arXiv subjects

Yuan Feng

Publications and source records attributed to Yuan Feng.

At least 19 recordsLinked to original sources

arXiv2026

Efficient Compilation for Hamiltonian Simulation via Global Binary Symplectic Form Simplification

Hamiltonian simulation is a core quantum workload, underpinning variational quantum algorithms and Trotterized time evolution. Such programs are expressed as Pauli exponential sequences, exhibiting structural patterns that are highly amenable to high-level synthesis and optimization. Existing compilers, however, fail to fully unlock the optimization potential of their global algebraic structure, even when employing advanced graph- or tableau-based methods. We present Symphony, a holistic compilation approach built on the binary symplectic form (BSF) representation of Pauli strings. Unlike prior group-wise BSF simplification and path-based Pauli network synthesis, Symphony applies generalized controlled-Pauli Clifford transformations directly to a global BSF tableau, progressively reducing active Pauli rows while adaptively emitting eligible two-qubit blocks in addition to single-qubit rotations in a forward Clifford frame. Following algebraic simplification, Symphony performs a causality-preserving block rescheduling heuristic that respects frame-induced dependencies while exposing extensive two-qubit block parallelism opportunities. This streamlined compilation style comprehensively exploits simultaneous simplification and commutativity opportunities, achieving efficient global optimization without relying on computationally expensive heuristics or long-horizon searches. Across the generic Hamiltonian simulation benchmarks in HamLib, Symphony achieves average reductions of 59% in two-qubit gate count and 91% in circuit depth. It Pareto-dominates prior state-of-the-art compilers, requiring 1.17--1.58$\times$ fewer two-qubit gates and especially shrinking two-qubit circuit depth by a substantial factor of 1.87--5.67$\times$ on average.

quant-ph↗

arXiv2026

VisionWeave: Weaving Elastic Visual Representations as a Native Capability of MLLMs

Multimodal large language models have become the dominant paradigm for visual understanding, but incur substantial costs by encoding inputs into dense, fixed-size patch tokens. However, visual information is unevenly distributed: some regions require fine-grained detail, while others admit compact representations. Downsampling sacrifices this detail, while existing token pruning and adaptive approaches remain limited in content-adaptive granularity, task generalization, and integration with modern MLLMs and serving infrastructure. Overcoming these limitations calls for foundation models that learn, end to end, where-and at what granularity-to allocate visual representations, a native capability we term elastic visual representation weaving. We introduce VisionWeave, establishing this capability in frontier-level MLLMs through large-scale training. It combines two components: a gated spatial pooler constructs coarse-grained representations alongside native fine-grained representations within a shared MRoPE coordinate, while a granularity router learns their content-adaptive allocation. Through self-distillation alone, we validate this capability on Qwen3.5-4B and scale to Qwen3.8-27B with over 30K A100 GPU-hours. Based on Qwen3.8-27B, VisionWeave adaptively adjusts token savings to visual content, saving 43.0% tokens on average while retaining 98.9% native performance across eight benchmarks, versus only 88% performance preserved for token pruning baselines with a fixed 50% savings target. Extensive evaluations confirm robust efficiency-quality trade-offs across diverse tasks, resolutions and video frames. When deployed on SGLang serving engine, our method achieves a 2.3x throughput gain while reducing mean TTFT by 54.4% and mean TPOT by 60.6%. Together, we believe these results position elastic visual weaving as a promising capability for next-generation multimodal models.

cs.CV↗

arXiv2026

OmniVChat: Synthesizing, Benchmarking, and Training for Native Audio-Visual Dialogue

We define OmniVChat (Omni Video Chat) as the task of native audio-visual dialogue between a user and an omni model. In OmniVChat, omni models directly and simultaneously receive audio and video from a user and return text. The user's query is embedded in the audio and video, without a separate text question, external captioning, or speech recognition. Direct audio-visual input reduces external latency and computation while preserving perceptual cues. However, research on OmniVChat faces two constraints: data availability and evaluation. Recordings of people using their own devices are scarce. Furthermore, good replies often depend on multimodal context and can be phrased in many ways, making keyword matching unreliable for evaluation. Recent progress in agent systems and video generation makes generation for comprehension viable, which means using synthesized dialogues for training and evaluation. Therefore, we present OmniVChat-Studio, a multi-agent data engine for synthesizing single- and multi-turn audio-visual dialogues. We use synthesized dialogues to build OmniVChat-Bench, an evaluation benchmark that evaluates omni models' basic dialogue abilities across five ability categories. Replies are judged by a large language model based on explicit scoring criteria. We also present OmniVChat-RL, a reinforcement learning reward design that jointly targets reply correctness, efficiency, and style in OmniVChat. Training Qwen3-Omni-Instruct with OmniVChat-RL on synthesized dialogues improves its performance on both OmniVChat-Bench and the human-recorded OmniVChat-Bench-Human. These gains validate the reward design and show transfer to real-world dialogues in training and evaluation.

