arXiv subjects

Fang Zhang

Publications and source records attributed to Fang Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Data-Driven Risk Fields for Safer End-to-End Autonomous Driving

Safety is a fundamental requirement for autonomous driving, yet existing end-to-end driving models still lack explicit risk-aware learning capacities. Existing rule-based risk models provide interpretable safety priors, yet their absolute risk scores depend on handcrafted functions, coefficients, and thresholds. Learning-based risk representations reduce part of this manual design, but their supervision often relies on occupancy-derived labels or heuristic cost values, which may not capture ego-conditioned planning risk. In this paper, we propose DRiF, a data-driven risk-field framework for safer end-to-end autonomous driving. DRiF learns a shared BEV feature with static map segmentation, dynamic risk prediction, and vehicle planning. For dynamic risk learning, DRiF converts rule-based safety priors into pairwise risk labels, and trains the risk field to preserve relative risk ordering instead of regressing handcrafted absolute scores. Experiments on Bench2Drive show that DRiF achieves competitive overall performance, with consistent improvements in driving score, success rate, and collision-related metrics. These results establish relative risk supervision as an effective way to connect explicit safety structure with end-to-end planning. The data and code will be publicly available.

cs.RO↗

arXiv2026

Roadside-Cooperative Autonomous Driving: From Data Platform to Vision-Language End-to-End Reasoning

Vehicle-to-Everything (V2X) cooperation enables beyond-line-of-sight perception, mitigating occlusions in single-vehicle sensing. However, existing V2X benchmarks provide limited support for closed-loop evaluation and language-grounded supervision, hindering the development of vision-language models (VLMs) for end-to-end cooperative driving. To address these limitations, we introduce V2XBench, a simulation platform featuring synchronized ego--roadside sensing and closed-loop evaluation, together with Chat-V2XBench, a progressively structured VQA dataset for cooperative reasoning. Building upon this benchmark infrastructure, we propose AURORA, an end-to-end cooperative driving framework. Equipped with a dual-view perception architecture, AURORA mitigates spatial and semantic discrepancies across ego and roadside viewpoints through a query-level Cross-View Query Alignment and Fusion (CQAF) module. Leveraging the resulting unified tokens, a LoRA-adapted VLM bridges semantic reasoning and generative trajectory planning. Extensive closed-loop evaluations on V2XBench demonstrate that AURORA achieves state-of-the-art performance in heavily occluded scenarios, with a Route Completion rate of 98.21% and a Driving Score of 76.02, while requiring low roadside communication bandwidth. Ultimately, this work pioneers an extensible V2X--VLM paradigm, paving the way for next-generation cooperative autonomous driving.

cs.RO↗

arXiv2026

RISE: Roadside Infrastructure Sequence Understanding across 3D Tracking and Structured Vision-Language Reasoning

We present RISE (Roadside Infrastructure Sequence Understanding and Evaluation), a framework spanning metric 3D tracking and structured vision-language reasoning in roadside sequences. For metric tracking, our image-only method combines SAM3 video identities with calibration-guided mask agreement for multi-view identity association, recovering persistent 3D tracks without LiDAR or task-specific 3D training. Its calibration-conditioned geometry allows the procedure to be instantiated at different calibrated multi-camera intersections without layout-specific retraining. On 20 human-reviewed clips from six intersections, the generated tracks achieve 66.9 MOTA within the defined multi-view evaluation scope. For structured vision-language reasoning, a human-reviewed MLLM pipeline mines high-value clips and uses a constrained full-context Oracle to construct bbox-grounded predictive QA without exposing future evidence to evaluated models. The resulting RISE-VQA dataset contains 33,910 QA pairs from 557 clips across 16 intersections and 61 roadside views. Its intersection-held-out RISE-Bench evaluates semantic choices, coordinates, future boxes, and interaction sets with deterministic task-specific metrics. Experiments show consistent benefits from domain adaptation and generally from temporal context, while revealing persistent challenges in spatial grounding, future localization, and interaction reasoning.

