Searcharxiv⌕ Search

arXiv subjects

Yue Yu

Publications and source records attributed to Yue Yu.

At least 19 recordsLinked to original sources
arXiv2026

Equilibrium-energy-scaled quantum lattice Boltzmann method for variable-velocity linear advection--diffusion equations

We develop a quantum lattice Boltzmann method for linear advection--diffusion equations with prescribed space- and time-dependent velocities. Direct block-encoding can accumulate normalization factors that suppress success probability exponentially, even when the classical propagator remains bounded. We control this loss by scaling each distribution component by the inverse square root of its local equilibrium weight. For uniformly positive weights and dimensionless relaxation times $τ^*\geq 1/2$, this equilibrium energy scaling gives the Bhatnagar--Gross--Krook collision operator unit spectral norm while preserving the discrete dynamics. Streaming normalization depends on weight ratios at the departure and arrival positions and times of each step; these ratios equal one for uniform steady advection. For periodic streaming, cumulative normalization stays bounded under mesh refinement at fixed physical time if logarithmic derivatives of the weights along streaming trajectories are bounded uniformly in space, time, and mesh resolution. A single-step block-encoding with two flag qubits supports coherent propagation of the full kinetic distribution without intermediate measurements. We compare oracle query bounds for time marching with those for a history-state linear system built from the same scaled operator, under specified assumptions on initial-state preparation and coefficient access. Tests in one and two dimensions and noiseless circuit simulations verify the construction. In a controlled comparison after 50 steps, scaling increases the probability of preparing the quantum state encoding the terminal distribution by approximately $450\times$. Classical refinement diagnostics up to $N_t=16{,}384$ steps confirm bounded cumulative normalization. The scaling controls additional normalization losses while retaining physical diffusive attenuation.

math-ph↗
arXiv2026

A hybrid quantum-classical lattice kinetic scheme with tensor-train feedback for incompressible flows

Nonlinear lattice kinetic simulation requires each updated distribution to supply the velocity and gradients that determine the next step. In a quantum-classical solver, reading out the full distribution incurs work proportional to the lattice size. We construct a tensor-train (TT) interface that retains reusable classical cores at the physical amplitude scale for a full-relaxation lattice kinetic scheme. The cores supply nonlinear coefficients and prepare each quantum input. Energy-scaled propagation and coordinate recovery encode the updated distribution, and alternating least-squares reconstruction converts signed Hadamard-test overlaps into output cores by restoring a known normalization factor. The next step starts from accepted cores without replaying previous quantum evolution. Under positive, unit-sum reference coefficients, positive lower and finite upper density bounds, and Lipschitz equilibrium sensitivity, perturbation bounds relate local defects to feedback errors. Resource bounds account for classical operations, circuit execution, sampling, and verification. Classical diagnostics expose why distribution accuracy alone is insufficient: in 80-step feedback on a $16\times16$ grid, rank cap 4 nearly eliminates the vortex despite a final distribution error below $1\%$, whereas rank cap 12 gives relative velocity error $1.03\times10^{-3}$. Four-step binomial readout simulations show that tightening core sampling tolerance threefold raises the prescribed sufficient budget from $5.97\times10^{12}$ to $5.37\times10^{13}$ shots as the result approaches the fixed-rank, fixed-sweep exact-overlap trajectory. The interface therefore provides a way to connect quantum propagation to nonlinear feedback and to assess the separate demands of compression, velocity accuracy, and readout precision.