eess.AS↗

arXiv2026

ScaleBlind: Point Cloud Completion under Unknown Scale

Point cloud completion aims to infer a complete 3D shape from a partial point cloud and serves as a fundamental building block for downstream tasks such as reconstruction, editing, and simulation. Despite the recent progress, existing learning-based methods often implicitly rely on access to the ground-truth shape scale (GT-scale) during both training- and testing-time normalization, assuming privileged information that is unavailable in real-world inference. This hidden assumption limits practical deployment and can lead to severe completion artifacts, e.g., over- or under-completion and nested shells, once the oracle GT-scale cue is removed. We observe that the recent foundation image generation models exhibit a strong capability of understanding objects and geometries, and producing multi-view consistent renderings, making them promising priors for GT-scale-free 3D completion. Motivated by this insight, we propose ScaleBlind, a novel framework that leverages foundation-model-based image completion to recover global scale directly from partial inputs and then faithfully produces the 3D completion. Specifically, ScaleBlind dreams out complete multi-view appearances from rendered partial views, lifts the inferred missing regions back into 3D to obtain a geometry-aware coarse completion, and further refines it via a powerful cross-modal fusion network with the original partial point cloud. By harnessing 2D foundation priors, our method eliminates the need for accessing GT-scale information at inference. Moreover, it provides a principled bridge between 2D generative priors and 3D point cloud completion. Extensive experiments demonstrate the superiority of our framework, making ScaleBlind the new state-of-the-art for the point cloud completion task.

cs.CV↗

arXiv2026

VideoMM: Adaptive Macro-Micro Inference for Efficient Video MLLMs

Scaling Multimodal Large Language Models (MLLMs) to long-form video understanding is bottlenecked by the explosion of visual tokens, which saturates context windows and incurs prohibitive costs. Current solutions predominantly rely on auxiliary models for token reduction but face a fundamental dilemma: lightweight encoder-driven approaches often overlook critical semantic information, whereas heavyweight MLLM-driven reduction negates the efficiency gains. {In this work, we identify a more fundamental inefficiency underlying this dilemma: while fine-grained visual details are essential for detailed understanding, they are largely redundant for the preliminary task of selecting semantically relevant regions. } Motivated by this, we introduce \textbf{VideoMM}, which marks a paradigm shift from model-centric downsizing to adaptive perceptual granularity. Specifically, our framework {decouples selection from reasoning} by executing semantic filtering on a cost-effective \textit{Macro Proxy} (derived from downscaled frames), and projecting the selected regions onto high-fidelity \textit{Micro Tokens} for detailed understanding only when necessary. Extensive evaluations show that VideoMM significantly outperforms existing solutions. It achieves a 6.13$\times$ speedup and a 7.4\% accuracy gain over full-context baselines on LongVideoBench, and further accelerates inference by 2.73$\times$ over current leading methods, establishing a highly scalable paradigm for long-video understanding. Our code is available at: https://github.com/adfh917k/VideoMM.

cs.AI↗

arXiv2026

LibraSpec: Dynamic Diffusion-Based Speculative Decoding via Marginal-Gain-Driven Optimization

Speculative decoding accelerates large language model inference by drafting multiple tokens for parallel verification, with efficiency critically determined by the speculative length selected at each decoding round. Existing dynamic speculation methods select the speculation length by estimating how many tokens will be accepted, which is reasonable for autoregressive drafters that generates tokens sequentially. The recent wave of diffusion-based drafters, however, generates candidate blocks in parallel at substantially lower drafting cost, shifting the key question from how many tokens to generate to how many generated tokens are worth verifying. We therefore reformulate dynamic speculative-length selection as expected-speedup optimization and derive a marginal criterion that extends the speculative sequence only when its acceptance gain outweighs the additional verification cost. Building on this criterion, we develop \textit{LibraSpec}, a training-free and plug-and-play algorithm that iteratively determines the speculative length using drafter confidence scores. Theoretically, we prove that LibraSpec monotonically converges toward the optimal speculative length. Experiments across six target models, three diffusion-based speculative decoding methods, and math, coding, and chat benchmarks show consistent improvements under both greedy and sampling settings, achieving a further $0.5\sim1.5\times$ improvement over baselines and up to $8.49\times$ speedup over autoregressive decoding.

