Searcharxiv⌕ Search

arXiv subjects

Qi Guo

Publications and source records attributed to Qi Guo.

At least 55 records · Page 3Linked to original sources
arXiv2026

AutoPPA: Automated Circuit PPA Optimization via Contrastive Code-based Rule Library Learning

Performance, power, and area (PPA) optimization is a fundamental task in RTL design, requiring a precise understanding of circuit functionality and the relationship between circuit structures and PPA metrics. Recent studies attempt to automate this process using LLMs, but neither feedback-based nor knowledge-based methods are efficient enough, as they either design without any prior knowledge or rely heavily on human-summarized optimization rules. In this paper, we propose AutoPPA, a fully automated PPA optimization framework. The key idea is to automatically generate optimization rules that enhance the search for optimal solutions. To do this, AutoPPA employs an Explore-Evaluate-Induce ($E^2I$) workflow that contrasts and abstracts rules from diverse generated code pairs rather than manually defined prior knowledge, yielding better optimization patterns. To make the abstracted rules more generalizable, AutoPPA employs an adaptive multi-step search framework that adopts the most effective rules for a given circuit. Experiments show that AutoPPA outperforms both the manual optimization and the state-of-the-art methods SymRTLO and RTLRewriter.

cs.LG↗
arXiv2026

LinuxArena: A Control Setting for AI Agents in Live Production Software Environments

We introduce LinuxArena, a control setting in which agents operate directly on live, multi-service production environments. LinuxArena contains 20 environments, 1,671 main tasks representing legitimate software engineering work, and 184 side tasks representing safety failures such as data exfiltration and backdooring, making it the largest and most diverse control setting for software engineering to date. We validate LinuxArena is useful for control research by running sabotage evaluations, which measure whether attackers can complete side tasks while working on main tasks, and monitor evaluations, which measure a monitor model's ability to detect sabotage attempts. Against a GPT-5-nano trusted monitor at a 1\% step-wise false positive rate, Claude Opus 4.6 achieves roughly a 23% undetected sabotage success rate. We additionally release LaStraj, a dataset of human-crafted attack trajectories that evade monitors at substantially higher rates than any model-generated attacks we elicited, showing that current attack policies do not saturate LinuxArena. These results suggest that LinuxArena has meaningful headroom for both attackers and defenders, making it a strong testbed for developing and evaluating future control protocols.

cs.CR↗
arXiv2026

LKV: End-to-End Learning of Head-wise Budgets and Token Selection for LLM KV Cache Eviction

Long-context inference in Large Language Models (LLMs) is bottlenecked by the linear growth of Key-Value (KV) cache memory. Existing KV cache compression paradigms are fundamentally limited by heuristics: heuristic budgeting relies on statistical priors rather than task objectives, causing resource misallocation, while heuristic selection relies on coupled query-key interactions or static inductive biases (e.g., attention sinks). To address this limitation, we introduce LKV (Learned KV Eviction), which formulates KV compression as an end-to-end differentiable optimization problem. LKV integrates LKV-H to learn task-optimized global budgets, and LKV-T to derive intrinsic KV importance without materializing attention matrices. This design bypasses heuristic proxies, strictly aligning compression with task objectives. Extensive evaluations demonstrate that LKV achieves state-of-the-art performance on both LongBench and RULER benchmarks at high compression rates. In particular, on LongBench, LKV achieves near-lossless performance with only 15\% KV cache retention. Crucially, our analysis identifies learned budgeting as the dominant driver of fidelity, demonstrating that data-driven allocation is essential to overcome the limitations of hand-crafted heuristics.

cs.LG↗
arXiv2026

Dynamically cold discs in high-redshift galaxies: comparison between ALMA observations and TNG50

Observations of highly rotationally supported gas discs in high redshift ($z$ > 3) star-forming galaxies challenge our understanding of galaxy formation, as the prevailing view holds that galaxies in the early universe are dynamically hot due to frequent mergers, gas accretion, and strong stellar feedback. We examined the kinematic properties of massive ($M_{\star} \geq 10^{10}\,M_{\odot}$) star-forming galaxies in the TNG50 cosmological hydrodynamical simulation in the redshift range $3\leq z \leq 5$. Mock emission line datacubes were constructed and analysed using the same methodology as for [CII] observations with ALMA. We measured the ratio of the gas rotation velocity ($V$) to velocity dispersion ($σ$) finding that most galaxies have $V/σ\sim$ $2-3$, lower than observed. However, a few simulated galaxies show $V/σ$ > 5. Such "cold" discs, selected at $z=4$, remain dynamically colder than most of the TNG population across $z=3-5$. A galaxy with $V/σ\gtrsim10$ appears in a transient phase that lasts $\leq200$ Myr. Dynamically cold disc formation in TNG50 is promoted by gas accretion with angular momentum aligned with the pre-existing disc, while most galaxies undergo misaligned accretion. Dynamically cold discs also show lower mass accretion rates and better aligned stellar and dark-matter angular momentum vectors. By tracing their evolution to $z = 0$, we find that one-third become massive disc galaxies and two-thirds become ETGs.