math-ph↗
arXiv2026

Spin-Consistency Constraints in Noncollinear Tensor TDA

TDDFT for open-shell systems, whether spin-conserving or spin-flip, has long suffered from spin contamination. This problem arises because the single-excitation space built upon a single Kohn-Sham determinant is not spin-complete. Adopting spin tensor reference states therefore offers an elegant and promising route to resolving this issue. In this work, we revisit the tensor TDDFT equations within the Tamm-Dancoff approximation (TDA) from a noncollinear perspective. We show that, for S = 1/2 reference states, the internal consistency of the spin tensor formulation can, with the aid of the zero-excitation-energy theorem, be recast as a set of constraints that the exchange-correlation kernel must satisfy. Standard noncollinear functionals, however, generally fail to meet these constraints. To address this, we propose a kernel reconstruction scheme that is independent of the specific functional form and free of empirical parameters. This scheme enforces the required constraints, restoring internal consistency in the full spin tensor structure, with spin adaptation following as a natural consequence. Furthermore, when extended to tensor reference states with other values of S, such as S = 1 for the oxygen molecule, the scheme eliminates the so-called artifact states, namely solutions with severely underestimated excitation energies. In addition, the scheme allows the target states that ROKS reference states aim to describe to be expressed and computed, at the TDA level, within the same unified framework as other states, a capability that spin-adapted spin-conserving TDDFT has so far lacked.

physics.chem-ph↗
arXiv2026

Massively parallel Schwarz methods for the high-frequency Helmholtz equation with O(k) convergence

We develop a one-level restricted additive Schwarz solver with perfectly matched layer transmission conditions (RAS--PML) for high-frequency Helmholtz systems. For problems with ${O}(k^d)$ unknowns, $d=2,3$, the solver uses Cartesian decompositions and ${O}(k^d)$ concurrent local solves for a single right-hand side, where $k$ is the wavenumber. Starting from the convergence theory for fixed decompositions established in~\texttt{arXiv:2404.02156}, we address the growing cost {as $k$ increases} of local solves through a nested RAS--PML construction. Under suitable stability and contraction assumptions, we prove that sufficiently accurate recursive local solves preserve convergence of the outer iteration. We also prove that a single inner iteration at each non-terminal level yields a correction algebraically equivalent to one-level RAS--PML, provided that the hierarchy-induced finest local problems and composite restriction and weighted extension operators are retained. This equivalence provides a hierarchical interpretation and permits a direct implementation without recursive transfers. We assess the one-level solver using PML and overlap widths of order ${O}(k^{-1}\log k)$. Constant-coefficient experiments cover $300\leq k\leq19200$ in two dimensions and $60\leq k\leq780$ in three dimensions, reaching $1.47\times10^9$ and $4.75\times10^8$ complex unknowns, respectively. Richardson iteration counts grow approximately linearly with $k$ over these ranges, while total parallel runtime, including setup and solution, exhibits near-linear growth at the larger tested wavenumbers. Further experiments assess the effects of mesh refinement and heterogeneous media.

math.NA↗
arXiv2026

The Root Finding Problem Revisited: Beyond the Robbins-Monro procedure

We study the sequential root-finding problem with repeated noisy measurements. We propose sequential-test bisection STB, which uses a nonparametric sequential sign test to determine when to move to a new design point. Compared with variants of stochastic approximation, STB offers broader adaptivity and faster convergence. When the regression function satisfies $|f(x)|\geq \textsf{c}_0|x-θ|^γ$ in a neighborhood of $θ$ for constants $\textsf{c}_0,γ>0$ where $θ$ is the root, i.e. $f(θ) = 0$, we establish non-asymptotic guarantees showing that STB, without using the knowledge of $γ$, achieves the minimax adaptive optimal convergence rate $\widetilde{O}(n^{-1/(2γ)})$, outperforming Robbins--Monro and related stochastic-approximation methods in convergence rate. Under sub-Gaussian noise, STB provides finite-sample confidence sequences for the root $θ$ without estimating a derivative. In the positive-slope model, we augment STB with {randomized weighted-section and a line fitted to all observations}, leading to asymptotic normality and consistent variance estimation. We formalize the efficiency by a convolution theorem of Hájek-Le Cam type. Furthermore, when the regression function is discontinuous at the root, we show that Robbins--Monro converges at rate $1/n$, whereas STB converges super-polynomially. We extend this construction to multivariate nonparametric level-set estimation by sampling along rays. As a technical byproduct, we establish a new central limit theorem under random stopping, with proof techniques that may be of independent interest.

math.ST↗
arXiv2026

Learning Nonlocal Interactions in Large-Scale Heterogeneous Metamaterials through Scientific Machine Learning

