challenge. While Federated Learning (FL) combined with Low-Rank Adaptation (LoRA) provides a resource-efficient paradigm for collaborative fine-tuning, practical deployments are hindered by the dual challenges of resource heterogeneity and data heterogeneity. Existing rank-heterogeneous methods primarily focus on bridging dimension mismatches for aggregation but typically provide a unified global model for all clients sharing the same rank, failing to capture client-specific features in non-IID scenarios. In this paper, we propose FedRoRA (Federated Rank-wise Personalized LoRA), a novel framework that enables fine-grained personalization within rank-heterogeneous federations. FedRoRA decouples adaptation into shared global directions and personalized rank-wise magnitudes governed by learnable diagonal scales. On the server side, it extracts a global subspace via singular value decomposition (SVD) and redistributes client-specific initializations through a personalized projection and top-$k$ selection mechanism. Extensive experiments on NLU and NLG benchmarks demonstrate that FedRoRA consistently outperforms state-of-the-art methods.

cs.LG↗
arXiv2026

When to Adapt: Conditional Memory Adapters for Retention-Preserving Domain Specialization

Large language models deployed in specialized domains must improve in-domain performance without sacrificing general capabilities. Existing parameter-efficient fine-tuning methods are typically always on: their learned perturbations are applied to every input, which can degrade out-of-domain (OOD) performance. We propose Engram Adapter, a framework that repurposes pretraining-time conditional memory as a post-hoc adapter for frozen LLMs. It uses multi-channel matching over local n-gram patterns with explicit occupancy tracking as a lightweight selectivity prior, making residual injection more likely on in-domain inputs while a learned scalar gate suppresses incoherent OOD retrievals. We evaluate on Qwen3-4B and Qwen3-8B with AG-News and MedMCQA as adaptation tasks and OOD benchmarks spanning reasoning, translation, code generation, and legal reasoning. Engram Adapter improves in-domain accuracy while preserving 99.4%--100.1% of average OOD performance; on LegalBench it slightly exceeds the frozen base model on average, whereas comparable always-on baselines degrade sharply. Mechanistic analyses show that although OOD activations are non-zero, gate and projection attenuation reduce residuals to approximately 0.08% of hidden-state norm, yielding small KL drift and negligible accuracy change. These results suggest conditional activation is a promising route toward modular, retention-preserving domain specialization over frozen backbones.

cs.CL↗
arXiv2026

Understanding and Enforcing Weight Disentanglement in Task Arithmetic

Task arithmetic provides an efficient, training-free way to edit pre-trained models, yet lacks a fundamental theoretical explanation for its success. The existing concept of ``weight disentanglement" describes the ideal outcome of non-interfering task composition but does not reveal its underlying cause. Crucially, what intrinsic properties of the pre-trained model ($θ_0$) or the task vectors ($τ_t$) enable this disentanglement remains underexplored. In this paper, we introduce Task-Feature Specialization (TFS), a model's ability to allocate distinct internal features to different tasks, as the fundamental principle. We first prove that TFS is a sufficient condition for weight disentanglement. More importantly, we find that TFS also gives rise to an observable geometric consequence: weight vector orthogonality. This positions TFS as the common cause for both the desired functional outcome (disentanglement) and a measurable geometric property (orthogonality). This relationship provides the key insight for our method: since the abstract TFS property is intractable to enforce directly, we can instead promote weight disentanglement by shaping its concrete geometric consequence, orthogonality. Therefore, we propose OrthoReg, a simple and effective regularization method that actively enforces an internal orthogonal structure on weight updates ($ΔW$) that constitute $τ_t$ during fine-tuning. And we theoretically prove that OrthoReg promotes disentanglement. Extensive experiments demonstrate that OrthoReg consistently and significantly enhances the performance of various task arithmetic methods. Code is available at \href{https://github.com/RL-MIND/OrthoReg}{https://github.com/RL-MIND/OrthoReg}.

cs.AI↗
arXiv2026

When Muon Meets Task Interference: A Spectral Perspective on Continual Learning and Model Merging

Continual learning (CL) and model merging (MM) both aim to obtain a single model that performs well across multiple tasks, challenged respectively by catastrophic forgetting and weight-disentanglement error. In the literature, these difficulties are merely treated separately and mitigated through a variety of solutions, while the geometry induced by the base optimizer is treated as an implementation detail. In this work, we show that the two difficulties are in fact two instances of the same phenomenon: a parameter update useful for one task shifts the model's outputs on another. We formalize this shared phenomenon as \textit{task interference} and reduce it to a common layer-wise Frobenius inner product $\langle ΔW_\ell, J_\ell(x)\rangle_F$. This quantity, in turn, is utilized to expose the role of the optimizer. We theoretically derive an upper bound that isolates the spectral norm $\|ΔW_\ell\|_2$ as an optimizer-controllable factor of task interference, and a per-mode analysis shows that this bound tracks the dominant part of the empirical interference. Specifically, we then identify the recent Muon optimizer as a mechanism that regulates this factor by construction. Our work reveals that its elegant control on spectral norm tightens the interference bound for both CL and MM, positioning Muon as a principled optimizer-centric approach complementary to existing solutions. Our theoretcal analysis is well validated by experimental results. Replacing the AdamW optimizer with Muon improves accuracy by up to +5.02 points on the eight-task model-merging benchmark across three CLIP backbones. For continual learning, Muon also delivers uniformly positive gains across ten class-incremental protocols, three task-incremental protocols, and the 11-task MTIL benchmark.

cs.LG↗
arXiv2026

Factorizations of Almost Simple Groups with Applications

The classification of factorizations $G=HK$ of finite almost simple groups, proposed by Wielandt in 1979 and pursued through several partial classifications, has remained open in one major case. We settle that case: for every finite almost simple classical group $G$, we determine all factorizations $G=HK$ in which both $H$ and $K$ have a unique nonsolvable composition factor, completing the classification of factorizations of finite almost simple groups. Among other consequences of this classification, we prove that the smallest dimension of a biperfect bicrossproduct Hopf algebra is $41287680$.

math.GR↗
arXiv2026

Multiplicity-dependent forward jet production in proton-nucleus collisions

Forward jet production in proton--nucleus collisions probes a dilute projectile scattering from a dense small-$x$ nuclear gluon field, while the charged-particle multiplicity inside the jet probes the final-state cascade in the jet cone. We develop a factorization framework for single-inclusive forward jets with an internal multiplicity measurement by combining the Color Glass Condensate (CGC) description of the production process with soft-collinear effective theory (SCET) semi-inclusive jet functions. The construction preserves the inclusive limit exactly: at zero multiplicity weight it reduces to the known inclusive next-to-leading order (NLO)/resummed CGC forward-jet cross section. We define the multiplicity-measured jet operator in a CGC background, formulate the matching that separates PDF, BK/JIMWLK, Sudakov, SiJF, and multiplicity-evolution radiation, and identify the Wilson-line resolution mechanism through which saturation can modify the internal multiplicity evolution. The resulting framework shows that high-multiplicity forward jets are sensitive both to the quark/gluon mixture generated by the CGC production kernel and to finite-$Q_s$ corrections when an early in-cone splitting is resolved by the target. A baseline numerical implementation validates the factorized structure and illustrates the resulting multiplicity-conditioned nuclear modification.

hep-ph↗