astro-ph.GA↗
arXiv2026

Compact single-shot ranging and near-far imaging using metasurfaces

We present a metasurface imaging system capable of simultaneously capturing two images at close range (1-2~cm) and an additional image at long range (about 40~cm) on a shared photosensor. The close-range image pair focuses at 1.4~cm and 2.0~cm, respectively, which forms a focal stack, enabling passive ranging with an accuracy of $\pm$1~mm from 12~mm to 20~mm through a computationally efficient depth-from-defocus algorithm for a simplified scenario. The entire system is compact, with a total track length of 15~mm, making it suitable for seamless integration into edge platforms for defense and other resource-constrained applications.

eess.IV↗
arXiv2026

DAT: Dual-Aware Adaptive Transmission for Efficient Multimodal LLM Inference in Edge-Cloud Systems

Multimodal large language models (MLLMs) have shown strong capability in semantic understanding and visual reasoning, yet their use on continuous video streams in bandwidth-constrained edge-cloud systems incurs prohibitive computation and communication overhead and hinders low-latency alerting and effective visual evidence delivery. To address this challenge, we propose DAT to achieve high-quality semantic generation, low-latency event alerting, and effective visual evidence supplementation. To reduce unnecessary deep reasoning costs, we propose a collaborative small-large model cascade. A lightweight edge-side small model acts as a gating module to filter non-target-event frames and perform object detection, triggering MLLM inference only for suspicious frames. Building on this, we introduce an efficient fine-tuning strategy with visual guidance and semantic prompting, which improves structured event understanding, object detection, and output consistency. To ensure low-latency semantic alerting and effective visual evidence supplementation under bandwidth constraints, we further devise a semantics and bandwidth-aware multi-stream adaptive transmission optimization method. Experimental results show that DAT achieves 98.83% recognition accuracy and 100% output consistency. Under severe congestion, it reduces weighted semantic alert delay by up to 77.5% and delivers 98.33% of visual evidence within 0.5 s, demonstrating the effectiveness of jointly optimizing cascade inference and elastic transmission.

cs.MM↗
arXiv2026

QiMeng-PRepair: Precise Code Repair via Edit-Aware Reward Optimization

Large Language Models (LLMs) achieve strong program repair performance but often suffer from over-editing, where excessive modifications overwrite correct code and hinder bug localization. We systematically quantify its impact and introduce precise repair task, which maximizes reuse of correct code while fixing only buggy parts. Building on this insight, we propose PRepair, a framework that mitigates over-editing and improves repair accuracy. PRepair has two components: Self-Breaking, which generates diverse buggy programs via controlled bug injection and min-max sampling, and Self-Repairing, which trains models with Edit-Aware Group Relative Policy Optimization (EA-GRPO) using an edit-aware reward to encourage minimal yet correct edits. Experiments show that PRepair improves repair precision by up to 31.4% under $\mathrm{fix}_1@1$, a metric that jointly considers repair correctness and extent, and significantly increases decoding throughput when combined with speculative editing, demonstrating its potential for precise and practical code repair.

cs.SE↗
arXiv2026

MSAO: Adaptive Modality Sparsity-Aware Offloading with Edge-Cloud Collaboration for Efficient Multimodal LLM Inference

Multimodal large language models (MLLMs) enable powerful cross-modal reasoning capabilities but impose substantial computational and latency burdens, posing critical challenges for deployment on resource-constrained edge devices. In this paper, we propose MSAO, an adaptive modality sparsity-aware offloading framework with edge-cloud collaboration for efficient MLLM Inference. First, a lightweight heterogeneous modality-aware via fine-grained sparsity module performs spatial-temporal-modal joint analysis to compute the Modality Activation Sparsity (MAS) metric, which quantifies the necessity of each modality with minimal computational overhead. Second, an adaptive speculative edge-cloud collaborative offloading mechanism dynamically schedules workloads between edge and cloud based on the derived MAS scores and real-time system states, leveraging confidence-guided speculative execution to hide communication latency. Extensive experiments on VQAv2 and MMBench benchmarks demonstrate that MSAO achieves a 30% reduction in end-to-end latency and 30%-65% decrease in resource overhead, while delivering a throughput improvement of 1.5x to 2.3x compared to traditional approaches, all without compromising competitive accuracy.

