arXiv subjects

Haoran Ma

Publications and source records attributed to Haoran Ma.

At least 19 recordsLinked to original sources

arXiv2026

Toward Executable Multi-framework Front-end Code Generation with Self-Correction

Generating codes from UI screenshots has recently benefited from MLLMs, yet existing methods centered on HTML targets and transfer poorly to multi-frameworks such as React, Vue, and Angular. However, unlike HTML, which does not entail compilation issues, generating executable code in multi-framework settings is more challenging due to framework-specific differences in syntax and failure modes, as well as the need to maintain global consistency across files. To address this problem, we study executable multi-framework front-end code generation and propose MESCoder, a tree-structured self-corrective generation framework. Our method first extracts a structural scaffold from the screenshot and then constructs sub-trees and assembly relations step by step, transforming large project generation into a sequence of localized decisions. On top of this scaffold, we introduce a unified Project Agent that continuously revises its own partial project under runtime sandbox feedback and dynamically chooses between local repair and cluster level repair. For training, we adopt a two-stage strategy that first initializes the policy offline to learn basic multi-framework generation and repair priors, and then refines it with online reinforcement learning to optimize long horizon executability and final page quality. The resulting framework models multi-framework executability as a conditioned project state transition process and improves the stability and quality of complex front end generation through self correction. Experiments show that our method improves executability, while narrowing the visual-fidelity gap to HTML-based generation. Code is available at https://github.com/leo-lilinxiao/MESCoder.

cs.LG↗

arXiv2026

Phase-wise MLLM Tuning for Multi-framework WebUI Code Generation

Multimodal large language models (MLLMs) achieve strong performance in translating WebUI screenshots into HTML/CSS. However, when they faced multiple frameworks (React/Vue/Angular), they often suffer from negative transfer that leads to compilation failures. These failures commonly stem from mixing framework specific syntax and violating framework constraints. This paper studies supervised fine-tuning for multi-framework WebUI code generation and proposes a phase-wise MLLM tuning method driven by compatibility and heterogeneity signals across frameworks. We first apply a normalization mapping that preserves each framework's native syntax while suppressing noise from variable naming and literal content. We then estimate cross framework heterogeneity and compatibility offline, and automatically derive training groups and their ordering by maximizing a phased objective. Finally, we perform phase-wise adapter tuning to reduce cross framework interference. Experiments on multiple benchmarks show that our proposed method improves both generation quality and compilation success rate compared with baselines.

cs.SE↗

arXiv2026

MulFCoder: Framework-conditioned Multi-agent for MLLM-based Multi-framework Front-end Code Generation

Generating runnable front-end code from UI screenshots is a long-standing goal in automated software engineering. Existing MLLM-based methods predominantly focused on HTML/CSS, leaving multi-framework generation for React/Vue/Angular underexplored. Naively modifying prompts leads to substantial performance gaps across multi-framework and highly framework-specific error modes. To address this, we propose MulFCoder, a framework-conditioned multi-agent method that explicitly encodes framework constraints to bring multi-framework differences into a decidable rule space. MulFCoder orchestrates four agents: Grounder constructs an ElementTable, ContentTable, and macro-layout regions from detected UI elements; Planner builds a DOM-like hierarchical layout tree, produces a task schedule, and derives a framework-specific file Contract; Writer generates structured file writes or patches within a restricted edit window; Judger enforces lightweight, framework-conditioned constraints to accept or reject updates and trigger bounded repairs, preventing drift and deadlocks without expensive builds. Experiments demonstrate that MulFCoder substantially improves compilation success rate and reduces framework-specific errors, with particularly pronounced gains on Angular.

cs.SE↗

arXiv2026

Photoelectrons resolve the phase-dependent variance of bright squeezed vacuum

The variance of bright squeezed vacuum (BSV) is not only strong but also phase dependent, yet recent strong-field experiments with BSV have recorded only its magnitude. A coherent field of the same color provides the phase reference that resolves this phase-dependent noise. Strong-field ionization of xenon by this light shows that the photoelectron yield follows the BSV quadrature along the coherent field, and that the nonlinearity of ionization amplifies the quadrature noise many times over in the electron yield. The noise is moreover distributed unevenly over the photoelectron momentum distribution, differing by an order of magnitude between ionization channels. Strong-field ionization thus offers a phase-sensitive, channel-resolved detector of quantum light, while the quadrature noise of quantum light offers a probe of the intensity dependence of individual ionization pathways.

