Papers

Let $M$ be a closed manifold equipped with a collection of smooth real vector fields $X_1,\ldots, X_k$ and a smooth measure $\mu$. Let $P:=\sum_j X_j^\dagger X_j $ be the associated sublaplacian. Under the assumption that each pair of points of $M$ can be connected by a path obtained by joining integral curves of the vector fields, the equation \[ Pu=f \] admits unique zero-average solutions $u$ in the natural energy space, for all zero average data $f\in L^2$. We say that exact regularity holds for the above equation if $u\in H^k(M)$ whenever $f\in H^k(M)$, where $H^k(M)$ is the $L^2$ based Sobolev space of any order $k\in \mathbb{N}$. We investigate how the presence of a characteristic submanifold, namely a submanifold tangent to all vector fields $X_j$, may cause a failure of exact regularity. We prove that this indeed happens for a class of "worm sublaplacians", which are real analogues of Kohn Laplacians on Diederich–Fornaess worm domains, of interest in several complex variables. Our main tool is a scaling lemma inspired by work of Barrett and Christ on the $\bar\partial$-Neumann problem.

We present WebFovea, a vision-based web agent that placed 2nd in the WebRetriever Challenge 2026 with a final score of 57.0 out of 100. The challenge evaluates agents end to end on Protocol III of the WebRetriever benchmark (arXiv:2607.06118): starting from an entry URL on a live website, the agent must operate the site's own interface and return a verifiable answer. A capable multimodal large language model (LLM) is necessary for this, but not sufficient. The model's decisions reach the browser through the harness, the code between the model and the page. At every step, four things must go right: the model's reply must be parsed into the intended action, the action must take effect on the page, the result must be reported back accurately, and the model must be shown the information it needs. On real websites, many of the failures we observed occurred at one of these four stages rather than in the model's reasoning. A coordinate-space mismatch placed every click at 3/4 of its intended coordinates; actions on native dropdowns, inside iframes, and in text boxes failed silently; and self-generated chat-template tokens contaminated 4.9% of task episodes. WebFovea hardens each stage and surrounds the loop with guardrails that keep the agent within the rules and its budget. The four-stage view does not depend on the model, although some individual fixes do. Because we used the same model in all four submissions, the rise of our official hidden-set score from 31.0 to 57.0 reflects changes to the harness, up to run-to-run variance on live sites. We describe the design, the evidence for each component (including negative results), a failure analysis, the limitations, and a roadmap that includes routing different steps to different models. Code is available at https://github.com/jianganghan/WebFovea.

Multimodal learning often suffers from modality imbalance, where the joint optimization process is dominated by a single modality. Existing methods typically estimate modality imbalance from score disparities derived from prediction uncertainty or optimization statistics. However, due to distinct prediction uncertainty and learning dynamics across modalities, direct comparison of such scores may misinterpret intrinsic modality differences as progress gaps, leading to biased imbalance estimation. In this paper, we propose Function-Space Guided Multimodal Optimization (FGMO), which leverages a function-space progress signal to assess modality-wise optimization progress and coordinate optimization across modalities to alleviate modality imbalance. Specifically, we introduce Functional Progress Estimation (FPE) to measure each modality's update-induced function-space response and calibrate it against a loss-aligned unimodal reference, producing a comparable progress signal. Based on this signal, Functional Response Control (FRC) redistributes modality-level function-space budgets and realizes the target responses through tensor-wise learning-rate adjustment. Theoretical analysis establishes a one-step target-contraction property of FRC under bounded controller-state mismatch, and extensive experiments demonstrate the effectiveness of FGMO across multiple multimodal benchmarks.