cs.DC↗
arXiv2026

Pupil Design for Computational Wavefront Estimation

Establishing a precise connection between imaged intensity and the incident wavefront is essential for emerging applications in adaptive optics, holography, computational microscopy, and non-line-of-sight imaging. While prior work has shown that breaking symmetries in pupil design enables wavefront recovery from a single intensity measurement, there is little guidance on how to design a pupil that improves wavefront estimation. In this work we introduce a quantitative asymmetry metric to bridge this gap and, through an extensive empirical study and supporting analysis, demonstrate that increasing asymmetry enhances wavefront recoverability. We analyze the trade-offs in pupil design, and the impact on light throughput along with performance in noise. Both large-scale simulations and optical bench experiments are carried out to support our findings.

eess.IV↗
arXiv2026

A census of quiescent galaxies across $0.5 < z < 8$ with JWST/MIRI: Mass-dependent number density evolution of quiescent galaxies in the early Universe

Recent JWST observations have revealed a large population of quiescent galaxies (QGs) at high redshift ($z \sim 4-8$), challenging current models of early galaxy formation and quenching. Accurate number density estimates are crucial but remain uncertain. We present a systematic study of QGs at $0.5 < z < 8$ using a mass-complete sample from the JWST/PRIMER survey with deep NIRCam and MIRI imaging. We demonstrate that MIRI photometry is essential for refining the QG sample: it helps to mitigate contamination from dusty star-forming galaxies in the high-mass regime at $z \sim 3-5$ and aids in recovering lower-mass QG candidates at $z > 5$ that are often missed without including MIRI data. We find that the evolution of the QG number density is strongly mass-dependent. The density of massive QGs ($\log (M_{\star}/M_{\odot}) > 10.6$) declines rapidly, falling from $n \approx 1.32\times10^{-5}~~\mathrm{Mpc^{-3}}$ at $z \sim 3-4$ to $n < 1 \times10^{-6}~~\mathrm{Mpc^{-3}}$ at $z \sim 6$, and becomes negligible at $z > 6$. In contrast, low-mass QGs ($9.5 < \log (M_{\star}/M_{\odot}) < 10.6$) exhibit a remarkably constant number density of $n \sim 3\times10^{-6}~\mathrm{Mpc^{-3}}$ across the redshift range $z = 4-8$. This plateau suggests that these high-redshift, low-mass QGs may be galaxies undergoing temporary quenching episodes, likely subject to rejuvenation upon future gas accretion. Comparisons with leading galaxy formation models reveal significant tensions: most models underestimate the abundance of massive QGs at $z > 4$ and fail to reproduce the flat density evolution observed for the low-mass population.

astro-ph.GA↗
arXiv2026

PersonalQ: Select, Quantize, and Serve Personalized Diffusion Models for Efficient Inference

Personalized text-to-image generation lets users fine-tune diffusion models into repositories of concept-specific checkpoints, but serving these repositories efficiently is difficult for two reasons: natural-language requests are often ambiguous and can be misrouted to visually similar checkpoints, and standard post-training quantization can distort the fragile representations that encode personalized concepts. We present PersonalQ, a unified framework that connects checkpoint selection and quantization through a shared signal -- the checkpoint's trigger token. Check-in performs intent-aligned selection by combining intent-aware hybrid retrieval with LLM-based reranking over checkpoint context and asks a brief clarification question only when multiple intents remain plausible; it then rewrites the prompt by inserting the selected checkpoint's canonical trigger. Complementing this, Trigger-Aware Quantization (TAQ) applies trigger-aware mixed precision in cross-attention, preserving trigger-conditioned key/value rows (and their attention weights) while aggressively quantizing the remaining pathways for memory-efficient inference. Experiments show that PersonalQ improves intent alignment over retrieval and reranking baselines, while TAQ consistently offers a stronger compression-quality trade-off than prior diffusion PTQ methods, enabling scalable serving of personalized checkpoints without sacrificing fidelity.

cs.AI↗
arXiv2026

LongCat-Flash-Prover: Advancing Native Formal Reasoning via Agentic Tool-Integrated Reinforcement Learning

