arXiv subjects

Pengfei Zhang

Publications and source records attributed to Pengfei Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Measuring High-Energy Cosmic Particles with the SKA

The origin of high-energy cosmic rays remain one of astrophysics' greatest unsolved mysteries. SKA-Low will be able to measure air showers initiated by cosmic rays with unprecedented precision in the PeV - EeV energy range, covering the critical transition region between Galactic and extragalactic sources. SKA-Low's densely instrumented core and broad bandwidth will allow for measurements of individual air showers with a level of detail unmatched by any existing or planned detector. The depth of shower maximum, the primary mass-sensitive observable, will be reconstructed with a resolution of better than 8~g/cm$^2$, a significant improvement over existing methods. Additionally, new reconstruction methods are expected to enable full air shower reconstruction across a wide energy range, down to PeV levels. At these energies, efficient photon/hadron separation may offer an opportunity to measure PeV gamma-ray air showers. Furthermore, SKA-Low opens a window into studying high-energy hadronic interactions, including via the unique channel of anomalous air showers. This combination of measurements provides a unique opportunity to investigate the origins and physics of high-energy cosmic rays. A dedicated particle detector array will provide triggered readout of raw antenna-level voltage buffers, enabling fully commensal cosmic-ray observations alongside regular operations. We outline our science case and discuss the observational strategy, signal properties and detector design underpinning these measurements. We also summarize the accompanying book chapters, which address composition measurements in the Galactic-to-extragalactic transition region, next-generation interferometric reconstruction techniques, hadronic interaction physics through anomalous air showers, the prospects for detecting PeV gamma-rays from Galactic sources, and the related project of imaging lightning using SKA-Low.

astro-ph.HE↗

arXiv2026

Using SKA-Low to Detect PeV Gamma-rays from Galactic Sources

Detecting so called PeVatrons is considered one of the prime goals of $γ$-ray astronomy. PeVatrons are astrophysical objects in the Galaxy that are sources of cosmic rays exceeding PeV ($10^{15}$ eV) energies, the highest in our Galaxy. Their nature is unknown as of now, with some candidates reaching barely above PeV energies just having been identified. Serendipitously, the energy threshold of air shower detection using radio emission, has been proven at 50 PeV. There is a case to be made that SKA-Low with its unprecedented number of antennas, can reach lower in energy, while the size of the core is sufficiently large provide a significant effective area to measure PeV fluxes. While this promises a novel angle towards understanding the cosmic ray accelerators in our Galaxy, it also would be the first detection of $γ$-ray air showers using radio emission.

astro-ph.HE↗

arXiv2026

Interferometric Analysis of Air-shower Radio Emission in the Near Field with an Information Field Theory Approach

Current reconstruction techniques for air-shower radio emission generated by cosmic rays have shown great success, having been applied to several radio detectors over the last decade. Nevertheless, they are limited by their high computational cost, simplified approximations, and signal information used for reconstruction. As such, advanced analyses are required to not only be able to perform a holistic reconstruction of all parameters, but also to conduct near-field interferometry of the air shower. This can be achieved through Information Field Theory (IFT), an imaging reconstruction framework based on Bayesian inference that can extract all available information within the signal to infer distributions of field-like quantities. In this chapter, we highlight current novel approaches that use IFT for air shower reconstruction, and the potential of their applicability towards SKA-Low.

astro-ph.IM↗

arXiv2026

From Sounds to Scenes: A Benchmark for Evaluating Context-Aware Auditory Scene Understanding in Large Audio Language Models

Recent Large Audio Language Models (LALMs) have achieved remarkable progress in audio perceptual tasks across individual acoustic layers, including speech, sound, and music. However, existing benchmarks predominantly evaluate these layers in isolation, overlooking the complex contextual relationships that arise when multiple acoustic sources co-occur in real-world auditory scenes. Real-world auditory interpretation requires Context-Aware Auditory Scene Understanding (CASU): the ability to comprehend the holistic scene by integrating sound layers. To evaluate this capability, we introduce the CASU benchmark, which assesses whether Audio LLMs can interpret auditory scenes composed of speech, acoustic events (e.g., announcements), and background environments (e.g., traffic), and reason about the logical relationships between these layers. We propose a scalable pipeline for constructing time-accurate, semi-synthetic audio streams by composing real-world scene sounds with synthetic speech. Building on this data, we design four tasks that probe scene understanding: contextual question answering, entity extraction from the scene, speaker role inference, and counterfactual reasoning where scene is manipulated. Experiments across multiple LALMs demonstrate that effective auditory scene understanding requires integration over all auditory layers, rather than reliance on speech or sound alone, underscoring the necessity of CASU for advancing complex audio understanding in LALMs.