physics.atom-ph↗

arXiv2026

UNWRAP: Single-shot phase analysis of bright squeezed vacuum by Newton-ring interferometry

The mean field of pulsed bright squeezed vacuum (BSV) vanishes, but the variance can be large, and is strongly phase-dependent. We employ ultrafast Newton-ring wavefront retrieval and analysis of phase (UNWRAP) to extract the field amplitude and phase profile of each BSV pulse, by Fourier analysis of the interference rings between the BSV and a coherent reference with different phase-front curvature. The concentric ring interference fringes are robust against pointing mismatch and drift, thereby simplifying the analysis. The measured distribution has two phase peaks separated by π radians. The rings sweep the delay between the two pulses, so a single frame can capture the temporal field profile of the BSV, even when it is composed of several temporal modes.

physics.optics↗

arXiv2026

Solving Every Step Is Not Enough: Milestone Oracles Reveal a Composition Gap in LLM Math Reasoning

Large language models (LLMs) can solve every intermediate step of a multi-step math problem on its own and still fail the full problem, even when given a roadmap of the steps and all of their answers. We introduce OracleLadder, a diagnostic evaluation that locates where LLM math reasoning fails by giving the model increasing levels of oracle help. For each problem, a teacher model writes a fixed roadmap of intermediate sub-goals (milestones), and a deterministic symbolic verifier grades every answer. Testing the model with no help, with the roadmap, with the roadmap plus the milestone answers, and on each milestone alone sorts each failure into one of five reasoning gaps. On 354 NuminaMath problems and six models from 8B to 671B parameters (Qwen3, gpt-oss, Llama 3.3, DeepSeek-V3.1), the largest gap for every model is the composition gap, a stricter form of the compositionality gap. It covers 33-48% of problems, and 24-37% after removing problems that an LLM review flags as grading errors. Accuracy and milestone-help recovery rank the two strongest models differently, and two RLVR runs with similar accuracy gains move problems differently. The roadmap effect replicates on MATH500 and AIME 2024/25, per-problem recovery agrees for 83-87% of problems under an independent second teacher, and the help ladder carries over to code generation. We release the data, roadmaps, prompts, and code at https://github.com/slark-prime/OracleLadder.

cs.CL↗

arXiv2026

Electron spectra from strong-field enhanced ionization in heavy water

Strong-field enhanced ionization (EI) is a phenomenon in which stretching of interatomic bonds into a distorted molecular geometry leads to an increase in the tunneling ionization rate driven by a strong field. While the nuclear dynamics leading to EI are well established, the electron dynamics of the ionization step remain largely unexplored. Isolating the momentum distribution of the electrons involved in EI is critical to fully characterizing the phenomenon. We have measured the electron momentum distribution associated with EI in triple ionization of D$_2$O using 6-fs pulse pairs together with full fragment momentum imaging and electron-ion correlation methods. We find that the EI electron momentum distribution differs substantially from that of standard strong-field tunneling from molecules, exhibiting an increased yield of electrons with large momentum in the direction of the laser polarization, and a change from the expected Gaussian distribution. These observations indicate that the instantaneous EI tunneling rate is maximized at a critical value of the laser electric field, rather than at the peak of an optical cycle. This finding distinguishes EI from Keldysh tunneling rate predictions, where tunneling rate increases monotonically with field strength. These pronounced differences between EI and non-EI electron spectra are critical tests of models of enhanced ionization and suggest a route towards control of the sub-cycle timing of electron emission.

