arXiv subjects

Qian Cheng

Publications and source records attributed to Qian Cheng.

At least 19 recordsLinked to original sources

arXiv2026

Sample-Guided Exact Top-K Selection for Long-Context Sparse Attention

Sparse attention bounds downstream attention work by retaining a fixed-size subset of indexed tokens, but its standalone exact Top-$K$ stage must still process materialized score rows whose length grows with context. Production radix selectors discover their first actionable boundary only after a complete-row pass, forcing another row-scale traversal before exact refinement. We observe that locating a compact upper tail requires substantially less resolution than identifying the exact rank boundary, and that fixed-stride partial views of the current row remain calibrated to the corresponding complete-row rank across ragged lengths. We present HPC-Ops Top-K, a sample-guided exact selector for ragged sparse-attention score rows. A fixed-stride view proposes a row-local coarse boundary; the mandatory complete-row pass certifies its sufficiency, forms the admitted candidate set, and initializes exact FP32 refinement over the unresolved frontier. A nested secondary boundary and exact recovery handle underfilled proposals before any output is committed, so sampling controls common-path work but never correctness. The GPU implementation fuses complete-row certification and candidate formation, and combines persistent, KV-split, and direct-exact execution behind graph-capturable ragged-row dispatch. We evaluate HPC-Ops Top-K on indexer scores from Hy4-Preview. It outperforms the fastest verified external exact baseline by $1.29$--$1.75\times$ across 20 operator configurations, with a $1.55\times$ geometric-mean speedup. It further achieves $1.36\times$ and $1.48\times$ speedups on two framework-derived sparse-attention traces. The implementation is available in HPC-Ops, Tencent's open-source high-performance operator library for LLM inference, at https://github.com/Tencent/hpc-ops.

cs.DC↗

arXiv2026

A Framework Integrating the Dynamic Stiffness Matrix with Physics-Informed Neural Networks for Solving Eigenvalue Problems and Analysing Dynamic Response

This paper introduces a framework that integrates the dynamic stiffness matrix (DSM) with physics-informed neural networks (PINN). The DSM-PINN embeds physical constraints within the model and demonstrates robustness, particularly when addressing limited datasets across diverse investigations. In this approach, deep neural network outputs approximate the displacement fields of element nodes. Unlike the finite element method (FEM), the element shape functions are homogeneous solutions to the governing partial differential equation, forming the basis of the exact dynamic stiffness matrix, thereby avoiding high-order derivative terms. This matrix also serves as a frequency-domain spectral element, resulting in a strong-form PINN. The loss function is produced by connecting neural networks with dynamic stiffness matrices. We focus on utilising PINNs to resolve eigenvalue problems by employing the Wittrick-Williams algorithm, which overcomes the challenge of neural networks failing to converge to higher-order eigenvalues. Additionally, the frequency domain-PINN method is used to analyse structural dynamic responses under moving and impulsive loads, addressing the limitation of neural networks in handling complex numbers. Theoretical convergence stability of the suggested approach is also analysed even DSM is an indefinite matrix after implementing the boundary condition. The numerical results validate the practicality and efficacy of the recommended approach.

math.NA↗

arXiv2026

Specula: Scaling formal specifications for autonomous model checking of system code

Specula is a push-button agentic system that generates high-quality formal specifications for large, complex system code and uses the specifications for highly effective model checking and bug finding. Specula employs large language model (LLM) based coding agents to autonomously develop TLA+ specifications, including invariants that describe correctness properties of the target system and formal models that describe the system implementation with the right level of abstractions. Specula is fully autonomous and thus eliminates the barrier of applying formal methods to real-world system code (as in traditional human-centric approaches). Meanwhile, Specula addresses limitations of LLM-driven techniques like reward hacking and hallucinations through self-evolving loops that iteratively improve specification quality by enabling the agents to deepen their understanding of system code and its behaviors. We have used Specula to check 48 open-source system projects; Specula found 249 bugs including many deep bugs that are hard to find by existing approaches. Specula has been used by several companies and is maintained at https://github.com/specula-org/Specula.

cs.SE↗

arXiv2026

Learning What to Remember: A Cognitively Grounded Multi-Factor Value Model for Agentic Memory