Diffusion models rely on numerical solvers requiring time-discretization, which has a large influence on the tradeoff between sampling cost and quality. However, the computational difficulty of the reverse process varies along the sampling trajectory and across data distributions, making the choice of discretization important. We adapt proportional-integral (PI) step-size control to diffusion, using our diffusion noise-normalised error estimator. Unlike existing adaptive methods in diffusion that respond only to the current error, the PI solver also incorporates the previous error, yielding smoother step adaptation. We further show that these per-sample trajectories exhibit shared structure and can be aggregated into a fixed schedule that retains much of the benefit of adaptive sampling. We evaluate both approaches on natural-image and language datasets, in terms of quality, measured by FID at a matched number of neural network evaluations (NFE), comparing them with widely used stochastic solvers and schedules. For images, our fixed discretization outperforms the commonly used EDM schedule in terms of sample quality when used with the stochastic Heun sampler, and with the EDM-churn sampler at low NFE. Additionally, our PI adaptive solver obtains better FID than most stochastic and adaptive baselines, although it does not beat the EDM-churn sampler at low NFE. Moreover, we find our solver outperforms both the EDM and the entropy schedule on language diffusion at low-to-medium NFE in terms of perplexity, with the drawback of lower token entropy. Lastly, we find that the benefit of per-sample adaptivity is problem-dependent. It is highly beneficial in 1D toy examples, while only marginal for image and language data, where the average schedule sometimes even outperforms the PI-adaptive solver. Code is available at https://github.com/ellakemperman/adaptive-second-order-diffusion-solvers

Large language model (LLM)-based agents increasingly operate in multi-agent systems (MAS) characterised by strategic interaction. However, little is known about whether, and to what extent, different types of messages affect the outcomes of strategic games. By investigating AI agents based on four popular LLMs, playing four games with different cooperation equilibria, we study whether messages of different kinds (natural language, numerical signals, or random sequences) significantly modify the levels of cooperation in each game, also depending on the agents' assigned personalities. We observe that structured messages alter the final payoffs for most games and LLMs, but without a predictable pattern; this challenges the assumption that AI agents can converge to stable equilibria regardless of additional capabilities. Moreover, we observe that agent-generated numerical messages depart from randomness, most strongly and consistently when agents are explicitly instructed to communicate; however, they introduce an additional interpretability challenge, as their symbol distributions are mostly associated with the payoff structure and typically become more concentrated with repetition, but are overall difficult for humans to interpret. Monitoring for coordination of AI agents through restricted channels should thus prioritise message-level fingerprints, which generalise across models, over behavioural decisions, which do not.

This paper investigates the secrecy performance of a cooperative non-orthogonal multiple access (NOMA) network under practical impairments. In the considered system, the source employs orthogonal space-time block coding (OSTBC) transmission, legitimate users adopt receive antenna selection (RAS), and the eavesdropper utilizes maximal-ratio combining (MRC) scheme. Moreover, an energy-constrained amplify-and-forward relay harvests energy from the received signal and allocates the harvested energy to information forwarding and artificial noise (AN) generation. To capture the realistic performance of a more practical transmission scenario, channel estimation errors (CEE), feedback delay (FBD), imperfect successive interference cancellation (SIC), and imperfect AN cancellation are taken into account. For this system, closed-form expressions for the legitimate-user signal-to-interference+noise ratios (SINR) cumulative distribution function (CDF) and the eavesdropper SINR probability density function (PDF) are derived, and a tractable analytical secrecy outage probability (SOP) expressions are obtained. In addition, the high-SNR secrecy behavior is characterized under ideal channel state information (CSI) conditions and practical impairments, revealing the achievable secrecy diversity order and the secrecy outage floor, respectively. Moreover, secrecy throughput and energy efficiency are analyzed based on the obtained SOP results. The numerical findings are in close agreement with the analytical results and further illustrate how the secrecy behavior of the proposed system is affected by key design parameters and practical impairments. The obtained results indicate that transmit and receive diversity and proper AN power allocation significantly improve the secrecy performance, whereas practical impairments lead to non-zero SOP floors in the high signal-to-noise ratio region.