physics.atom-ph↗

arXiv2026

EMERGE-Policy: A Robot Mind Emerges Beyond a Single Policy

A robot's effective ``mind'' need not reside in a single policy. It can emerge when specialized components perceive, reason, predict, act, verify, and remember within a shared orchestration process. EMERGE-Policy turns this perspective into a graph-structured agentic framework that coordinates both capability invocation and information exchange. A Main Agent retains task-level state within an active context window, while role-specific Sub Agents process perception, execution monitoring, verification, and memory consolidation in isolated contexts and return structured, task-relevant evidence. Role-specific contexts control information load by exposing only decision-relevant evidence to the Main Agent, while the functional Skill interface composes heterogeneous backends as Operational, Imagination, and Evaluation Skills. Criterion-grounded verification, textual failure diagnosis, and Branch Stack recovery provide localized correction, with token-aware external memory preserving task-relevant state. Together, their closed-loop interaction realizes the system-level policy captured by the name EMERGE-Policy. Without additional fine-tuning, we achieved outstanding performance on several public benchmark that have had a wide-reaching impact, and conducted a series of real robot experiments. These system-level results suggest that through the division of different functional sub-tasks among multiple agents and their concurrent collaboration, as well as the technical paradigm where the model is regarded as a skill and called within the framework, EMERGE-Policy can extend the robust robot policies beyond isolated runs.

cs.RO↗

arXiv2026

Tracking molecular hydrogen formation from ionized water in real time

Removing an electron from a water molecule can drive its two hydrogen atoms to pair up and depart as molecular hydrogen. However, even for this elementary reaction, the route from start to finish has remained hidden because measurements have yet to follow the electronic and nuclear motion simultaneously. Combining correlated photoelectron and ion imaging, few-femtosecond pump--probe measurements, and nonadiabatic simulations, we track the complete pathway in isolated heavy water (D$_2$O) molecules. The reaction takes an indirect route and dissociates along three distinct pathways (direct, roaming, and delayed) with formation times of about 34 and 72 femtoseconds for the direct and delayed branches. Yet bond formation requires the molecule to first break its own symmetry. Only random asymmetric motion enables the electronic-state switch at a conical intersection, joining the two hydrogen atoms before the oxygen--hydrogen bond breaks. These results establish a time-resolved picture of molecular hydrogen formation from water and provide a general strategy for linking electronic excitation to chemical outcomes in settings from radiation damage to hydrogen production.

physics.chem-ph↗

arXiv2026

MRUF: Multi-granularity Routing with Uncertainty-Aware Fusion for Robust Multimodal Sentiment Analysis

Multimodal sentiment analysis relies on language, visual, and acoustic cues, but utterance-level modality quality may vary due to occlusion, background noise, motion blur, or imperfect transcripts, causing conventional fusion to over-trust unreliable modalities. We propose MRUF, a reliability-aware fusion method that combines multi-granularity routing with uncertainty-aware calibration. MRUF summarizes sentiment-relevant representations, performs subspace- and modality-level routing, and supervises modality routing with leave-one-out error increases to estimate utterance-level modality importance. It further predicts modality-wise uncertainty and refines modality gates through inverse-variance reweighting, while modality-invariant contrastive alignment stabilizes the shared representation space. Experiments on CMU-MOSI and CMU-MOSEI under aligned and unaligned settings show consistent improvements over strong baselines, and mechanism analysis verifies that modalities with higher predicted uncertainty receive lower fusion weights.

cs.AI↗

arXiv2026

Uncertainty-Aware Prediction of Lung Tumor Growth from Sparse Longitudinal CT Data via Bayesian Physics-Informed Neural Networks

This work studies lung tumor growth prediction from sparse and irregular longitudinal computed tomography (CT) observations with measurement variability. A Bayesian physics-informed neural network is developed by combining Gompertz growth dynamics with low-dimensional Bayesian inference in the log-volume domain. The framework employs a two-stage inference strategy combining maximum a posteriori (MAP) estimation and Hamiltonian Monte Carlo (HMC) sampling to estimate posterior predictive distributions and uncertainty intervals. The method was evaluated on longitudinal data from the National Lung Screening Trial (30 patients). Results show that the model captures heterogeneous tumor growth patterns while maintaining reasonable prediction accuracy under limited observations. Compared with deterministic modeling approaches, the proposed approach additionally provides calibrated uncertainty estimates. The inferred posterior parameter correlations were consistent with expected biological growth behavior. The proposed framework achieved a cohort-level log-space RMSE of approximately 0.20 together with well-calibrated 95% credible interval coverage across 30 patients. These findings suggest that Bayesian physics-informed modeling may be useful for uncertainty-aware tumor growth assessment when only limited longitudinal follow-up scans are available.