We introduce LongCat-Flash-Prover, a flagship 560-billion-parameter open-source Mixture-of- Experts (MoE) model that advances Native Formal Reasoning in Lean4 through agentic tool-integrated reasoning (TIR). We decompose the native formal reasoning task into three independent formal capabilities, i.e., auto-formalization, sketching, and proving. To facilitate these capabilities, we propose a Hybrid-Experts Iteration Framework to expand high-quality task trajectories, including generating a formal statement based on a given informal problem, producing a whole-proof directly from the statement, or a lemma-style sketch. During agentic RL, we present a Hierarchical Importance Sampling Policy Optimization (HisPO) algorithm, which aims to stabilize the MoE model training on such long-horizon tasks. It employs a gradient masking strategy that accounts for the policy staleness and the inherent train-inference engine discrepancies at both sequence and token levels. Additionally, we also incorporate theorem consistency and legality detection mechanisms to eliminate reward hacking issues. Extensive evaluations show that our LongCat-Flash-Prover sets a new state-of-the-art for open-weights models in both auto-formalization and theorem proving. Demonstrating remarkable sample efficiency, it achieves a 97.1% pass rate on MiniF2F-Test using only 72 inference budget per problem. On more challenging benchmarks, it solves 70.8% of ProverBench and 41.5% of PutnamBench with no more than 220 attempts per problem, significantly outperforming existing open-weights baselines.

cs.AI↗
arXiv2026

Modeling the Spectral Energy Distribution of Active Galactic Nuclei: Implications for Cosmological Simulations of Galaxy Formation

Modeling the spectral energy distribution (SED) of active galactic nuclei (AGN) plays a very important role in constraining modern cosmological simulations of galaxy formation. Here, we utilize an advanced supermassive black hole (SMBH) accretion disk model to compute the accretion flow structure and AGN SED across a wide range of black hole mass ($M_{\rm SMBH}$) and dimensionless accretion rates $\dot{m}(\equiv \dot{M}_{\rm acc}/\dot{M}_\mathrm{Edd})$, where $\dot{M}_{\rm acc}$ is the mass flow rate through the disk and $\dot{M}_\mathrm{Edd}$ is the Eddington mass accretion rate. We find that the radiative efficiency is mainly influenced by $\dot m$, while contributions of $M_{\rm SMBH}$ and $\dot{m}$ to the bolometric luminosity are comparably important. We have developed new scaling relationships that relate the bolometric luminosity of an AGN to its luminosities in the hard X-ray, soft X-ray, and optical bands. Our results align with existing literature at high luminosities but suggest lower luminosities in the hard and soft X-ray bands for AGNs with low bolometric luminosities than commonly reported values. Combining with the semi-analytical model of galaxy formation \textsc{L-Galaxies} and Millennium dark matter simulation for the distribution of ($M_{\rm SMBH}, \dot{m}$) at different redshift, we find the model predictions align well with observational data at redshifts below 1 but deviates for higher redshifts regarding AGN detection fraction and luminosity functions. This deviation may arise from improper treatment of SMBH growth at high redshifts in the model or bias from limited observational data. This AGN SED calculation can be readily applied in other cosmological simulations.

astro-ph.GA↗
arXiv2026

SpiderCam: Low-Power Snapshot Depth from Differential Defocus

We introduce SpiderCam, an FPGA-based snapshot depth-from-defocus camera which produces 480x400 sparse depth maps in real-time at 32.5 FPS over a working range of 52 cm while consuming 624 mW of power in total. SpiderCam comprises a custom camera that simultaneously captures two differently focused images of the same scene, processed with a SystemVerilog implementation of depth from differential defocus (DfDD) on a low-power FPGA. To achieve state-of-the-art power consumption, we present algorithmic improvements to DfDD that overcome challenges caused by low-power sensors, and design a memory-local implementation for streaming depth computation on a device that is too small to store even a single image pair. We report the first sub-Watt total power measurement for passive FPGA-based 3D cameras in the literature.