Long-running LLM agents accumulate interaction histories far larger than any context window, forcing a standing decision: what to encode deeply, what to forget, and what to retrieve under a fixed memory budget. Production systems answer with semantic similarity or recency -- both mis-specified for the forgetting decision, which is made at consolidation time before the future query is known. We propose a multi-factor memory value function V(m)=\sum_i w_i f_i(m) over seven interpretable factors (emotional intensity, goal relevance, value alignment, self/user relevance, task utility, reliability, and usage history) drawn from cognitive psychology, whose weights are learned from a downstream objective by a gradient-free optimiser, and whose single scalar uniformly controls encoding depth, forget risk, and retrieval rank. We make a methodological point: on LongMemEval, scoring goal relevance against the held-out evaluation question saturates gold-evidence retention at \approx 0.98 -- this measures retrieval, not forgetting. In the realistic blind regime, a learned multi-factor value retains 0.770 \pm 0.011 of gold evidence across 479 usable cases, versus 0.657 for uniform weights, 0.518 for the best single factor, and 0.368 for recency; every paired gap's 95% bootstrap CI is above zero, and a neural network over the same factors ties the linear model. The learned weights are interpretable -- reliability, emotional intensity, and self/user relevance dominate, while query-time goal similarity is correctly down-weighted for the forgetting decision. A controlled synthetic task with planted confounds confirms the learner recovers a separating weighting (1.00 retention) where uniform weighting fails (0.62). The substrate is open-source; all experiments run on a single CPU with no API calls.

cs.AI↗

arXiv2026

GaussianDream: A Feed-Forward 3D Gaussian World Model for Robotic Manipulation

Vision-language-action (VLA) policies have advanced language-conditioned robotic manipulation by transferring semantic priors from pretrained vision-language models to action generation. However, standard action-imitation learning often lacks sufficient modeling of explicit 3D spatial information, dense geometric supervision, and future environment evolution, all critical for precise robotic interaction. To address this, we propose \textbf{GaussianDream}, a feed-forward 3D Gaussian world-model plug-in. Specifically, we introduce learnable GaussianDream Queries in the encoder, enabling the model to capture current-frame 3D spatial structure and short-horizon future evolution. During training, the latent GaussianDream prefix is processed by a static reconstruction head and a future prediction head to produce current 3D Gaussian scene states and future Gaussian evolution states. The current branch is supervised by RGB rendering and depth, while the future branch uses future RGB, depth, and pseudo 3D scene-flow signals. During inference, GaussianDream discards all auxiliary heads and retains only the learned prefix to condition action generation, without test-time Gaussian reconstruction or future prediction. Experimental results demonstrate that GaussianDream achieves state-of-the-art performance across multiple robotic manipulation benchmarks, reaching \textbf{98.4\%} on LIBERO, \textbf{54.8\%} on RoboCasa Human-50, and \textbf{50.0\%} on real-robot tasks. Compared with existing 3D-enhanced VLA methods, GaussianDream achieves strong accuracy while providing higher inference efficiency than video-based world-model approaches.

cs.RO↗

arXiv2026

Physics-informed neural networks for quantitative assessment of cancellous bone microstructure from photoacoustic signals

Artificial intelligence (AI) empowers innovative diagnostic tools for common diseases, yet its clinical application in skeletal health evaluation is constrained by unsatisfactory accuracy, owing to the inherent porous and poroelastic biophysical features of bone. To address such bottlenecks amid global population aging, this study targets skeletal health and develops a reliable AI framework for precise bone microstructural characterization. We proposed Biot-PINN, a physics-informed neural network embedded with Biot's poroelasticity theory to characterize mechanical responses and wave propagation in poroelastic bone tissues. By decoding photoacoustic signals encoding bone mineral and microstructural features, the framework enables automatic bone microstructural grading. Experimental results reveal that Biot-PINN reaches an accuracy of 97%, markedly surpassing traditional data-driven approaches and providing a robust solution for early skeletal health diagnosis.

physics.med-ph↗

arXiv2026

A Physics-Constrained Learning Framework for Wave Propagation in Complex Poroelastic Multilayered Media

Wave propagation through complex poroelastic multilayered media is difficult to model and invert because pronounced heterogeneity, scattering, mode conversion and fluid-solid coupling jointly distort acoustic signals during propagation. Here we present Physics-Constrained Learning for Complex Multilayered Media (PCL-CMM), a general framework that integrates Biot's poroelastic theory with the elastic wave equation to bridge the gap between physically rigorous wave modelling and data-driven learning. PCL-CMM constructs a high-fidelity digital twin that dynamically computes an effective acoustic stiffness tensor for forward wave modelling and incorporates the resulting physical constraint as a loss term to regularize the training of deep neural networks. We demonstrate PCL-CMM on transcranial photoacoustic imaging, where skull-induced acoustic distortions severely degrade image formation. Across simulations and ex vivo experiments, PCL-CMM effectively compensates for these distortions and improves SSIM by more than 0.06 compared with purely data-driven neural networks. This work establishes a physics-constrained learning framework for acoustic wave modelling in complex poroelastic multilayered media.