cs.CL↗

arXiv2026

An Interpretation of Bunched Logic for Reasoning about Heap-Manipulating Quantum Programs

We introduce heap manipulation into quantum programming languages to enable flexible quantum memory management, which in turn poses new challenges for reasoning about program correctness. To address these challenges, we develop a novel quantum interpretation of bunched logic by extending Birkhoff--von Neumann quantum logic to Hilbert spaces whose dimensions vary to accommodate the dynamic allocation and deallocation of variables on the quantum heap. Beyond the separating conjunction ($*$), we present, for the first time, a quantum interpretation of the separating implication ($-\mkern-3mu*$) to support backward reasoning. This interpretation preserves the adjunction between ($*$) and ($-\mkern-3mu*$), ensuring that both operators capture spatial properties and are precisely aligned with the semantics of quantum heap manipulations. Building on this foundation, we establish a quantum separation logic that supports local reasoning and is relatively complete.

quant-ph↗

arXiv2026

Resource Estimation for Fault-Tolerant Quantum Programs

Fault-tolerant quantum computation enables the deployment of practical quantum algorithms but incurs substantial overhead from error correction, making resource estimation a central concern. Beyond case-by-case analyses, existing quantum programming languages either require programmers to manipulate low-level hardware details, rendering fault-tolerant implementations cumbersome, or abstract away the underlying error-correction schemes, reducing the effectiveness of resource utilization and estimation. To address these limitations while preserving programmability, we present a quantum programming language that enables efficient resource utilization, together with a resource-estimation framework for comprehensive resource analysis. Our framework features programmer-visible abstractions of error-correction schemes and cross-layer program-hardware analysis, allowing systematic exploration of resource trade-offs. We evaluate our approach on detailed fault-tolerant implementations of practical large-scale quantum algorithms, including components typically treated as black boxes in existing frameworks. The results demonstrate that our framework enables substantial resource savings while delivering detailed, fine-grained, and accurate resource estimates for fault-tolerant quantum programs.

quant-ph↗

arXiv2026

A Unified Causal Inference Framework for the Desirability of Outcome Ranking Paradigm in Benefit-Risk Evaluation

We developed a unified covariate-adjusted causal inference framework for estimating the desirability of outcome ranking (DOOR) probability for benefit-risk evaluation in randomized trials and observational studies. The framework expresses the DOOR probability as a bilinear functional of the marginal ordinal outcome distributions under the two treatment strategies, estimates conditional ordinal distributions through sequential risk-set hazards, and derives the efficient influence function (EIF) of the DOOR probability. The point-estimation simulations compared G-computation, normalized inverse probability weighting (IPW), augmented IPW (AIPW), and targeted maximum likelihood estimation (TMLE), with nuisance functions estimated using generalized linear models or Super Learner (SL). TMLE-SL showed the strongest and most consistent point-estimation performance, with AIPW-SL ranking second. EIF-based inference was then evaluated for AIPW-SL and TMLE-SL, with and without cross-fitting, across settings varying in overlap, treatment-effect heterogeneity, and treatment allocation. CVTMLE-SL showed the strongest overall performance across DOOR-scale bias, recovery of the underlying ordinal distributions, standard-error accuracy, and confidence-interval coverage. We illustrate the methodology using data from the multidrug-resistant organism network of the Antibacterial Resistance Leadership Group.

stat.ML↗

arXiv2026

REAL: REtrieval-reAsoning and Logic-constructed Attention Behaviors for Long-Context KV Cache Compression

The growing sequence length of large language models poses significant challenges for key-value (KV) caches. Existing state-of-the-art cache eviction methods primarily analyze the inference behavior of attention heads in successful retrieval-reasoning cases, often overlooking diverse behaviors in failure cases, such as bias and distraction. This oversight limits the potential to leverage heterogeneous head behaviors for improved eviction performance. Inspired by the confusion matrix, we introduce an Attention Behavior Matrix to comprehensively analyze attention head behaviors in both success and failure scenarios. By maximizing the signal-to-noise ratio -- strengthening valid reasoning pathways in success cases while inhibiting noise from bias and distraction in failure cases -- we propose REtrieval-reAsoning and Logic-constructed (REAL) KV cache eviction, the first method to leverage multi-behavior analysis. Comprehensive evaluations show that REAL achieves remarkable performance across various models and benchmarks; notably, on LongBench v2, it achieves comparable accuracy to the strongest baseline, HeadKV-R2, while requiring 32x less space (Figure 1). By offering a novel perspective on behavior analysis, we pave the way for a shift from success-only to comprehensive, failure-aware methods in long-context modeling. Our code is available at https://github.com/yonseicasl/REAL.