Monocular semantic scene completion (SSC) for quadruped robots remains underexplored in real crowded indoor environments, where human-scene occlusion disrupts static geometry and human occupancy predictions are often incomplete or spatially misplaced. We present CrowdOcc, an RGB-D dataset and monocular SSC framework for this setting. CrowdOcc contains 25.1K frames from 11 indoor scenes, with semantic occupancy annotations constructed through static dynamic decoupling. Our framework combines: (i) Normal Guided Scene Geometry Fusion (NGSGF) to complement depth-aware lifting with surface-normal cues for occlusion robust geometry; and (ii) Human-Centric Sparse Interaction (HCSI) to selectively model human-human and local human scene relations in 3D. Our method achieves state-of-the-art SSC performance on CrowdOcc's scene-disjoint test set, reaching 15.80 IoU, 11.40 mIoU, and 46.23 Human IoU, demonstrating generalization to unseen indoor scenes.

The matroid parity problem generalizes both graph matching and matroid intersection. Although the general version is intractable, Lovász (1981) developed a polynomial-time algorithm for the linear matroid parity problem, assuming the availability of matrix representations. Subsequently, Gabow and Stallmann (1986) presented an augmenting path algorithm, which has long been recognized as one of the fastest deterministic algorithms. Since shortest augmenting paths improved algorithms for graph matching (Micali & Vazirani, 1980) and linear matroid intersection (Cunningham, 1986), extending these techniques to linear matroid parity appears to be a natural progression. However, such an algorithm has remained elusive for four decades. We present the first shortest augmenting path algorithm for linear matroid parity. Our approach synthesizes the augmenting path algorithm of Gabow and Stallmann with the synchronized blossom formation of Micali and Vazirani, whose correctness was recently proven by Vazirani (2024). Our key technical contributions are threefold: (i) a linear-algebraic argument that bounds the lengths of shortest augmenting paths for linear matroid parity, which generalizes Cunningham's bound for linear matroid intersection; (ii) an a priori characterization of shortest search paths through lower bounds on their lengths; and (iii) an extension of the structural properties for graph matching established by Izumi, Kitamura, and Yamaguchi (2025) to the linear matroid parity setting. Our algorithm deterministically solves the linear matroid parity problem in ${\rm O}(nr^2\log r)$ time, where $n$ is the ground set size and $r$ is the matroid rank. By incorporating fast matrix multiplication, this complexity can be further reduced to ${\rm O}(nr^2)$. These results improve upon the long-standing deterministic bounds of ${\rm O}(nr^3)$ and ${\rm O}(nr^\omega)$.

Gene set analysis is a cornerstone of functional genomics, yet it remains labor-intensive and heavily dependent on manual curation and expert biological interpretation. While Large Language Models (LLMs) have emerged as powerful tools for genomic reasoning and annotation, most existing approaches rely on symbolic gene names and fail to capture domain-specific biological structure, particularly protein sequence information that governs molecular activity, interactions, and downstream gene function. In this work, we propose SoftGene, a novel framework for LLM-based gene set annotation that leverages the hierarchical structure of gene sets. First, we use a hierarchical attention-based encoder built on ESM, a protein language model, to represent each gene set using protein-level amino acid sequence information. Second, we construct a hybrid prompting scheme that combines soft prompts derived from gene set embeddings with hard prompts containing auxiliary context generated by an LLM, and feed the resulting prompt into a local LLM for annotation. We evaluate our framework on two benchmark datasets: Gene Ontology (GO) and the Molecular Signatures Database (MSigDB). Our results show that integrating protein-sequence representations with textual context improves gene set annotation overall, while per-domain analyses reveal that the contribution of protein embeddings varies across biological domains.

We consider the $\phi^4$ theory with $\phi(x)\in\mathbb{R}$ in three Euclidean dimensions. For a variety of self-couplings $\hat{\lambda}$, the critical (bare) parameter $\hat{\mu}_{0\mathrm{c}}^2(\hat{\lambda})$ is determined where the lattice-regulated system changes from the symmetric phase to the broken phase. Next, we send the box volume to infinity and switch to a universal scheme; this yields $\hat{\mu}_\mathrm{c}^2(\hat{\lambda})$ for each simulated $\hat{\lambda}$. Finally, a continuum extrapolation gives $f_\mathrm{c}=\lim_{\hat{\lambda}\to0} \hat{\lambda}^2/\hat{\mu}_\mathrm{c}^2(\hat{\lambda})$, and we find $f_\mathrm{c} = 18.808(64)_{\mathrm{stat}}(45)_{\mathrm{sys}} = 18.808(78)_{\mathrm{tot}}$. In addition, $\Delta\langle\phi^2\rangle_\mathrm{c}/\lambda = -0.00303(12)_{\mathrm{stat}}(12)_{\mathrm{sys}} = -0.00303(17)_{\mathrm{tot}}$ is determined in the combined continuum and infinite-volume limit.