cs.SD↗

arXiv2026

DreamX-World 1.0: A General-Purpose Interactive World Model

DreamX-World 1.0 is a general-purpose interactive text/image-to-video world model for controllable long-horizon generation. It supports camera navigation, revisits to previously observed regions, and promptable events across photorealistic, game-style, and stylized domains. Our data engine combines camera-accurate Unreal Engine rendering, action-rich gameplay recordings, and real-world videos with recovered camera geometry. For camera control, we introduce E-PRoPE, a lightweight variant of projective positional encoding that retains PRoPE's projective camera geometry while applying camera-aware attention to spatially reduced tokens. We convert a bidirectional video generator into a few-step autoregressive world model using causal forcing, DMD-style distillation, and long-rollout training. Training on self-generated long-horizon contexts exposes the model to its own generated history and reduces the style and color drift that accumulates across autoregressive chunks. Memory-Conditioned Scene Persistence retrieves earlier views through camera-geometry-based retrieval, while residual recycling makes the conditioning path less sensitive to imperfect memory latents. Event Instruction Tuning adds composable event control, and reinforcement learning alignment recovers camera control and visual quality after distillation. With mixed-precision DiT execution, residual reuse, 75\%-pruned VAE decoding, and asynchronous pipeline parallelism, DreamX-World 1.0 reaches up to 16\,FPS on eight RTX\,5090 GPUs. On our 5-second basic evaluation, DreamX-World 1.0 achieves a camera-control score of 73.75 and an overall score of 84.76, outperforming HY-WorldPlay 1.5 and LingBot-World in overall score, which achieve 80.79 and 80.45, respectively.

cs.CV↗

arXiv2026

Post-Selection Probability and Fidelity of Bidirectional Teleportation

Understanding the scrambling of quantum information is central to many areas of quantum physics, including quantum thermalization, entanglement growth, and quantum information processing. Insights from these studies have, in turn, inspired the development of novel quantum protocols and algorithms. Recently, a bidirectional teleportation protocol was proposed to implement a digital SWAP operation between qubits by leveraging chaotic Hamiltonian evolution combined with measurement and post-selection. In this work, we provide a comprehensive study of two central quantities that characterize the protocol, the post-selection probability and the fidelity, taking into account possible errors in time-reversed dynamics. We show that these quantities can be expressed in terms of standard diagnostics in quantum dynamics, including the Loschmidt echo and its subsystem variant. The results unveil (1) the initial-state dependence of the fidelity and (2) the stability of the post-selection probability in integrable models. Our findings offer practical guidance for the implementation of the protocol on realistic quantum devices.

quant-ph↗

arXiv2026

AP-GRPO: Anchor-Gated Phonetic Alignment with Policy Optimization for Pathological Speech Reconstruction

Pathological speech from patients with neurodegenerative and neuromotor disorders is often acoustically distorted and linguistically fragmented, making pathological speech reconstruction necessary to recover intended textual content from distorted and incomplete speech recordings. Crucially, such recordings are rarely uniformly degraded: some words or short phrases remain reliable and can serve as audible anchors for reconstructing the corrupted surrounding content. We introduce Anchor-gated Phonetic Group Relative Policy Optimization (AP-GRPO), a GRPO framework with phonetic reward that aligns speech language models (SLMs) through audible-anchor preservation and inter-anchor phonetic compatibility to the original speech signal. AP-GRPO consists of: (i) an anchor-gated reward that matches reliable audible anchors in clear regions; and (ii) an inter-anchor phonetic alignment reward that evaluates whether recovered contents are phonetically supported by the corresponding corrupted inter-anchor speech span. Across four disease conditions, AP-GRPO improves faithful speech reconstruction, and the learned anchor constraint automatically adapts to each condition and thus reveals interpretable disease-specific profiles: conditions with severe articulatory degradation require stronger anchor enforcement, whereas milder impairment or linguistically impaired conditions rely more on phonetic alignment for inter-anchor recovery.

cs.SD↗

arXiv2026

Enhancing Many-Body Chaos via Entropy Injection from Environment

In closed quantum systems, local information spreads throughout the entire system and becomes highly complex under unitary evolution. In contrast, when the system is embedded in an environment, system-environment coupling can transfer information from the system into the environment, thereby reducing the rate of complexity growth within the system. This leads to the environment-induced scrambling transition established in previous works. In this work, we identify entropy injection from the environment as a different physical process that instead enhances many-body chaos. Our setup consists of coupling a system that is already in equilibrium with one environment to another environment, which serves as an entropy reservoir and drives the system into a non-equilibrium state. When entropy flows into the system through either heat transfer or particle transfer, the effective Hilbert space explored by the system enlarges, a mechanism that can enhance many-body chaos. We explicitly demonstrate this idea by constructing a solvable complex Brownian SYK model, in which both the relaxation toward the steady state and the steady-state quantum Lyapunov exponent can be computed analytically. Our results provide a controllable mechanism for tuning quantum scrambling through entropy flow in quantum many-body systems coupled to environments.