cs.CL↗

arXiv2026

Coherent manipulation of the biphoton generation in cavity-QED system

We theoretically investigate the coherent manipulation of biphoton generation via spontaneous four-wave mixing in a cavity-QED system with a single atom. The atom is driven by pumping, coupling, and driving fields, and the generation of the Stokes and anti-Stokes photons are enhanced by two cavities. By solving the master equation in the steady state, we analyze the spectral brightness, as well as the degree of the auto-correlation and cross-correlation. Our results show that when the pumping and driving fields are in two-photon resonance, the dark state established between the ground and Rydberg states. efficiently enhances the controllability of the driving field over the biphoton generation and the quantum statistics. In contrast, under large two-photon detuning, the control capability of the driving field is significantly reduced. The coupling field, which directly relates to the electromagnetically induced transparency, modifies the linewidth of the biphoton, while the atom-cavity coupling strength only changes the brightness without affecting the linewidth.

quant-ph↗

arXiv2026

CriticalKV: Optimizing KV Cache Eviction from an Output Perturbation Perspective

Large language models have revolutionized natural language processing but face significant challenges of high storage and runtime costs, due to the transformer architecture's reliance on self-attention, particularly the large KV cache for long-sequence inference. Recent efforts to reduce KV cache size by pruning less critical entries based on attention weights remain empirical and lack formal grounding. This paper presents a formal study on identifying critical KV cache entries by analyzing attention output perturbation. Our analysis reveals that, beyond attention weights, the value states within KV entries and pretrained parameter matrices are also crucial. Based on this, we propose a perturbation-constrained selection algorithm that optimizes the worst-case output perturbation to identify critical entries. We demonstrate that our algorithm is a universal, plug-and-play enhancement that incurs negligible computational overhead. When integrated with three state-of-the-art cache eviction methods on three distinct LLMs, our algorithm significantly reduces the compression loss by more than \textit{half} on average across 29 datasets from the Ruler and LongBench benchmarks. Further perturbation analysis, at both the head and layer levels, confirms the principles underlying our effectiveness. This work offers a new, formally grounded perspective to cache eviction , opening promising avenues for future research. The code is publicly available at https://github.com/FFY0/DefensiveKV.

cs.CL↗

arXiv2026

Effects of magnetically driven shocks on nucleosynthesis and kilonovae from neutron star mergers

Neutron-star mergers can launch mildly relativistic to moderately relativistic outflows whose interaction with the ejecta can reshape kilonova emission. We parametrically study magnetically powered outbursts from long-lived merger remnants, such as flare-like eruptions and collapse-driven shocks, and quantify their impact on ejecta dynamics, composition, and observables. Using two-dimensional special-relativistic magnetohydrodynamic simulations, we follow magnetized blast waves injected into expanding merger ejecta for early- and late-launch scenarios across a range of shock strengths. We then post-process Lagrangian tracer histories with the nuclear reaction network WinNet and the radiative-transfer code SuperNu with realistic opacities, to connect shock heating directly to nucleosynthesis and kilonova light curves. We find that sufficiently strong shocks can reheat portions of the ejecta to nuclear statistical equilibrium, increase the electron fraction in the shocked material, and deposit entropy, leading to systematic changes in $r$-process yields. These thermodynamic and compositional changes can leave observable imprints on kilonova emission -- especially in color evolution and late-time light-curve behavior -- indicating that magnetically driven remnant variability can potentially contribute to kilonova diversity.

astro-ph.HE↗

arXiv2026

4th Workshop on Maritime Computer Vision (MaCVi): Challenge Overview

The 4th Workshop on Maritime Computer Vision (MaCVi) is organized as part of CVPR 2026. This edition features five benchmark challenges with emphasis on both predictive accuracy and embedded real-time feasibility. This report summarizes the MaCVi 2026 challenge setup, evaluation protocols, datasets, and benchmark tracks, and presents quantitative results, qualitative comparisons, and cross-challenge analyses of emerging method trends. We also include technical reports from top-performing teams to highlight practical design choices and lessons learned across the benchmark suite. Datasets, leaderboards, and challenge resources are available at https://macvi.org/workshop/cvpr26.

cs.CV↗

arXiv2026

VA-FastNavi-MARL: Real-Time Robot Control with Multimedia-Driven Meta-Reinforcement Learning