The key-value (KV) cache becomes a major memory bottleneck in long-context LLM inference, placing substantial pressure on memory capacity and bandwidth. To mitigate this bottleneck, vector quantization (VQ) has emerged as a promising approach for aggressive KV cache compression. However, existing VQ methods degrade substantially in the 1-bit regime. At such extreme compression, each codebook must represent a larger group of channels with a limited set of centroids, making effective use of its capacity increasingly challenging. To address this, we introduce $\textbf{TaSQ}$, which tailors the VQ target space by combining query-guided channel weighting, cross-head normalization, and covariance-aware channel grouping to better reflect the error sensitivity and statistical structure of cached activations. Since these transforms are RoPE-compatible and can be easily merged into projection weights and codebooks, TaSQ preserves the conventional VQ lookup structure and adds negligible serving overhead. Across general, long-chain-of-thought reasoning, and long-context retrieval benchmarks, TaSQ consistently outperforms existing low-bit KV cache VQ baselines while preserving reasoning stability. On a single RTX 6000 Ada GPU, its SGLang implementation supports up to $14\times$ larger batch sizes and achieves $1.87\times$ higher peak throughput compared to the BF16 baseline.

In a world fraught with uncertainty, it behooves us to consider the most disastrous possible outcomes, in order that we can best mitigate against them. Often these are extreme risk events – of high consequence, but low probability. Yet, the rarity of such events limits our ability to study them; there is a paucity of past occurrences to draw upon, and sampling such rare events from models is typically costly in time. Here, we introduce an approach for rare-event sampling based on quantum stochastic models. This integrates the memory advantages of quantum stochastic modelling – and its associated efficient means of preparing quantum sample states – with the `speed-ups' of quantum amplitude amplification, to yield simultaneous reductions in spatial and temporal computational complexity compared to classical Monte Carlo-based approaches.

What makes a short story gripping; a news article newsworthy; or a math proof elegant? These constructs resist articulation or verification; their meaning is at least partially tacit. However, modern AI models are improved primarily via articulated constitutions, rubrics and verifiers (i.e. in RLAIF and RLVR); tacit components of preferences are typically understudied. We introduce a large, labeled preference dataset CreativePreferences, containing 2.8M texts labeled by 317M human preference judgments across 7 creative domains, with 42 benchmark tasks. We model these labels with executable programs, rubric banks and densely trained models (V, A and VAT, respectively). We observe robust articulability gaps, VAT-VA; and verifiability gaps, VAT-V; we estimate upper and lower bounds for each gap with a novel measurement approach that discovers articulable and verifiable metrics, identifies spurious variables and estimates the value of undiscovered metrics using capture-recapture. These gaps occur across all domains, even in domains traditionally treated as fully verifiable: correctness-centered domains (i.e. mathematics and software engineering) and claim- and novelty-centric domains (i.e. news, patents, peer review). The size of the gap varies based on domain (e.g. peer review and creative writing have the largest articulability gaps) and widens as more people take part in the judgment, consistent with Collins' collective tacit knowledge. We show two consequences: (1) on human generations, the full model more closely matches human preferences, often in disagreement with articulated criteria, and (2) in an analogy to Goodhart's law, articulating preference shifts it away from the tacit dimension. Articulability and verifiability gaps are consequential; we give recommendations on when tasks can be prompted; how learning mechanisms might improve; and when to leave judgments with humans.