quant-ph↗

arXiv2026

5% > 100%: Flatness Preference is All You Need for Multimodal Parameter-Efficient Fine-Tuning

Parameter-Efficient Fine-Tuning (PEFT) methods provide a streamlined and efficient tool for adapting large models to domain-specific multimodal downstream tasks. Although these methods proved their tangible effects in practice, their principal aspects remain under-explored. Therefore we remain curious about the underlying generalization mechanisms in various PEFT methods and how they can be further enhanced. In this paper, we reveal the flatness preference widely present in various PEFTs, where a small fraction of sharp dimensions dominates the generalization of PEFT. This finding suggests an appealing possibility: we may be satisfied with a better generalization by merely attending to this small fraction of sharp dimensions instead of all of them. Furthermore, we propose Flatness Preference Optimization (FlatPO) to flatten these key sharpness dimensions, leading various PEFTs toward better generalization. Extensive experiments demonstrate the effectiveness of our findings and the proposed method. Code is available at https://github.com/Can-Lin/FlatPO.

cs.CV↗

arXiv2026

Adaptive Loss Balancing for Noise-Robust GRPO in Generative Recommendation

Reinforcement learning (RL) presents a promising avenue for enhancing generative recommendation beyond supervised imitation, leveraging reward signals to guide policy improvement. However, its efficacy is critically contingent on the trustworthiness of the reward model for the samples it evaluates. In practice, production rankers, the widely adopted reward models, are trained on exposure-biased logs, leading to sample-dependent inaccuracies that violate this assumption. Our stratified analysis uncovers a consistent pattern: reward guidance is most beneficial when the policy exhibits uncertainty and the ranker can effectively discriminate the ground-truth item from rollout negatives. On other samples, the reward signal is either negligible or detrimental, highlighting the risk of uniform RL application. To address such an issue, we introduce AdaGRPO, a novel framework that treats reward-guided optimization as selective admission rather than uniform pressure. Training is anchored in supervised negative log-likelihood, while the GRPO objective is gated by a binary, per-sample clip determined by two rollout diagnostics: policy-side difficulty and reward discriminability. Instances failing either diagnostic default to pure supervision, ensuring stability and mitigating the amplification of noisy gradients. We validate AdaGRPO on a large-scale e-commerce dataset. At the best intermediate checkpoint, it elevates HR@10 from 11.01% to 12.18% while constraining hallucination below 0.22%, and maintains robustness at the final checkpoint (HR@10 11.63%, hallucination 0.27%), outperforming fixed NLL--GRPO mixtures across the retrieval--validity frontier. In production A/B tests, AdaGRPO achieves statistically significant gains in click-through rate and dwell time, confirming its practical utility.

cs.LG↗

arXiv2026

Can the Environment Speak for Itself? $T^{2}$-GRPO: A Turn-Trajectory Group Relative Policy Optimization for Caregiver Agents

Optimizing large language models (LLMs) for long-horizon caregiver agents requires balancing delayed task objectives with immediate environment dynamics, such as patient distress and resistance. In dementia care, this balance is especially difficult: trajectory level rewards are too sparse for turn level credit assignment, while external LLM-based evaluators are costly and can misread fragmented or indirect patient responses. To address this issue, we propose \textbf{T}urn-\textbf{T}rajectory \textbf{G}roup \textbf{R}elative \textbf{P}olicy \textbf{O}ptimization (\textbf{T$^{2}$-GRPO}), a framework that decouples caregiver RL into two normalized reward horizons and enforces safety through a binary hard veto. $T^2$-GRPO derives dense turn-level rewards directly from environment state transitions, measuring changes in patient distress and resistance from a frozen dementia patient simulator. These environment-grounded rewards are combined with trajectory-level evaluations through independent centered-rank normalization, which preserves heterogeneous reward signals and mitigates reward collapse. Extensive experiments on dementia caregivers show that T $^{2}$-GRPO outperforms competitive baselines, indicating a substantial improvement for emotionally sensitive caregiver scenarios that effectively handles immediate patient feedback, long-term care outcomes, and safety constraints.