cs.CV↗
arXiv2026

Reproducing Abell 2744 with the HyperMillennium Simulation

We present the Hyper Millennium (HM) simulation, an extremely large cosmological simulation designed to support next-generation galaxy surveys. The simulation follows 4.2 trillion dark matter particles in a comoving box of $2.5\ h^{-1}{\rm Gpc}$, with a mass resolution of $3.2 \times 10^8\, {h^{-1}\rm M_{\odot}}$ and a force resolution of $3.0\ h^{-1}{\rm kpc}$. Its combination of scale and resolution is ideal for studying large-scale structures and rare cosmic objects. In this first paper of the HM project, we explore whether the massive galaxy cluster Abell~2744 (A2744) can be reproduced in detail in the simulation. Pixel-based statistics of galaxy number density $N_{\rm gal}$, luminosity density $L_{\rm gal}$, and projected mass density $κ$ show excellent agreement between A2744 and its analogues down to $\sim 50$ kpc, once field-selection biases toward high galaxy surface density are accounted for. This concordance, achieved in one of the most extreme known galaxy environments, is a validation of the underlying $Λ{\rm CDM}$ model in the extreme regime of A2744. It also showcases the robustness and accuracy of the HM simulation, which, when coupled with a sophisticated semi-analytic galaxy formation model, is capable of producing galaxy and mass catalogues of comparable quality out to high redshift across its full comoving volume of 50.4 ${\rm Gpc^3}$.

astro-ph.CO↗
arXiv2026

DriveFix: Spatio-Temporally Coherent Driving Scene Restoration

Recent advancements in 4D scene reconstruction, particularly those leveraging diffusion priors, have shown promise for novel view synthesis in autonomous driving. However, these methods often process frames independently or in a view-by-view manner, leading to a critical lack of spatio-temporal synergy. This results in spatial misalignment across cameras and temporal drift in sequences. We propose DriveFix, a novel multi-view restoration framework that ensures spatio-temporal coherence for driving scenes. Our approach employs an interleaved diffusion transformer architecture with specialized blocks to explicitly model both temporal dependencies and cross-camera spatial consistency. By conditioning the generation on historical context and integrating geometry-aware training losses, DriveFix enforces that the restored views adhere to a unified 3D geometry. This enables the consistent propagation of high-fidelity textures and significantly reduces artifacts. Extensive evaluations on the Waymo, nuScenes, and PandaSet datasets demonstrate that DriveFix achieves state-of-the-art performance in both reconstruction and novel view synthesis, marking a substantial step toward robust 4D world modeling for real-world deployment.

cs.CV↗
arXiv2026

FAR-Drive: Frame-AutoRegressive Video Generation in Closed-Loop Autonomous Driving

Despite rapid progress in autonomous driving, reliable training and evaluation of driving systems remain fundamentally constrained by the lack of scalable and interactive simulation environments. Recent generative video models achieve remarkable visual fidelity, yet most operate in open-loop settings and fail to support fine-grained frame-level interaction between agent actions and environment evolution. Building a learning-based closed-loop simulator for autonomous driving poses three major challenges: maintaining long-horizon temporal and cross-view consistency, mitigating autoregressive degradation under iterative self-conditioning, and satisfying low-latency inference constraints. In this work, we propose FAR-Drive, a frame-level autoregressive video generation framework for autonomous driving. We introduce a multi-view diffusion transformer with fine-grained structured control, enabling geometrically consistent multi-camera generation. To address long-horizon consistency and iterative degradation, we design a two-stage training strategy consisting of adaptive reference horizon conditioning and blend-forcing autoregressive training, which progressively improves consistency and robustness under self-conditioning. To meet low-latency interaction requirements, we further integrate system-level efficiency optimizations for inference acceleration. Experiments on the nuScenes dataset demonstrate that our method achieves state-of-the-art performance among existing closed-loop autonomous driving simulation approaches, while maintaining sub-second latency on a single GPU.

cs.CV↗
arXiv2026

QiMeng-CodeV-SVA: Training Specialized LLMs for Hardware Assertion Generation via RTL-Grounded Bidirectional Data Synthesis

SystemVerilog Assertions (SVAs) are crucial for hardware verification. Recent studies leverage general-purpose LLMs to translate natural language properties to SVAs (NL2SVA), but they perform poorly due to limited data. We propose a data synthesis framework to tackle two challenges: the scarcity of high-quality real-world SVA corpora and the lack of reliable methods to determine NL-SVA semantic equivalence. For the former, large-scale open-source RTLs are used to guide LLMs to generate real-world SVAs; for the latter, bidirectional translation serves as a data selection method. With the synthesized data, we train CodeV-SVA, a series of SVA generation models. Notably, CodeV-SVA-14B achieves 75.8% on NL2SVA-Human and 84.0% on NL2SVA-Machine in Func.@1, matching or exceeding advanced LLMs like GPT-5 and DeepSeek-R1.

cs.CL↗