Quantum entanglement in $\tau^+\tau^-$ production provides a direct probe of non-classical spin correlations in a relativistic quantum system, with the $\tau$ decay products serving as spin analyzers. We investigate the detector-level feasibility of such measurements in $e^+e^-\to\tau^+\tau^-$ at the Super Tau-Charm Facility (STCF) at $\sqrt{s}=7$ GeV, using the $\rho\rho$ and $\pi\rho$ decay channels. Signal and dominant background processes are simulated with the full STCF detector and reconstruction framework. Dedicated reconstruction and event-selection procedures yield overall signal efficiencies of approximately $4.5\%$ and $4.8\%$, with corresponding purities of about $87.6\%$ and $93.2\%$ for the $\rho\rho$ and $\pi\rho$ channels, respectively. From the reconstructed decay kinematics, the $\tau^+\tau^-$ spin-density matrix is inferred and the concurrence $C[\rho]$ together with the Bell-sensitive quantity $m_{12}[C]$ are evaluated in a selected fiducial region. For the benchmark requirement $|\cos\theta|<0.2$ and an integrated luminosity of $1~\mathrm{ab}^{-1}$, the projected relative precision on $C[\rho]$ is about $2.2\%$ in both channels, while that on $m_{12}[C]$ is approximately $1.4\%$ for $\rho\rho$ and $1.5\%$ for $\pi\rho$, including statistical and detector-resolution uncertainties. The reconstructed $m_{12}[C]$ shows a larger separation from its reference threshold in the $\rho\rho$ channel, making this topology particularly sensitive to Bell-type spin correlations. These results demonstrate the detector-level feasibility of probing quantum entanglement and non-classical spin correlations in hadronic $\tau$ decays at STCF.

We develop a representation-theoretic approach to determine the Morse index and nullity of an embedded minimal surface based only on its topology and symmetry group. The method combines an equivariant comparison between the second variations of area and energy with the representation theory of the dihedral group. In the Euclidean unit ball we compute for every $k\geq3$ the Morse index and nullity of any free boundary minimal $k$-noid with prismatic symmetry of order $4k$, and of any free boundary minimal surface of genus one with connected boundary and antiprismatic symmetry of order eight. In the round sphere we obtain an alternative computation for the Morse index and nullity for the first family of the Lawson surfaces, using only half of their symmetry group. The results are independent of any particular construction of the surfaces.

Simultaneous Sign Language Translation (SLT) is critical for real-time communication, yet existing methods remain largely confined to sentence-level, offline settings that assume pre-segmented inputs. These assumptions hinder deployment in realistic scenarios involving continuous, unsegmented video streams. We present ReSCUE, a unified framework for simultaneous SLT on unsegmented long-form sign language videos that aligns training and inference with realistic streaming conditions. ReSCUE combines inference-aware training to handle partial inputs, non-signing pauses, and multi-sentence contexts, stabilized re-translation to enable low-latency yet revisable predictions with reduced output flicker, and a sentence commitment mechanism for online segmentation and memory management. Experiments on standard sentence-level benchmarks show that ReSCUE achieves lower latency and the best translation quality under low-latency settings. On long-form unsegmented datasets, ReSCUE approaches the translation quality of oracle offline systems that use ground-truth sentence boundaries, while operating at substantially lower latency, demonstrating its practicality for real-world streaming scenarios.

This paper presents a construction of a small universal machine within the framework of Multi-set Reaction Systems. We encode the machine state using elements that represent registers and instruction labels, and we enforce sequential execution by ensuring that reaction chains for a given instruction can only activate after the corresponding label element appears. We implement a universal register machine with 8 registers and 23 instructions, showing that the priority-based model requires 87 distinct elements. The results demonstrate that multiset reaction systems are capable of simulating universal computation.