physics.med-ph↗

arXiv2025

LSRE: Latent Semantic Rule Encoding for Real-Time Semantic Risk Detection in Autonomous Driving

Real-world autonomous driving must adhere to complex human social rules that extend beyond legally codified traffic regulations. Many of these semantic constraints, such as yielding to emergency vehicles, complying with traffic officers' gestures, or stopping for school buses, are intuitive for humans yet difficult to encode explicitly. Although large vision-language models (VLMs) can interpret such semantics, their inference cost makes them impractical for real-time deployment. This work proposes LSRE, a Latent Semantic Rule Encoding framework that converts sparsely sampled VLM judgments into decision boundaries within the latent space of a recurrent world model. By encoding language-defined safety semantics into a lightweight latent classifier, LSRE enables real-time semantic risk assessment at 10 Hz without per-frame VLM queries. Experiments on six semantic-failure scenarios in CARLA demonstrate that LSRE attains semantic risk detection accuracy comparable to a large VLM baseline, while providing substantially earlier hazard anticipation and maintaining low computational latency. LSRE further generalizes to rarely seen semantic-similar test cases, indicating that language-guided latent classification offers an effective and deployable mechanism for semantic safety monitoring in autonomous driving.

cs.RO↗

arXiv2025

Are All Data Necessary? Efficient Data Pruning for Large-scale Autonomous Driving Dataset via Trajectory Entropy Maximization

Collecting large-scale naturalistic driving data is essential for training robust autonomous driving planners. However, real-world datasets often contain a substantial amount of repetitive and low-value samples, which lead to excessive storage costs and bring limited benefits to policy learning. To address this issue, we propose an information-theoretic data pruning method that effectively reduces the training data volume without compromising model performance. Our approach evaluates the trajectory distribution information entropy of driving data and iteratively selects high-value samples that preserve the statistical characteristics of the original dataset in a model-agnostic manner. From a theoretical perspective, we show that maximizing trajectory entropy effectively constrains the Kullback-Leibler divergence between the pruned subset and the original data distribution, thereby maintaining generalization ability. Comprehensive experiments on the NuPlan benchmark with a large-scale imitation learning framework demonstrate that the proposed method can reduce the dataset size by up to 40% while maintaining closed-loop performance. This work provides a lightweight and theoretically grounded approach for scalable data management and efficient policy learning in autonomous driving systems.

cs.RO↗

arXiv2025

SysMoBench: Evaluating AI on Formally Modeling Complex Real-World Systems

Formal models are essential to specifying large, complex computer systems and verifying their correctness, but are notoriously expensive to write and maintain. Recent advances in generative AI show promise in generating certain forms of specifications. However, existing work mostly targets small code, not complete systems. It is unclear whether AI can deal with realistic system artifacts, as this requires abstracting their complex behavioral properties into formal models. We present SysMoBench, a benchmark that evaluates AI's ability to formally model large, complex systems. We focus on concurrent and distributed systems, which are keystones of today's critical computing infrastructures, encompassing operating systems and cloud infrastructure. We use TLA+, the de facto specification language for concurrent and distributed systems, though the benchmark can be extended to other specification languages. We address the primary challenge of evaluating AI-generated models by automating metrics like syntactic and runtime correctness, conformance to system code, and invariant correctness. SysMoBench currently includes eleven diverse system artifacts: the Raft implementation of Etcd and Redis, the leader election of ZooKeeper, the Spinlock, Mutex, and Ringbuffer in Asterinas OS, etc., with more being added. SysMoBench enables us to understand the capabilities and limitations of today's LLMs and agents, putting tools in this area on a firm footing and opening up promising new research directions.

cs.AI↗

arXiv2025

DRARL: Disengagement-Reason-Augmented Reinforcement Learning for Efficient Improvement of Autonomous Driving Policy