In multiscale systems, nonlocal interactions among spatially distributed constituents can unlock collective, potentially emergent behavior irreducible to individual constituent responses. Beyond characterization, nonlocality in multiscale systems is increasingly viewed as a design element across fracture mechanics, wave physics, and disordered systems. Existing nonlocality theories provide rigorous descriptions and mechanistic insight, but often require prescribed analytic forms, range, symmetries, or periodicity. Machine learning offers a data-driven route to learn interactions, yet principled approaches that tie the network design to physical theory remain nascent. Here we present domain-decomposition attention neural operator (DANO), which learns configuration-dependent interactions in heterogeneous systems governed by linear field equations from field data alone. Through a case study in large-scale optical metasurfaces, we demonstrate DANO's full-field prediction under geometric heterogeneity and across drive conditions, small-to-large domain transfer, and multifunctional inverse design, with learned interactions validated through physics-based interrogation. As an AI-for-Science tool, it supports interaction discovery and scalable, customizable design of nonlocal multiscale systems, including virtual-reality optics.

cs.CE↗
arXiv2026

CoWalk: Cooperative Walks for Agentic Batch Queries on Disk-Resident Graphs

Workloads such as query rewriting, deep research and group-relative policy optimization (GRPO) issue batches of semantically related vector queries. Existing disk-resident approximate nearest neighbor (ANN) systems process these queries independently. On three workloads, 31 to 67 percent of the I/Os of a batch fetch a page that another query of the batch has already fetched. A page cache shared by the batch removes these I/Os, yet it makes a single-thread search only $1.1$ to $1.2{\times}$ faster, because the search time goes to distance computation and the cache does not change the walk of any query. We present CoWalk, which runs the queries of a batch as cooperating walks on one thread. Every query keeps its own candidate list, a page is read once per batch, and the full-precision vector on every fetched page is ranked by every query of the batch. A query therefore reaches the target recall with a shorter candidate list, which computes fewer distances and issues fewer I/Os. At recall@10 of 0.95, CoWalk answers a batch on one thread $1.6$ to $1.9{\times}$ faster than the fastest per-query baseline and issues $2.8$ to $6.9{\times}$ fewer I/Os. On a server it serves $3.0$ to $6.9{\times}$ the throughput of that baseline at a similar batch latency.

cs.DB↗
arXiv2026

HOLogicBench: Evaluating Higher-Order Logical Reasoning in LLMs

Existing LLM logical-reasoning benchmarks largely focus on object-level inference under fixed predicates and rules, leaving higher-order reasoning over rules and policies underexplored. We introduce HOLogicBench, a 1,379-instance benchmark that systematically evaluates five representative higher-order reasoning operations, with Isabelle/HOL formalizations and proof-verified answers. Beyond answer accuracy, HOLogicBench provides solver-verified reference reasoning paths for each instance, enabling fine-grained evaluation of model-generated reasoning against valid derivations. Across 16 model configurations, average accuracy is 53.86\%, with the best configuration reaching 65.99\%. Further analyses reveal that LLMs exhibit limited alignment with valid derivations, remain vulnerable to minimal counterfactual perturbations, and display distinct failure patterns across higher-order reasoning types.

cs.AI↗
arXiv2026

Exploring More, Reasoning Better: Stepwise Risk-Sensitive GRPO for Diffusion Language Models

Diffusion large language models (dLLMs) generate text by denoising a sequence or successive blocks, allowing several tokens to be revealed in parallel. Reinforcement learning with verifiable rewards (RLVR) reuses terminal feedback across these decisions, even as their conditioning context changes. We propose stepwise risk-sensitive GRPO (StepRS-GRPO), which varies the risk coefficient of the group-advantage transformation across denoising states while retaining the underlying trainer. For binary rewards, we show that this transformation is exactly a prompt- and state-dependent rescaling of centered outcome advantages. A capability-based calibration suggests a coefficient scale, while endpoint and interpolation ablations guide schedule selection. Across multiple dLLM backbones and mathematical reasoning benchmarks, StepRS-GRPO improves both pass@1 accuracy and pass@k coverage over centered GRPO, while increasing answer diversity. In our ablation studies, mass-matched controls support the contributions of state allocation and schedule direction, and the gains persist after matching the root mean square (RMS) of the advantages to that of centered GRPO. Reasoning-trace diagnostics further show that the diversity gains from StepRS-GRPO extend beyond final-answer strings.