cs.AI↗

arXiv2026

A superconducting surface-code processor with lattice-surgery logical operations

Fault-tolerant logical operations are fundamental for scalable quantum computation. Here, we report the experimental realization of lattice-surgery operations between a pair of distance-three surface-code logical qubits on a planar superconducting processor. During repeated syndrome extraction cycles, the logical qubits exhibit per-cycle error rates of $0.0365(2)$ and $0.0282(1)$, respectively, after leakage events are rejected. By leveraging joint initialization and lattice splitting, we deterministically prepare a logical Bell state, confirming genuine bipartite entanglement via the error-corrected logical state fidelity. We further execute a two-qubit Deutsch-Jozsa algorithm at the logical level to demonstrate algorithmic utility in a fault-tolerant framework. Finally, to achieve universal control, we implement magic-state injection and gate teleportation to realize continuous non-Clifford rotations about the logical $X$ axis. For the logical $R_{X}(π/4)$ gate, we achieve a logical gate fidelity of $0.943_{-9}^{+10}$ conditioned on the absence of detected errors. These results establish lattice surgery as a practical and versatile paradigm for logical computation in near-term surface-code architectures, representing a critical milestone toward scalable fault-tolerant quantum advantage in superconducting circuits.

quant-ph↗

arXiv2026

Ten Headache Specialists versus Artificial Intelligence for Clinical Literature Summarization: A Critical Evaluation and Comparison

Summarizing the latest medical literature to guide clinical decision-making is essential for evidence-based medicine and high-quality patient care. Yet clinicians face increasing challenges due to limited time with patients and a rapidly growing volume of published articles. Although retrieval-augmented large language models (LLMs) have shown promise in clinical summarization, human evaluations of their effectiveness in synthesizing broader scientific literature and direct comparisons to expert-written syntheses remain scarce. We constructed a RAG-based agentic AI framework using three state-of-the-art LLMs: Sonnet, GPT-4o, and Llama 3.1. A headache specialist created 13 questions, three for prompt optimization and ten for evaluation. Ten headache specialists across the United States and Canada each wrote a summary for one question, yielding four summaries per question (expert, Sonnet, GPT-4o, and Llama). The experts, blinded to authorship, critically evaluated the summaries, excluding the topic for which they wrote a summary, based on correctness, completeness, conciseness, and clinical utility, scoring each from 1 to 10 using standardized rubrics. They also ranked the summaries by preference and indicated whether they believed each summary was written by an expert or an LLM. Our study, comparing LLM- and expert-written literature summaries evaluated by headache specialists, showed that expert-written summaries were preferred, although experts sometimes found it challenging to distinguish between human- and AI-generated summaries. We also identified key expert-valued features beyond standard evaluation metrics that can guide future refinement of both human and AI literature summarization pipelines.

cs.AI↗

arXiv2026

EvoCut: Multi-Layer Evolution-Aware Visual Token Compression for Efficient Large Vision-Language Models

Large vision-language models (LVLMs) achieve strong performance on image and video understanding tasks, but their inference efficiency is constrained by the large number of visual tokens produced by vision encoders. Most existing visual token compression methods estimate token importance from attention scores or representation properties at specific layers, overlooking how visual tokens evolve across the vision encoder. Such layer-specific criteria may provide incomplete importance estimates and limit performance preservation after compression. To address this issue, we analyze layer-wise visual token evolution directions and observe that tokens form multiple group evolution directions across vision-encoder layers. Our analysis further shows that informative tokens tend to exhibit persistent deviations from common group evolution directions. Based on this observation, we propose EvoCut, a training-free and attention-free visual token compression method that estimates token importance from multi-layer evolution deviation. Experimental results show that EvoCut can retain only 11.1\% of the visual tokens on LLaVA-1.5-7B while preserving 94.4\% of the average performance, demonstrating its effectiveness in balancing efficiency and accuracy.

cs.CV↗

arXiv2026

The Invitation Trap: Proactive Availability Backdoor in LLMs via Conversational Induction

Current backdoor attacks against LLMs are typically manipulated by the attacker and remain passive. In this paper, we introduce the \textbf{Proactive Availability Backdoor (PAB)}, a novel paradigm that shifts the attack vector from passive waiting to active social engineering. By weaponizing the inherent helpfulness of aligned LLMs, PAB proactively traps users into executing trigger-implanted queries by offering suggestions, achieving high aggressiveness, precision and stealthiness. To rigorously evaluate its threat in a real-life context, we introduce a dual-agent ecological simulation framework based on selected dimensions of the Five-Factor Model, and deploy PAB with few-shot prompts. Being validated on different models and domains, PAB performs remarkably and its effective attack success rate, which calculates the joint probability of attack incidence rate and attack success rate, goes to \textbf{73.1\%}. We also introduce \textbf{Anti-PAB}, a defense method tailored for PAB. Our findings reveal that the helpfulness of LLMs can be weaponized to compromise availability, exposing a serious hidden threat to LLMs users. We release all the scripts and datasets in the experiments at \texttt{https://anonymous.4open.science/r/PAB-ANONYMOUS/}.