With the increasing presence of automated vehicles on open roads under driver supervision, disengagement cases are becoming more prevalent. While some data-driven planning systems attempt to directly utilize these disengagement cases for policy improvement, the inherent scarcity of disengagement data (often occurring as a single instances) restricts training effectiveness. Furthermore, some disengagement data should be excluded since the disengagement may not always come from the failure of driving policies, e.g. the driver may casually intervene for a while. To this end, this work proposes disengagement-reason-augmented reinforcement learning (DRARL), which enhances driving policy improvement process according to the reason of disengagement cases. Specifically, the reason of disengagement is identified by a out-of-distribution (OOD) state estimation model. When the reason doesn't exist, the case will be identified as a casual disengagement case, which doesn't require additional policy adjustment. Otherwise, the policy can be updated under a reason-augmented imagination environment, improving the policy performance of disengagement cases with similar reasons. The method is evaluated using real-world disengagement cases collected by autonomous driving robotaxi. Experimental results demonstrate that the method accurately identifies policy-related disengagement reasons, allowing the agent to handle both original and semantically similar cases through reason-augmented training. Furthermore, the approach prevents the agent from becoming overly conservative after policy adjustments. Overall, this work provides an efficient way to improve driving policy performance with disengagement cases.

cs.RO↗

arXiv2025

Enhanced oil recovery in reservoirs via diffusion-driven $\text{CO}_{2}$ flooding: Experimental insights and material balance modeling

$\text{CO}_{2}$ flooding is central to carbon utilization technologies, yet conventional waterflooding models fail to capture the complex interactions between CO$_2$ and formation fluids. In this study, one- and two-dimensional nuclear magnetic resonance experiments reveal that $\text{CO}_{2}$ markedly enhances crude oil mobility during miscible displacement via multiple synergistic mechanisms, yielding a recovery factor of $60.97\%$, which surpasses that of immiscible displacement (maximum $57.53\%$). Guided by these findings, we propose a convection-diffusion model that incorporates the diffusion coefficient ($D$) and porosity ($\phi$) as key parameters. This model captures the spatiotemporal evolution of the $\text{CO}_{2}$ front and addresses a key limitation of conventional formulations-the omission of diffusion effects. It improves predictions of gas breakthrough time and enables optimized injection design for low-permeability reservoirs. Extending classical material balance theory, we develop an enhanced $\text{CO}_{2}$ flooding equation that integrates critical transport phenomena. This formulation incorporates $\text{CO}_{2}$ diffusion, oil phase expansion, reservoir adsorption, and gas compressibility to describe the dynamic transport and mass compensation of injected $\text{CO}_{2}$. Validation through experimental and numerical data confirms the model's robustness and applicability under low-permeability conditions. The proposed framework overcomes limitations of physical experiments under extreme environments and offers theoretical insight into oil recovery enhancement and $\text{CO}_{2}$ injection strategy optimization.

physics.flu-dyn↗

arXiv2024

Detecting collagen by machine learning improved photoacoustic spectral analysis for breast cancer diagnostics: feasibility studies with murine models

Collagen, a key structural component of the extracellular matrix, undergoes significant remodeling during carcinogenesis. However, the important role of collagen levels in breast cancer diagnostics still lacks effective in vivo detection techniques to provide a deeper understanding. This study presents photoacoustic spectral analysis improved by machine learning as a promising non-invasive diagnostic method, focusing on exploring collagen as a salient biomarker. Murine model experiments revealed more profound associations of collagen with other cancer components than in normal tissues. Moreover, an optimal set of feature wavelengths was identified by a genetic algorithm for enhanced diagnostic performance, among which 75% were from collagen-dominated absorption wavebands. Using optimal spectra, the diagnostic algorithm achieved 72% accuracy, 66% sensitivity, and 78% specificity, surpassing full-range spectra by 6%, 4%, and 8%, respectively. The proposed photoacoustic methods examine the feasibility of offering valuable biochemical insights into existing techniques, showing great potential for early-stage cancer detection.

physics.optics↗

arXiv2024

Longitudinal photoacoustic monitoring of collagen evolution modulated by cancer-associated fibroblasts: simulation and experiment studies

Noninvasive in vivo detection of collagen facilitates the investigation of mechanisms by which cancer-associated fibroblast (CAF) regulates the extracellular matrix. This study explored the feasibility of photoacoustic spectrum analysis (PASA) in identifying longitudinal changes of collagen modulated by CAFs using simulations and experiment studies. Optical and acoustic simulations in tissues were performed based on the histological slides of maximum cross-sections of murine malignancies to verify the effectiveness of photoacoustic (PA) detection system and the parameter "relative area of power spectrum density (APSD)". Experiments were conducted on three groups of mouse models with incremental ratios of CAFs and breast cancer cells at 3 continuous time points. Results discovered that the system configuration and APSD were capable of reflecting the evolution of collagen during cancer growth. Furthermore, cancers receiving a high dose of CAFs exhibited a suppressed collagen level. The presented methods show great potential for clinical translation of PASA in the field of cancer therapies targeting CAFs.

physics.bio-ph↗

arXiv2024

Photoacoustic tracking of photo-magnetically powered nanoparticles for cancer therapy