cs.CV↗

arXiv2026

DriveCode: Domain Specific Numerical Encoding for LLM-Based Autonomous Driving

Large language models (LLMs) have shown great promise for autonomous driving. However, discretizing numbers into tokens limits precise numerical reasoning, fails to reflect the positional significance of digits in the training objective, and makes it difficult to achieve both decoding efficiency and numerical precision. These limitations affect both the processing of sensor measurements and the generation of precise control commands, creating a fundamental barrier for deploying LLM-based autonomous driving systems. In this paper, we introduce DriveCode, a novel numerical encoding method that represents numbers as dedicated embeddings rather than discrete text tokens. DriveCode employs a number projector to map numbers into the language model's hidden space, enabling seamless integration with visual and textual features in a unified multimodal sequence. Evaluated on OmniDrive, DriveGPT4, and DriveGPT4-V2 datasets, DriveCode demonstrates superior performance in trajectory prediction and control signal generation, confirming its effectiveness for LLM-based autonomous driving systems.

cs.CV↗

arXiv2026

LOCUS: Local Visual Cue Search for Enhancing Fine-Grained Perception in Multimodal Large Language Models

Multimodal Large Language Models (MLLMs) remain unreliable on fine-grained visual perception, even when high-resolution inputs preserve the necessary local details. We identify this limitation as visual context rot: decisive evidence may exist in the full image, yet fail to be reliably selected and used amid redundant visual context. We propose LOCUS (LOcal visual CUe Search), a training framework that teaches MLLMs to internalize local evidence search through a verifiable proxy task. During training, LOCUS provides a local crop as a visual cue and optimizes the model to recover its spatial support in the full image using an IoU-based reward. The visual cue is used only during training, leaving the standard image-question inference interface unchanged. Experiments across fine-grained perception, hallucination, general understanding, and reasoning benchmarks show that LOCUS improves localization-sensitive visual understanding while preserving broad capabilities. Attention analyses further indicate stronger focus on task-relevant evidence regions, suggesting that training-time visual cue search provides an effective route to internalized fine-grained evidence selection.

cs.CV↗

arXiv2026

AgentSociety 2: An Integrated Research Environment for Executable Social Science

AI scientist systems are beginning to automate parts of scientific research, but social science poses a distinct challenge: its objects of inquiry are not merely datasets or laboratory protocols, but integrated social processes involving situated participants, interaction contexts, interventions, and outcomes. Yet a critical link is missing: existing systems either assist isolated research tasks or simulate agents as experimental subjects, leaving the research workflow and simulated society decoupled. Here we introduce AgentSociety 2, an Integrated Research Environment for executable social science. It couples two roles of LLM agents in the same runtime: AI social scientists that coordinate literature grounding, hypothesis generation, experiment design, simulation execution, result interpretation, and manuscript drafting; and silicon participants that generate behavioral responses within configurable social environments. This dual-role design turns hypotheses into auditable agent behaviors, environment rules, interventions, and measurements, thereby supporting an end-to-end workflow. Across seven illustrative studies spanning micro-level social-science laboratory experiments, meso-level dynamics in social media, and macro-level urban scenarios, we demonstrate its capacity to support diverse disciplinary questions, reproduce major qualitative patterns from prior studies, identify informative deviations, and enable large-scale simulations through optimized agent-environment interactions. By preserving human researchers' high-level agency while delegating procedural orchestration to agentic systems, it provides a human-in-the-loop and controllable infrastructure for next-generation computational social science, with broader applications in scalable computational social experimentation and AI-enabled social governance platforms.

cs.CY↗

arXiv2026

Quantum Theory of Exciton Magnetic Moment: Interaction and Topological Effects

Combining magnetometry with optical spectroscopy has uncovered novel quantum phenomena and is emerging as a powerful probe of quantum materials. However, the theory of the magnetic response of excitons, correlated electron-hole pairs in insulators, remains incomplete due to insufficient treatment of electron-hole interactions and quantum geometric effects. In biased bilayer graphene, for instance, theoretical predictions of valley g-factors for p-excitons deviate from experiment by nearly an order of magnitude. Here, we develop a quantum theory of the exciton orbital magnetic moment, based on first-order perturbation theory within the GW plus Bethe-Salpeter Equation approach and a rigorous treatment of the position operator in the response of exciton states to a magnetic field. Our formalism reveals three distinct contributions that go beyond the heuristic approaches used in the literature: a Berry-phase-corrected single-particle electron and hole moment difference, a term from envelope-function winding linked to electron-hole relative motion, and a center-of-mass correction from exciton band quantum geometry, with the latter two being completely new effects not considered in previous studies. Our ab initio calculations yield results in excellent agreement with experiment, establishing the importance of interaction and quantum geometric effects in the magnetic response of excitons.