cs.CR↗

arXiv2026

Noise Suppression and Radio Frequency Interference Rejection for Self-Triggered Radio Detectors of Extensive Air Showers

Self-triggered radio detection of ultra-high-energy cosmic rays and neutrinos offers a scalable and cost-effective approach for next-generation astroparticle observatories, but remains challenging under realistic radio-frequency interference (RFI) conditions. In the classical air-shower radio band, the achievable sensitivity and trigger reliability are critically limited by the balance between external sky background noise and internal detector-unit noise, as well as by non-stationary anthropogenic interference. In this work, we present an end-to-end design and experimental characterization of a self-triggered radio detector unit explicitly optimized to operate in a galactic-noise-dominated regime. Rather than focusing on a single hardware component or trigger algorithm, we adopt a system-level methodology that coherently integrates sky-noise modeling, RF-chain noise budgeting, electromagnetic compatibility (EMC) mitigation, and measurement-driven validation. By using the galactic radio background as a quantitative reference, we assess the internal noise performance of the detector unit and demonstrate conditions under which extensive air shower (EAS) radio signals can be distinguished from anthropogenic interference at the system-response level. We further introduce an indirect noise-quantification method to estimate the low-noise amplifier contribution within the complete RF chain based on differential internal-noise measurements evaluated at the ADC level. The proposed detector unit is validated through laboratory and on-site measurements, demonstrating operation close to the galactic-noise limit in the core frequency band. These results provide a practical and transferable methodology for the design and deployment of large-scale self-triggered radio arrays such as GRAND.

astro-ph.IM↗

arXiv2026

RISE: Reliable Improvement in Self-Evolving Vision-Language Models

Vision-language models (VLMs) have achieved strong multimodal reasoning capabilities, but further improving them still relies heavily on large-scale human-constructed supervision for post-training. Such supervision is costly to obtain, especially for reasoning-intensive multimodal tasks where questions, answers, and feedback signals must be carefully designed. This motivates self-evolving learning, where a model improves itself through a dual-role closed loop: a questioner autonomously poses questions and a solver learns to solve them. However, we observe that current VLM self-evolving methods still face three major challenges: coarse-grained role alternation delays the interaction between question generation and solver adaptation; generated questions can progressively degrade in quality; and question types may collapse toward a narrow distribution. These issues limit the efficiency and reliability of self-evolution. Thus, we propose \textbf{RISE}, a reliable self-evolving framework for vision-language models. RISE is built on three complementary designs: fine-grained role alternation, which shortens the feedback loop between the questioner and the solver to improve efficiency; a quality supervisor, which improves question validity and pseudo-label reliability; and skill-aware dynamic balancing, which mitigates mode collapse and maintains broad skill coverage during evolution. Together, these components enable more reliable and effective self-evolution from unlabeled images. Experiments on two VLM backbones across seven benchmarks show that RISE consistently improves the base models, yielding broad and sustained gains. Our code is publicly available at https://github.com/AMAP-ML/RISE.

cs.CV↗

arXiv2026

Instruction Anchor: Dissecting the Mechanistic Dynamics of Modality Arbitration

Modality following is the ability to selectively leverage multimodal contexts based on user instructions. It is fundamental to the safety and reliability of multimodal large language models (MLLMs) in real-world deployments. However, the internal mechanisms governing this decision-making process remain largely under-explored. In this work, we investigate the mechanism underlying modality following through an information flow perspective. Our findings reveal that instruction tokens serve as structural anchor for modality arbitration: Shallow attention layers perform undifferentiated information transfer, aggregating multimodal cues to instruction tokens as a latent buffer; in contrast, deep attention layers selectively strengthen the instruction-compliant subspace and resolve modality arbitration according to the instruction-specified intent, with a sparse subset of attention heads driving this process. Targeted attention-head interventions further validate the functional specificity of these heads: blocking only $5\%$ of the identified heads substantially degrades modality following while preserving general visual and language capabilities, whereas targeted amplification can restore failed modality-following samples by up to approximately $60\%$. Together, this work provides a mechanistic account of modality following and informs future efforts to improve how MLLMs integrate and utilize multimodal evidence under user instructions.

cs.CL↗