cs.LG↗
arXiv2026

Harness Self-Evolution for Long-Horizon Terminal Agents

Long-horizon terminal agents may solve a task in one run yet fail across repeated runs. Stochastic errors can propagate through actions and context, creating a gap between capability and reliable execution. We present RiseHarness, a framework for iteratively improving agent harnesses with model weights fixed. It focuses on flipping tasks with both successful and failed runs under the same configuration, using successful runs as references for failure analysis. RiseHarness combines agent trajectories with internal harness traces from a runtime monitor to examine how model actions interact with command execution, feedback, context management, and completion decisions. A Harness Handbook links this evidence to harness behaviors and source code, helping identify recurring problems in mechanisms shared across tasks. RiseHarness then searches other harness frameworks for relevant implementations and adapts them as design references to the current model and execution protocol. Candidate changes undergo behavioral checks and repeated task evaluation; versions with measured gains seed the next iteration. This loop integrates suitable mechanisms from different frameworks into a single harness and uses new execution evidence to guide further changes. We evaluate task success and reliability across models, ablate harness execution evidence and external research, and test transfer to unseen benchmarks with the same model. RiseHarness treats a specific model as the unit of harness adaptation, studying how changes to shared interaction mechanisms can make existing capabilities more reliable across tasks.

cs.AI↗
arXiv2026

WaveLLM: Wavelet-Domain LLM Entropy Modeling for Lossless Image Compression

Lossless image compression fundamentally relies on accurate entropy modeling, driving a recent shift toward large language model (LLM)-based methods. However, existing approaches mainly emphasize the modeling capacity of LLMs while directly serializing images into pixel or residual sequences, leaving heterogeneous signal patterns entangled within a single prediction stream. In this paper, we argue that effective entropy modeling depends not only on model capacity, but also on the predictability of the input representation. Motivated by this insight, we propose WaveLLM, a wavelet-domain LLM-based entropy model for lossless image compression. By decomposing an image into multi-level wavelet subbands, WaveLLM converts a heterogeneous prediction problem into a set of more structured and statistically homogeneous subproblems that are more amenable to autoregressive probability estimation. To bridge numerical coding symbols with the LLM's native text token space, we introduce a dedicated numerical interface featuring value embeddings and a coefficient prediction head. Furthermore, a task-adaptive prompting strategy is designed to enable a single shared model to adapt to diverse subbands within a unified framework. Built on Llama-3.2-3B with LoRA adaptation, WaveLLM achieves 2.51, 2.24, and 2.27 bpsp on Kodak, CLIC Pro, and DIV2K, respectively, outperforming the PixelCNN-based codec using the same wavelet transform by 0.43, 0.36, and 0.46 bpsp, and surpassing recent LLM-based lossless compressors on all three benchmarks.

cs.CV↗
arXiv2026

PairSight: Characterizing and Predicting Operator-Level Interference in GPU Sharing for DNN Inference

Serving DNN inference is costly, yet a single task seldom saturates both the compute units and the memory bandwidth of a GPU. GPU sharing raises utilization by co-locating several tasks on one GPU, at the cost of interference among them. Existing systems estimate this interference with binary compute-memory rules, standalone scaling curves, or model-level predictors. However, these estimates rest on a limited understanding of interference. They leave the software and hardware properties of co-located operators largely unexplored and give no slowdown prediction that is both interpretable and accurate. We conduct an operator-level study of interference under SM partitioning and use hardware counters to trace it to contention on the shared memory path, where the slowdown depends jointly on both co-located operators. Based on these findings, we propose PairSight, a hardware-aware interference predictor that learns how the hardware behaviors of co-located tasks interact and predicts the latency of each task before co-location. On six DNN models including LLMs, PairSight predicts latency under sharing with 4.3% mean error, against 19.0% for a compute-memory intensity rule, and adapts to unseen models and inputs with a few samples.

cs.PF↗
arXiv2026

ConStore: Search-Overlapped Prefetch and Co-Retrieval-Aware Content Storage for Disk-Based ANNS