Long-term agents must remember not only what is true about a user, but also how a particular agent should work with that user as their shared history evolves. Existing benchmarks primarily supervise user facts and preferences or experience reusable across users, leaving this relationship-specific agent memory implicit. Additionally, most prior works measure the model solely with final-answer QA over long interaction histories, making the assessment still incomplete and unreliable. To this end, we introduce DyadMem with the proposed new definition User-conditioned Relational Agent Memory (URAM). DyadMem jointly annotates user-side memory and URAM along the same multi-session trajectories, resulting in 6 memory categories. To summarize, it includes 3,065 episodes, 50,961 sessions, and 61,210 QA instances, with extensive session-level Capture and Update gold annotations, query-level Recall support, and two QA settings: Gold-Memory and Full-Pipeline. Across 16 open-weight and 4 proprietary models, Gold-Memory QA is consistently strong, yet Full-Pipeline QA drops sharply. Such a gap explicitly supports our fine-grained evaluation design. Additionally, several quantitative results further reveal low Capture recall, incomplete Recall, and unsafe-deletion issues arising from even the frontier LLMs. We further conduct a rigorous experiment to validate the effectiveness of our URAM and observe the positive effects for all 20 models. In summary, DyadMem is a dual-domain, full-pipeline memory benchmark with extensive annotation efforts for advancing the domain's development.

Hybrid statistical-neural pipelines often assume that a successful statistical first stage leaves a cleaner and more learnable residual target. We examine that assumption in short-horizon volatility forecasting through a signal-forecast-system diagnostic framework. Across five liquid U.S. assets, a volatility-aligned HAR-style model outperforms AR, MA, and ARIMA. Within expanding training windows, the pre-standardization fitted residual process used to construct residual-LSTM sequences has about 82% lower variance than the corresponding target and near-zero lag-1 autocorrelation; independently, rolling pseudo-out-of-sample HAR errors show about 74% variance reduction and similarly weak lag-1 dependence. Residual-only LSTM augmentation nevertheless raises mean squared error from 0.3049 to 0.3594 on average, with deterioration on every asset. Pure LSTM records the lowest selected pseudo-out-of-sample MSE, 0.2649, while the residual hybrid requires substantially more end-to-end runtime without improving accuracy. We describe this pattern as forecaster-preconditioner asymmetry: first-stage forecasting success and statistical residual simplification need not translate into useful downstream neural preconditioning.

Fission fragments are born rotating rapidly, but the evolution of this spin with excitation energy and its impact on fission observables remain largely unknown. Combining state-of-the-art microscopic and statistical reaction theory, we simulate the full fission process - from the initial state, through scission, to fragment decay - and report the first microscopic, energy-resolved spin and parity distributions of the fragments produced in $^{239}$Pu(n,f). From thermal energy to $E_n = 10$ MeV, the spin of individual fragments evolves non-monotonically, tracking their deformation. The parity content, strongly non-statistical at thermal energy, remains correlated with spin throughout this range. For the first time, we quantify the impact of microscopic spin-parity distributions on prompt photon multiplicities, demonstrating the predictive potential of parameter-free microscopic inputs.

This work presents an automatic speech recognition (ASR) system personalized for a Czech speaker with a permanent tracheal stoma and severe dysarthria rendering their speech unintelligible to untrained listeners. We release a public dataset containing 33 annotated hours of the speaker's speech, collected using a novel "artificial conversation" protocol designed for high engagement and dialogue realism. We propose a multi-stage training pipeline based on Whisper Base: fine-tuning on standard Czech speech, acoustically simulated tracheostomic speech, and the speaker's data. We evaluate the system across three near real-time scenarios: scripted conversations, question answering, and spontaneous dialogue, achieving a 50% relative reduction in Character Error Rate compared to Whisper Base baseline and surpassing the average recognition accuracy of their assistants in acoustic recognition of isolated utterances. We demonstrate that even for severely impeded speech, a helpful ASR is achievable, as evidenced by the quantitative results and the feedback from the speaker.