cs.LG↗

arXiv2026

D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Drafting

Speculative decoding accelerates LLM inference by having a small drafter propose tokens that a larger target model verifies in parallel. Recent diffusion-based parallel drafters such as DFlash predict the full B-token block in one forward pass, enabling deeper drafters and longer accepted blocks. However, existing multi-token drafter objectives often use fixed position-dependent weighting schedules, such as head-dependent weights or block-position decays, which do not adapt as the positions limiting acceptance change during training. To address this, we derive per-position training weights from a differentiable surrogate of expected accepted draft length, matching the weight of each position to its log-probability gradient contribution. The resulting loss, D-PACE (Dynamic Position-Aware Cross-Entropy), shifts training signal toward positions that currently limit acceptance as the drafter improves. Across six benchmarks, two Qwen3-4B draft depths, two decoding temperatures, and two additional target models, D-PACE consistently improves both wall-clock speedup and average emitted length, with 2.3\% measured training-time overhead and no changes to the drafter architecture or inference procedure.

cs.LG↗

arXiv2026

Photonic-Implemented Efficient Deep Quantum Neural Network via Virtual-Driven Hilbert Space Expansion

The growing computational demands of classical neural networks have intensified the search for energy-efficient and powerful computational alternatives. Quantum neural networks (QNNs) implemented on integrated photonic platforms offer a compelling avenue, offering exceptional computational power enhancements, with inherent programmability and scalability of integrated architectures. A critical challenge, however, is implementing the fundamental non-unitary and nonlinear activation function of QNNs within a linear quantum photonic system. Existing strategies, such as the adding ancillary qubits and measurement-based feedback or forward are constrained by high qubit resource costs, overhead devices, and poor cascadability. Here, we propose a novel deep photonic QNN with an expanded computational Hilbert space via input replication and mode expansion, which enables the realization of effective non-unitary and nonlinear activation on a linear programmable quantum photonic chip. This approach eliminates the need for physical ancillary qubits, measurement-induced qubit consumption and the measurement device burden, thereby significantly reduce resource costs. The fabricated chip integrates four high-quality entanglement sources and a programmable high-dimensional interferometric network, enabling a two-hidden-layer QNN that exhibits dimension-enhanced expressivity over the existing QNN architectures. We demonstrate its capabilities across diverse tasks, including nonlinear classification, image generation, and quantum Gibbs state preparation. This work establishes a scalable and efficient architecture toward practical quantum deep learning systems capable of tackling problems beyond the reach of classical computation.

quant-ph↗

arXiv2026

Transformer-based prediction of two-dimensional material electronic properties under elastic strain engineering

Strain engineering provides a powerful route for tuning the electronic properties of two-dimensional (2D) materials, but exploring the full multidimensional strain space with density functional theory (DFT) is computationally prohibitive due to the nonlinear coupling between normal and shear components. In this work, we introduce a Transformer-based, multi-target surrogate model framework that achieves DFT-level bandgap prediction accuracy, reaching a mean absolute error of 0.0103 eV while retaining full interpretability through attention-weight analysis. The learned self-attention map consistently identifies shear strain as the interaction center that influences both bandgap and phonon stability, an insight not readily captured by classical feature-importance metrics. This work establishes attention-based architectures as physically interpretable surrogate models for multi-property prediction, offering a generalizable strategy for accelerating deep elastic strain engineering in materials informatics.

cond-mat.mtrl-sci↗

arXiv2026

DiffVP: Differential Visual Semantic Prompting for LLM-Based CT Report Generation

While large language models (LLMs) have advanced CT report generation, existing methods typically encode 3D volumes holistically, failing to distinguish informative cues from redundant anatomical background. Inspired by radiological cognitive subtraction, we propose Differential Visual Prompting (DiffVP), which conditions report generation on explicit, high-level semantic scan-to-reference differences rather than solely on absolute visual features. DiffVP employs a hierarchical difference extractor to capture complementary global and local semantic discrepancies into a shared latent space, along with a difference-to-prompt generator that transforms these signals into learnable visual prefix tokens for LLM conditioning. These difference prompts serve as structured conditioning signals that implicitly suppress invariant anatomy while amplifying diagnostically relevant visual evidence, thereby facilitating accurate report generation without explicit lesion localization. On two large-scale benchmarks, DiffVP consistently outperforms prior methods, improving the average BLEU-1-4 by +10.98 and +4.36, respectively, and further boosts clinical efficacy on RadGenome-ChestCT (F1 score 0.421). All codes will be released at https://github.com/ArielTYH/DiffVP/.