Disk-based approximate nearest-neighbor search (ANNS) supports retrieval-augmented generation (RAG), search, recommendation, and advertising over corpora that exceed memory capacity. Existing ANNS research often focuses on vector search at small top-K, whereas industrial-scale applications require larger retrieval sets. Fetching the corresponding content from storage remains comparatively underexplored. Our profiling across million- to billion-vector corpora shows that, at large K, content fetch latency can match or exceed vector-search latency and cannot be ignored in end-to-end retrieval. We present ConStore, a content storage engine combining two complementary techniques to reduce end-to-end retrieval latency. Speculative Inline Content Prefetch (SICP) commonly abstracts graph-based and cluster-based ANNS as progressive SSD I/O that refines an in-flight candidate queue. It prefetches content for stable leading candidates under a bounded I/O budget, overlapping content reads with search without changing the final top-K results. Co-Retrieval-Pattern-Aware Grouping (RPAG) uses query logs to place frequently co-retrieved documents in contiguous segments. Workload-Adaptive Segment Reorganization (WASR) adapts this layout to workload shifts by selectively regrouping segments under a bounded rewrite budget. Group Content Fetch (GCF) exploits this layout to coalesce per-document random reads into batched, page-aware I/O requests. SICP works independently of the content layout, while RPAG and GCF reduce fetch cost even without prefetching. Across three multi-hop QA workloads with K from 10 to 1600, ConStore reduces end-to-end retrieval latency over the best per-document-read baseline by up to 48% on DiskANN and 25% on SPANN. Separate experiments on 100M- and billion-vector indexes demonstrate SICP's applicability at larger scales.

cs.DB↗
arXiv2026

ReTrace: Cross-Stage Failure Attribution for Self-Improving Robotic Manipulation

General-purpose multimodal foundation models are increasingly capable of directly controlling robots, yet improving their manipulation capabilities through autonomous interaction remains challenging. A key challenge to such self-improvement is learning from failures in complex manipulation tasks involving cross-stage dependencies, where an action may succeed locally but cause a later stage to fail. To address this challenge, we introduce ReTrace, a self-improvement framework for robotic manipulation through autonomous physical interaction and retrospective reasoning, without provided demonstrations or model-parameter updates. ReTrace retrospectively traces downstream failures to earlier decisions, proposes targeted corrections, and evaluates them through subsequent interactions. Across attempts, ReTrace progressively consolidates validated corrections into a reusable manipulation skill that guide subsequent execution. We evaluate ReTrace on four real-world manipulation tasks using a Franka Research 3 robot. On the two tasks with strong cross-stage dependencies, GPT-6 Astra-low improves from 0% to 80% success on plug insertion and from 0% to 100% on ball pouring, while GPT-6 Sol-xhigh improves from 0% to 10% and 0% to 90%, respectively. The learned skills also transfer across model and reasoning configurations and reduce inference-time decision cost in most evaluated settings.

cs.RO↗
arXiv2026

Scalable Circuit Learning for Interpreting Large Language Models

A prominent research direction in mechanistic interpretability is learning sparse circuits over LLM components to reveal how they jointly produce model behavior. However, raw neurons are polysemantic, making learned circuits hard to interpret. Sparse autoencoder (SAE) features alleviate this, but their high dimensionality makes existing intervention-based circuit learning methods computationally prohibitive. We propose CircuitLasso, a scalable circuit learning approach based on sparse linear regression. CircuitLasso recovers circuits whose structural accuracy matches that of state-of-the-art intervention-based methods on the benchmark data, at a fraction of the computational cost. For interpretability, CircuitLasso efficiently uncovers relationships among SAE features, showing how human-interpretable semantic features propagate through the model and influence its predictions. Finally, we validate the utility of our learned circuits by leveraging their insights to achieve comparable performance at substantially lower cost on a domain-generalization task.