In this paper, we propose a Role-aware Stimulus-guided (RASG) framework for interlocutor emotion recognition, which predicts listener emotions from listener-only videos and speaker-only audios. RASG consists of Role-aware Visual Transfer (RVT) and Stimulus-guided Boundary Reasoning (SBR) modules, which address supervision mismatch due to the lack of labeled listener data and ambiguity among visually similar listener reactions whose interpretation depends on speaker context, respectively. More specifically, RVT selects speaker samples whose facial expressions support their emotion labels. It then filters listener tracks and uses reliable pseudo-labels to train a listener-centric visual expert. SBR uses a two-class language reasoner only when the visual model is uncertain. It treats speaker audio and text as context rather than direct emotion evidence to distinguish similar listener reactions. Experiments conducted on MER-Cross dataset shows that RASG achieves 76.25% on MER-Cross and improves the performance of the baseline over 17%. Our team ranks second in Track 1 (MER-Cross) of the MER Grand Challenge at ACM MM 2026.

Accurate 3D detection is essential for mobile embodied agents, while Vision Foundation Models (VFMs) offer transferable visual and geometric priors. Yet existing VFM-based 3D detectors rely on narrow-view monocular images or discrete perspective views, limiting coherent surround perception; equirectangular projection (ERP) instead encodes a continuous 360 scene in a single image. Direct transfer remains difficult because ERP organizes geometry and visual information differently, making object-relevant cues hard to model, localize, and preserve. We propose OmniAct3D, a framework that adapts perspective-trained VFM detectors to ERP while preserving transferable VFM priors. To resolve geometric mismatch, the ERP-Ray Geometry Adapter (ERGA-Ray) models spherical viewing rays and periodic spatial structure. To localize evidence in scene-wide context, the Visual-Action Reasoning Chain (VARC) grounds each hypothesis in relevant panoramic evidence and converts it into a structured geometric action. To recover local cues lost under fixed token budgets, the Appearance-Guided Heading Expert (AGHE) re-encodes object regions at higher resolution for heading estimation. Experiments show that OmniAct3D improves over the previous best 3D detector by 2.96 NDS points on Spheriverse and over the unadapted VFM baseline by 24.87 mAP points on PanoMMOcc. With target-specific geometry adaptation, VARC retains 95–98% of the same-configuration mAP, indicating reusable object-level 3D reasoning across sensing configurations. The source code will be made publicly available at https://github.com/FeiT-FeiTeng/OmniAct3D.

Large language model (LLM)-based agents increasingly connect model-generated decisions to security-sensitive software capabilities such as command execution, filesystem access, network communication, browser control, and external tools. Existing analyses often use predefined sensitive operations as anchors, but operation identity alone is insufficient to determine security implications. We present AgentSecGraph, a security-aware static analysis framework that constructs a candidate-centered Security-Aware Agent Dependency Graph (Security-ADG) for each security-sensitive operation. It augments operation identity with agent relevance, source and dependency evidence, trust-boundary context, guard evidence, and external-effect semantics. We further introduce AgentSecBench, a corpus of 67 real-world LLM-agent repositories spanning 11 ecosystems and 37,542 source files. The current analyzer identifies 23,866 static security-sensitive operation candidates across 65 repositories and emits one Security-ADG artifact per candidate. Corpus-wide analysis recovers source-to-operation dependency evidence for 9,821 candidates (41.15%) and potential guard evidence for 3,075 (12.88%), completing in 50.8 minutes. Using a separate reproduction-backed evaluation layer, we establish 22 security-sensitive behaviors across 13 repositories: one confirmed vulnerability, one pending disclosure candidate, and 20 guarded behaviors. In nine held-out cases, Security-ADG preserves 91.1% of the reference context and all five observed guards, compared with 20.0% for a sink-only view and 40.0% for a simplified ADG. These results show that security-aware dependency and contextual evidence enable distinctions that cannot be recovered from sensitive-operation identity alone.