cond-mat.mes-hall↗

arXiv2026

Bunny Codes: Broadening Superconducting Quantum Error Correction Capability through Advanced Control Engineering

Drawing on advances in superconducting qubit control schemes that unlock enriched native gate sets at the hardware level, we systematically examine how harnessing this enlarged physical two-qubit gate pool---specifically CNOT and CXSWAP---streamlines syndrome extraction for certain qLDPC codes with nonlocal stabilizers. Through an exhaustive search, we discover a set of qLDPC codes with various stabilizer weights and distances that can be implemented on the two-dimensional nearest-neighbor qubit connectivity native to superconducting hardware while achieving performance equivalent to that of the direct CNOT implementation requiring long-range interactions. We refer to those codes as Bunny codes. Across all code distances we examine, the best Bunny codes with weight-6 stabilizers in periodic boundary conditions have a code rate approximately $3\times$ that of the toric code; when converted to open boundary conditions, they retain an approximately $2\times$ code rate advantage over the rotated surface code. In circuit-level simulation, we find that some Bunny codes exhibit logical error rates an order of magnitude lower than toric codes with comparable code rates. Our results demonstrate that high-performance quantum error correction can be achieved using an expanded gate set rather than long-range couplers, thereby significantly reducing hardware complexity.

quant-ph↗

arXiv2026

Benchmarking fault-tolerant quantum computing hardware via QLOPS

It is widely recognized that quantum computing has profound impacts on multiple fields, including but not limited to cryptography, machine learning, materials science, etc. To run quantum algorithms, it is essential to develop scalable quantum hardware with low noise levels and to design efficient fault-tolerant quantum computing (FTQC) schemes. Currently, various FTQC schemes have been developed for different hardware platforms. However, a comprehensive framework for the analysis and evaluation of these schemes is still lacking. In this work, we propose Quantum Logical Operations Per Second (QLOPS) as a metric for assessing the performance of FTQC schemes on quantum hardware platforms. This benchmarking framework will integrate essential relevant factors, e.g., the code rates of quantum error-correcting codes, the accuracy, throughput, and latency of the decoder. Through a resource analysis of factoring RSA-2048, we demonstrate that QLOPS reflects the practical requirements of quantum algorithm execution. This framework will enable the identification of bottlenecks in quantum hardware, providing potential directions for their development. Moreover, our results will help establish a comparative framework for evaluating FTQC designs. As this benchmarking approach considers practical applications, it may assist in estimating the hardware resources needed to implement quantum algorithms and offers preliminary insights into potential timelines.

quant-ph↗

arXiv2026

Driving risk emerges from the required two-dimensional joint evasive acceleration

Most autonomous driving safety benchmarks use time-to-collision (TTC) to assess risk and guide safe behaviour. However, TTC-based methods treat risk as a one-dimensional closing problem, despite the inherently two-dimensional nature of collision avoidance, and therefore cannot faithfully capture risk or its evolution over time. Here, we report evasive acceleration (EA), a hyperparameter-free and physically interpretable two-dimensional paradigm for risk quantification. By evaluating all possible directions of collision avoidance, EA defines risk as the minimum magnitude of a constant relative acceleration vector required to alter the relative motion and make the interaction collision-free. Using interaction data from five open datasets and more than 600 real crashes, we derive percentile-based warning thresholds and show that EA provides the earliest statistically significant warning across all thresholds. Moreover, EA provides the best discrimination of eventual collision outcomes and improves information retention by 54.2-241.4% over all compared baselines. Adding EA to existing methods yields 17.5-95.5 times more information gain than adding existing methods to EA, indicating that EA captures much of the outcome-relevant information in existing methods while contributing substantial additional nonredundant information. Overall, EA better captures the structure of collision risk and provides a foundation for next-generation autonomous driving systems.

cs.RO↗

arXiv2026

AgentSociety: Large-Scale Simulation of LLM-Driven Generative Agents Advances Understanding of Human Behaviors and Society