The in vivo propulsion and monitoring of nanoparticles (NPs) have received tremendous achievements in the past decade. Developing functional NPs that can be efficiently manipulated inside the human body with a non-invasive tracking modality is critical to clinical translation. This study synthesized a photo-magnetically powered nanoparticle (PMN) with a Fe3O4 core and gold spiky surface. The Au-nanotips ensure PMNs have a strong light absorption in the second near-infrared (NIR) window and produce outstanding photoacoustic signals. The Bio-transmission electron microscopy and simulation results prove that the assembly of PMNs under a magnetic field further enhances the photothermal conversion in cells, contributing to the reduction of ambient viscosity. Photoacoustic imaging (PAI) realized real-time monitoring of PMN movements and revealed that laser plus magnetic coupling couldimprove intratumoral distribution and retention. The proposed methods exhibit excellent potential for the clinical research of cancer nanotherapies.

physics.bio-ph↗

arXiv2024

Universal Murray's law for optimised fluid transport in synthetic structures

Materials following Murray's law are of significant interest due to their unique porous structure and optimal mass transfer ability. However, it is challenging to construct such biomimetic hierarchical channels with perfectly cylindrical pores in synthetic systems following the existing theory. Achieving superior mass transport capacity revealed by Murray's law in nanostructured materials has thus far remained out of reach. We propose a Universal Murray's law applicable to a wide range of hierarchical structures, shapes and generalised transfer processes. We experimentally demonstrate optimal flow of various fluids in hierarchically planar and tubular graphene aerogel structures to validate the proposed law. By adjusting the macroscopic pores in such aerogel-based gas sensors, we also show a significantly improved sensor response dynamic. Our work provides a solid framework for designing synthetic Murray materials with arbitrarily shaped channels for superior mass transfer capabilities, with future implications in catalysis, sensing and energy applications.

physics.app-ph↗

arXiv2024

A Survey on Monocular Re-Localization: From the Perspective of Scene Map Representation

Monocular Re-Localization (MRL) is a critical component in autonomous applications, estimating 6 degree-of-freedom ego poses w.r.t. the scene map based on monocular images. In recent decades, significant progress has been made in the development of MRL techniques. Numerous algorithms have accomplished extraordinary success in terms of localization accuracy and robustness. In MRL, scene maps are represented in various forms, and they determine how MRL methods work and how MRL methods perform. However, to the best of our knowledge, existing surveys do not provide systematic reviews about the relationship between MRL solutions and their used scene map representation. This survey fills the gap by comprehensively reviewing MRL methods from such a perspective, promoting further research. 1) We commence by delving into the problem definition of MRL, exploring current challenges, and comparing ours with existing surveys. 2) Many well-known MRL methods are categorized and reviewed into five classes according to the representation forms of utilized map, i.e., geo-tagged frames, visual landmarks, point clouds, vectorized semantic map, and neural network-based map. 3) To quantitatively and fairly compare MRL methods with various map, we introduce some public datasets and provide the performances of some state-of-the-art MRL methods. The strengths and weakness of MRL methods with different map are analyzed. 4) We finally introduce some topics of interest in this field and give personal opinions. This survey can serve as a valuable referenced materials for MRL, and a continuously updated summary of this survey is publicly available to the community at: https://github.com/jinyummiao/map-in-mono-reloc.

cs.RO↗

arXiv2023

Theoretical and experimental study of attenuation in cancellous bone

Photoacoustic (PA) technology can provide information on both the physical structure and chemical composition of bone, showing great potential in bone assessment. However, due to the complex composition and porous structure of cancellous bone, the PA signals generated and propagated in cancellous bone are complex and difficult to be directly used in cancellous bone analysis. In this paper, a photoacoustic differential attenuation spectrum (PA-DAS) method is proposed. By eliminating the PA spectrum of the optical absorption sources, the propagation attenuation characteristics of cancellous bone are studied theoretically and experimentally. An analytical solution for the propagation attenuation of broadband ultrasound waves in cancellous bone is given by applying high-frequency and viscous corrections to Biot's theory. An experimental system of PA-DAS with an eccentric excitation differential detection system is established to obtain the PA-DAS of cancellous bone and its acoustic propagation characteristic on the rabbit osteoporosis model. The PA-DAS quantization parameter slope is further extracted to quantify the attenuation of high and low frequency components. The results show that the PA-DAS can distinguish osteoporotic bone from normal bone, enabling quantitative assessment of bone mineral density and the diagnosis of osteoporosis.

physics.med-ph↗