cs.LG↗
arXiv2026

Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading

AI agents have become capable of autonomously completing short, well-specified tasks. However, existing terminal benchmarks largely focus on simple problems that finish within minutes and are evaluated only by their final outcome. This setup overlooks intermediate progress and partial solutions, yielding sparse reward signals and an incomplete picture of agent capability. We introduce Long-Horizon-Terminal-Bench, a terminal benchmark of 46 long-horizon tasks spanning nine categories, including experiment reproduction, software engineering, multimodal analysis, interactive games, and scientific computing. Each task follows a Terminal-Bench-style setup with a reference solution or simulation engine, but is further decomposed into fine-grained graded subtasks. This design enables dense intermediate rewards and partial credit, allowing evaluation to capture not only whether an agent reaches the final goal, but also how far it progresses on open-ended workflows. Tasks in Long-Horizon-Terminal-Bench typically require hundreds of episodes and minutes to hours of execution, stressing long-horizon planning, long-context management, and iterative debugging rather than one-shot problem solving. We evaluate 15 frontier models and find that agents consume on average 9.9M tokens per task, with roughly 231 episodes and 85.3 minutes of execution time per run, making Long-Horizon-Terminal-Bench more demanding than prior terminal-based benchmarks. Even the strongest tested model achieves 15.2% pass@1 at a partial-reward threshold of 0.95 and 10.9% at a perfect-reward threshold of 1.0, while the mean pass rate across models is 4.3% and 1.7% under the two thresholds, respectively. These results reveal headroom for improvement. We further analyze failure modes and error patterns, and release Long-Horizon-Terminal-Bench to support future progress on long-horizon terminal agents.

cs.AI↗
arXiv2026

Long-Horizon-Terminal-Bench 2.0: Automating Diverse Task Construction and Measuring Sustained Agent Progress

Long-horizon terminal benchmarks measure whether an agent can sustain progress through extended workflows with process rewards, not only whether its final state passes a binary check. We present Long-Horizon-Terminal-Bench 2.0 (LHTB 2.0), an extension of Long-Horizon-Terminal-Bench that preserves its 46 original tasks while correcting task and evaluation-infrastructure defects and extending 31 tasks. LHTB 2.0 contains 77 tasks across 13 domains and eight task types. The extended tasks are produced through a systematic reconstruction-task pipeline. We start from a working scientific or engineering artifact, then use a parameterized reference implementation to generate multiple scenarios, reveal six as behavioral evidence, hold out three for evaluation, and use dense progress ladders to reward partial progress while preventing memorization of public outputs. We also extend the evaluation horizon from 90 minutes to three hours and record verifier checkpoints every 30 minutes, enabling direct measurement of sustained progress within a run. Across 1,145 scored model-task runs from 15 frontier models, GPT-6-astra leads with dense mean reward 0.69 and 37 tasks at R at least 0.95, followed by GPT-5.6-sol with dense mean reward 0.67 and 32 tasks at the same threshold. On the 30 reconstruction tasks, binary verifier scores average 0.221, whereas dense ladder scores average 0.653, showing that outcome-only evaluation discards substantial intermediate progress. LHTB 2.0 provides a broader, hardened, and process-aware testbed for studying reliable agent execution over hours.

cs.AI↗
arXiv2026

SpatialHarness: Test-Time Spatial Scaffolding for Fine Robotic Manipulation

Frontier multimodal foundation models (e.g., GPT-6 Astra) have recently shown strong potential for direct robotic control, yet their performance on fine manipulation remains limited. We argue that an important source of failure is not necessarily insufficient policy capability, but insufficient spatial observability, where task-critical spatial relationships may be poorly revealed by the existing physical camera setup. We introduce SpatialHarness, a test-time embodied harness that provides test-time spatial scaffolding for fine robotic manipulation without policy fine-tuning or changes to the physical sensing setup. SpatialHarness maintains an online simulated scene synchronized with real-world execution, identifies task-critical spatial relationships, and renders complementary virtual views that expose them to a frozen multimodal policy. To keep the simulated scene aligned during interaction, we develop interaction-aware scene synchronization that distinguishes static, held, and transition modes. We evaluate SpatialHarness on four real-robot manipulation tasks spanning precise geometric alignment, object-relative placement, and articulated-object interaction. Using the same frozen GPT-6 Astra policy, SpatialHarness substantially improves task success, including from 26.7% to 66.7% on plug insertion and from 0% to 100% on Tower of Hanoi. These results indicate that improving spatial observability at test time can unlock fine-manipulation capabilities already present in strong multimodal foundation models. Project website: https://emilia113.github.io/SpatialHarness/.

cs.RO↗