Understanding human behavior and society is a central focus in social sciences, with the rise of generative social science marking a significant paradigmatic shift. By leveraging bottom-up simulations, it replaces costly and logistically challenging traditional experiments with scalable, replicable, and systematic computational approaches for studying complex social dynamics. Recent advances in large language models (LLMs) have further transformed this research paradigm, enabling the creation of human-like generative social agents and realistic simulacra of society. In this paper, we propose AgentSociety, a large-scale social simulator that integrates LLM-driven agents, a realistic societal environment, and a powerful large-scale simulation engine. Based on the proposed simulator, we generate social lives for over 10k agents, simulating their 5 million interactions both among agents and between agents and their environment. Furthermore, we explore the potential of AgentSociety as a testbed for computational social experiments, focusing on five key social issues: polarization, the spread of inflammatory messages, the effects of universal basic income policies, the impact of external shocks such as hurricanes, and urban sustainability. These five issues serve as valuable cases for assessing AgentSociety's support for typical research methods -- such as surveys, interviews, and interventions -- as well as for investigating the patterns, causes, and underlying mechanisms of social issues. The alignment between AgentSociety's outcomes and real-world experimental results not only demonstrates its ability to capture human behaviors and their underlying mechanisms, but also underscores its potential as an important platform for social scientists and policymakers.

cs.SI↗

arXiv2026

$L^p$-estimates for the wave equation with partial inverse-square potentials

This paper investigates $L^p$-estimates for solutions to the wave equation perturbed by a scaling-critical partial inverse-square potential. We study a model in which the singularity of the potential appears only in a subset of the variables, corresponding to the Schrödinger operator $\mathcal{H}_a = -Δ_x - Δ_y + a/|x|^2$ on $\mathbb{R}^{2+n}$. Using spectral analysis, we establish the $L^p$-boundedness of the wave propagator $(1+\sqrt{\mathcal{H}_a})^{-γ} e^{it\sqrt{\mathcal{H}_a}}$ for a range of exponents $γ$ and $p$ satisfying $|1/p -1/2| < γ/(n+1)$. The key ingredients are the spectral measure kernel of the partial inverse-square operator $\mathcal{H}_a$ and the complex interpolation argument.

math.AP↗

arXiv2026

Routing-based technique for defect mitigation in quantum error correction

As quantum chips scale up for large-scale computation, hardware defects become inevitable and must be carefully addressed. In this work, we introduce Halma, a defect mitigation technique empowered by an expanded native gate set that incorporates the iSWAP gate alongside the conventional CNOT gate. Halma emerges as a supplementary technique within the defect mitigation toolbox, offering effective mitigation of ancilla qubit defects encountered during surface code stabilizer measurements while maintaining compatibility with existing superstabilizer-based methodologies. Halma introduces zero reduction in the spacelike distance of the code without further sacrifice to the timelike distance. Numerical simulation suggests that in comparison to previous methods, Halma could provide an order of magnitude improvement in the average logical error rate under realistic experimental settings, leading to a $\sim3\times$ reduction in the footprint of a teraquop. These results clearly demonstrate the capability of Halma in easing the near-term realization of fault-tolerant quantum computing on hardware with fabrication defects, and exemplifies how leveraging intrinsic hardware capabilities can enhance quantum hardware performance.

quant-ph↗

arXiv2026

Equi-RO: A 4D mmWave Radar Odometry via Equivariant Networks

Autonomous vehicles and robots rely on accurate odometry estimation in GPS-denied environments. While LiDARs and cameras struggle under extreme weather, 4D mmWave radar emerges as a robust alternative with all-weather operability and velocity measurement. In this paper, we introduce Equi-RO, an equivariant network-based framework for 4D radar odometry. Our algorithm pre-processes Doppler velocity into invariant node and edge features in the graph, and employs separate networks for equivariant and invariant feature processing. A graph-based architecture enhances feature aggregation in sparse radar data, improving inter-frame correspondence. Experiments on an open-source dataset and a self-collected dataset show Equi-RO outperforms state-of-the-art algorithms in accuracy and robustness. Overall, our method achieves 10.7% and 13.4% relative improvements in translation and rotation accuracy, respectively, compared to the best baseline on the open-source dataset.

cs.RO↗

arXiv2026

Learning to Decode in Parallel: Self-Coordinating Neural Network for Real-Time Quantum Error Correction