Interpreting dynamic, heterogeneous multimedia commands with real-time responsiveness is critical for Human-Robot Interaction. We present VA-FastNavi-MARL, a framework that aligns asynchronous audio-visual inputs into a unified latent representation. By treating diverse instructions as a distribution of navigable goals via Meta-Reinforcement Learning, our method enables rapid adaptation to unseen directives with negligible inference overhead. Unlike approaches bottlenecked by heavy sensory processing, our modality-agnostic stream ensures seamless, low-latency control. Validation on a multi-arm workspace confirms that VA-FastNavi-MARL significantly outperforms baselines in sample efficiency and maintains robust, real-time execution even under noisy multimedia streams.

cs.RO↗

arXiv2026

TrustGeoGen: Formal-Verified Data Engine for Trustworthy Multi-modal Geometric Problem Solving

Geometric problem solving (GPS) requires precise multimodal understanding and rigorous, step-by-step logical reasoning. However, developing capable Multimodal Large Language Models (MLLMs) for GPS is heavily bottlenecked by the scarcity of high-quality, verifiable data. Existing data acquisition paradigms either suffer from modality incompleteness and unverified logical gaps ("leaps-of-faith"), or rely on formal engines that generate rigid, structurally homogeneous data, failing to produce high-difficulty problems or foster genuine natural-language reasoning. To overcome these limitations, we introduce TrustGeoGen, an autonomous and formalized geometric data generation engine. TrustGeoGen strictly guarantees reasoning trustworthiness through formal verification while generating multimodally integrated data, including premises, visual diagrams, and solutions. To systematically scale problem difficulty, we incorporates difficulty-aware filtering and iterative bootstrapping mechanism. Furthermore, we propose "connection thinking" to bridge the semantic gap between rigid formal logic and fluent human-like reasoning, ensuring coherent logical transitions. We also introduce the GeoExplore family of sampling algorithms to extract diverse problem-solving trajectories based on various thinking templates. Extensive experiments demonstrate that training models on our synthesized dataset, GeoTrust, substantially enhances deep geometric reasoning capabilities and yields significant performance gains across out-of-distribution (OOD) benchmarks, including GeoQA, Geometry3K, and OlympiadBench.Our code and data can be found at https://github.com/InternScience/TrustGeoGen

cs.AI↗

arXiv2026

Social Robots for People Living with Dementia: A Scoping Review on Deception from Design to Perception

As social robots are increasingly introduced into dementia care, their embodied and interactive design may blur the boundary between artificial and lifelike entities, raising ethical concerns about robotic deception. However, it remains unclear which specific design cues of social robots might lead to social robotic deception (SRD) in people living with dementia (PLwD), and which perceptions and responses of PLwD might indicate that SRD is taking place. To address these questions, we conducted a scoping review of 26 empirical studies reporting PLwD interacting with social robots. We identified three key design cue categories that might contribute to SRD and one that might break the illusion. However, the available literature does not provide sufficient evidence to determine which specific design cues lead to SRD. Thematic analysis of user responses reveals six recurring patterns in how PLwD perceive and respond to social robots. However, conceptual limitations in existing definitions of robotic deception make it difficult to identify when and to what extent deception actually occurs. Building on the results, we propose a dual-process interpretation that clarifies the cognitive basis of false beliefs in human-robot interaction and distinguishes SRD from anthropomorphism or emotional engagement.

cs.HC↗

arXiv2026

Refinement orders for quantum programs

Refinement is a fundamental technique in the verification and systematic development of computer programs. It supports a disciplined approach to software construction through stepwise refinement, whereby an abstract specification is gradually transformed into a concrete implementation that satisfies the desired requirements. Central to this methodology is the notion of a refinement order, which guarantees that each refinement step preserves program correctness. This paper presents the first comprehensive study of refinement orders for quantum programs, covering both deterministic and nondeterministic settings under total and partial correctness criteria. We investigate three natural classes of quantum predicates: projectors, representing qualitative properties; effects, capturing quantitative properties; and sets of effects, modeling demonic nondeterminism. For deterministic quantum programs, we show that refinement with respect to effect-based and set-of-effects based specifications coincides with the standard complete-positivity order on superoperators, whereas refinement induced by projector-based specifications can be characterized by the linear span of Kraus operators. For nondeterministic quantum programs with set-of-effects based specifications, we establish precise correspondences with classical domain-theoretic notions: the Smyth order characterizes refinement under total correctness, while the Hoare order characterizes refinement under partial correctness. Moreover, effect-based and projector-based specifications lead to strictly weaker refinement orders.

cs.LO↗