cs.CV↗

arXiv2026

SRFS: Parallel Processing Fault-tolerant ROS2-based Flight Software for the Space Ranger CubeSat

Traditional Real-Time Operating Systems (RTOS) often suffer from limited parallel performance, whereas thread monitoring in Linux-based systems remains challenging. To overcome these limitations, this paper presents a satellite flight software system design based on the Robot Operating System (ROS), which utilizes its reliable built-in publish-subscribe messaging mechanism to facilitate inter-application communication. In response to the complex functional demands of modern small satellites, the proposed design integrates both hardware and software architectures, along with system scheduling and error-correction strategies. This integration supports efficient parallel data processing, enhances system reliability, and shortens the development cycle through code reuse. The system was rigorously evaluated through comprehensive tests covering time delay, system management, fault tolerance, and maintenance procedures. Experimental results confirm the system's effectiveness in telemetry, remote control, integration of new features, and autonomous error recovery. The findings underscore the high reliability and maintainability of the ROS-based satellite flight software, offering a valuable reference for the rapid development of high-performance small satellite systems.

eess.SY↗

arXiv2025

Cyber Racing Coach: A Haptic Shared Control Framework for Teaching Advanced Driving Skills

This study introduces a haptic shared control framework designed to teach human drivers advanced driving skills. In this context, shared control refers to a driving mode where the human driver collaborates with an autonomous driving system to control the steering of a vehicle simultaneously. Advanced driving skills are those necessary to safely push the vehicle to its handling limits in high-performance driving such as racing and emergency obstacle avoidance. Previous research has demonstrated the performance and safety benefits of shared control schemes using both subjective and objective evaluations. However, these schemes have not been assessed for their impact on skill acquisition on complex and demanding tasks. Prior research on long-term skill acquisition either applies haptic shared control to simple tasks or employs other feedback methods like visual and auditory aids. To bridge this gap, this study creates a cyber racing coach framework based on the haptic shared control paradigm and evaluates its performance in helping human drivers acquire high-performance driving skills. The framework introduces (1) an autonomous driving system that is capable of cooperating with humans in a highly performant driving scenario; and (2) a haptic shared control mechanism along with a fading scheme to gradually reduce the steering assistance from autonomy based on the human driver's performance during training. Two benchmarks are considered: self-learning (no assistance) and full assistance during training. Results from a human subject study indicate that the proposed framework helps human drivers develop superior racing skills compared to the benchmarks, resulting in better performance and consistency.

cs.RO↗

arXiv2025

ConServe: Fine-Grained GPU Harvesting for LLM Online and Offline Co-Serving

Large language model (LLM) serving demands low latency and high throughput, but high load variability makes it challenging to achieve high GPU utilization. In this paper, we identify a synergetic but overlooked opportunity to co-serve latency-critical online requests alongside latency-tolerant offline tasks such as model benchmarking. While promising, existing serving systems fail to co-serve them efficiently, as their coarse-grained resource management at the request or iteration level cannot harvest millisecond-level GPU idle cycles without introducing interference that violates online latency objectives. ConServe is a new LLM co-serving system that achieves high throughput and strong online latency guarantees by managing resources at finer granularities. ConServe introduces three techniques: (1) a latency-aware token-level scheduler that precisely sizes offline batches and tokens to fit within online latency objectives; (2) sub-iteration, layer-wise preemption that allows offline tasks to yield to online load spikes; and (3) incremental KV cache management that enables preempting and resuming offline requests at near-zero cost. Evaluations with Llama-3.1 and Qwen-2.5 models on real-world workloads show that ConServe delivers an average of 2.2$\times$ higher throughput and reduces online serving tail latency by 2.9$\times$ on average compared to state-of-the-art systems.

cs.DC↗