Fast, reliable decoders are pivotal components for enabling fault-tolerant quantum computation (FTQC). Neural network decoders like AlphaQubit have demonstrated potential, achieving higher accuracy than traditional human-designed decoding algorithms. However, existing implementations of neural network decoders lack the parallelism required to decode the syndrome stream generated by a superconducting logical qubit in real time. Moreover, integrating AlphaQubit with sliding window-based parallel decoding schemes presents non-trivial challenges: AlphaQubit is trained solely to output a single bit corresponding to the global logical correction for an entire memory experiment, rather than local physical corrections that can be easily integrated. We address this issue by training a recurrent, transformer-based neural network specifically tailored for parallel window decoding. While it still outputs a single bit, we derive training labels from a consistent set of local corrections and train on various types of decoding windows simultaneously. This approach enables the network to self-coordinate across neighboring windows, facilitating high-accuracy parallel decoding of arbitrarily long memory experiments. As a result, we overcome the throughput bottleneck that previously precluded the use of AlphaQubit-type decoders in FTQC. Our work presents the first scalable, neural-network-based parallel decoding framework that simultaneously achieves SOTA accuracy and the stringent throughput required for real-time quantum error correction. Using an end-to-end experimental workflow, we benchmark our decoder on the Zuchongzhi 3.2 superconducting quantum processor on surface codes with distances up to 7, demonstrating its superior accuracy. Moreover, we demonstrate that, using our approach, a single TPU v6e is capable of decoding surface codes with distances up to 25 within 1us per decoding round.

quant-ph↗

arXiv2025

Lifting Biomolecular Data Acquisition

One strategy to scale up ML-driven science is to increase wet lab experiments' information density. We present a method based on a neural extension of compressed sensing to function space. We measure the activity of multiple different molecules simultaneously, rather than individually. Then, we deconvolute the molecule-activity map during model training. Co-design of wet lab experiments and learning algorithms provably leads to orders-of-magnitude gains in information density. We demonstrate on antibodies and cell therapies.

q-bio.BM↗

arXiv2025

Quantum Design Automation: Foundations, Challenges, and the Road Ahead

Quantum computing is transitioning from laboratory research to industrial deployment, yet significant challenges persist: system scalability and performance, fabrication yields, and the advancement of algorithms and applications. We emphasize that in building quantum computers -- spanning quantum chips, system integration, instruction sets, algorithms, and middleware such as quantum error correction schemes -- design is everywhere. In this paper, we advocate for a holistic design perspective in quantum computing, a perspective we argue is pivotal to unlocking innovative co-design opportunities and addressing the aforementioned key challenges. To equip readers with sufficient background for exploring co-optimization opportunities, we detail how interconnected computational methods and tools collaborate to enable end-to-end quantum computer design. This coverage encompasses critical stages -- such as chip layout design automation, high-fidelity system-level simulation, Hamiltonian derivation for quantum system modeling, control pulse simulation, decoherence analysis, and physical verification and testing -- followed by quantum instruction set design. We then proceed to quantum system and software development, including quantum circuit synthesis, quantum error correction and fault tolerance, and logic verification and testing. Through these discussions, we illustrate with concrete examples -- including co-optimizing quantum instruction sets with algorithmic considerations, customizing error correction circuits to hardware-specific constraints, and streamlining quantum chip design through tailored code design, among others. We hope that the detailed end-to-end design workflow as well as these examples will foster dialogue between the hardware and software communities, ultimately facilitating the translation of meaningful research findings into future quantum hardware implementations.

quant-ph↗

arXiv2025

Learning Neural Decoding with Parallelism and Self-Coordination for Quantum Error Correction

Fast, reliable decoders are pivotal components for enabling fault-tolerant quantum computation. Neural network decoders like AlphaQubit have demonstrated significant potential, achieving higher accuracy than traditional human-designed decoding algorithms. However, existing implementations of neural network decoders lack the parallelism required to decode the syndrome stream generated by a superconducting logical qubit in real time. Moreover, integrating AlphaQubit with sliding window-based parallel decoding schemes presents non-trivial challenges: AlphaQubit is trained solely to output a single bit corresponding to the global logical correction for an entire memory experiment, rather than local physical corrections that can be easily integrated. We address this issue by training a recurrent, transformer-based neural network specifically tailored for sliding-window decoding. While our network still outputs a single bit per window, we derive training labels from a consistent set of local corrections and train on various types of decoding windows simultaneously. This approach enables the network to self-coordinate across neighboring windows, facilitating high-accuracy parallel decoding of arbitrarily long memory experiments. As a result, we resolve the throughput limitation that previously prohibited the application of AlphaQubit-type decoders in fault-tolerant quantum computation.

quant-ph↗