Category-level object pose estimation predicts the rotation, translation, and metric size of unseen instances within known categories. Many accurate RGB-D methods rely on external instance segmentation and crop-based pose estimation, introducing separate stages and object-dependent processing costs that hinder real-time inference. To bring accurate pose estimation into real time, we present \ours{}, an end-to-end trainable query-based RGB-D set predictor. It jointly detects and segments objects and estimates their \mbox{9-DoF} poses without explicit CAD-derived shape priors or a separately trained instance segmentor. Shared image and scene encoding avoids repeated per-object crop encoding. A query-conditioned geometry pathway associates observed 3D points and RGB features with object queries and incorporates shared scene context. Object-centric refinement uses the resulting point descriptors to update an explicit pose state through pose-conditioned cross-attention and recurrent residual corrections. On NOCS, \ours{} substantially improves on published RGB-D joint detection and pose estimation results. It achieves competitive performance compared with two-stage methods under all-object evaluation on REAL275 and HouseCat6D, while enabling real-time full-frame pose estimation at $31.8$ FPS on an RTX~A6000. Project page: https://yopo-series.github.io/RYOPO-project-page/.

Most human motion generation methods encode motion as tokens on a uniform temporal grid, where every token spans the same fixed time window. Human motion, however, is temporally heterogeneous: slowly evolving global trajectories coexist with rapid transient events such as foot contacts and joint impulses. Forcing such multi-scale dynamics onto tokens of identical temporal resolution entangles motion frequencies, leaving slow regions redundant while smoothing out the rapid details that distinguish realistic motion. We propose FreqMo, a scale-adaptive motion representation that decomposes motion into wavelet frequency bands, separating dynamics across temporal scales while preserving temporal localization and exact reconstruction. Unified Frequency Residual Quantization (UFRQ) then encodes all bands within a single shared codebook, compressing the token sequence threefold and enabling stable single-stage generation. Experiments show FreqMo attains SOTA fidelity with substantially improved high-frequency preservation, and the same decomposition transfers to continuous diffusion backbones.

In this paper, which originates from Chapter 4 of the author's PhD thesis [1, Chap. 4], we investigate the global well-posedness of a class of partially diffusive hyperbolic systems in critical homogeneous Besov spaces. Our main objective is to extend the analysis developed in our previous work [2] to a functional framework in which the high-frequency com- ponents of the solution are controlled in $L^p$-based spaces with $p > 2$. This broader framework allows us to establish global existence under weaker smallness assumptions on the initial data and provides more precise information on the qualitative behavior of the resulting solutions. In particular, our analysis yields refined regularity. As an application, our global existence theorem applies to the multidimensional compressible Fourier-Navier-Stokes system

Large language models (LLMs) are increasingly used in software engineering, including agentic systems that coordinate multiple agents, but impose higher computational and environmental costs. In this paper, we present a comprehensive empirical study of agentic LLM systems across five software engineering tasks: code generation, technical debt identification, code vulnerability detection, log parsing, and log analysis. For each task, we compare LLM configurations that range from a non-agentic single-query baseline to multi-agent workflows, using six open-weight LLMs, two prompt strategies, and three hardware platforms. We assess each configuration in terms of accuracy, inference latency, and energy consumption. Our results reveal substantial trade-offs between agentic complexity and energy efficiency: multi-agent designs consume on average 6.36$\times$ as much energy and run 6.07$\times$ as long as the non-agentic baseline, with worst-case slowdowns of up to 160$\times$ for individual task–hardware pairs. Accuracy gains from additional agents are limited and task-specific: multi-agent improves average vulnerability-detection accuracy, but lightweight non-agentic and single-agent configurations still dominate the Pareto front, accounting for 59 of 66 Pareto-optimal configurations. Model and prompt choice act as task-specific levers whose effective direction varies between tasks rather than as global defaults. We translate these findings into design guidelines for sustainable, task-aware LLM-based development tools.

Quantum entanglement, one of the most intrinsic properties of quantum mechanics, corresponds to a key resource for quantum information. Exceptional points (EPs) are special phase transition points in non-Hermitian systems. Nonreciprocity achieves one-way motion by breaking reciprocal symmetries. However, the connections among quantum entanglement, EPs, and nonreciprocity remain unexplored in non-Hermitian systems. In this paper, we investigate the manipulation of optomechanical entanglement by non-Hermitian coupling between two counter-propagating optical modes based on a whispering-gallery-mode (WGM) resonator coupled with two Rayleigh scatterers. EPs can periodically appear by controlling the relative angle between two nanoparticles. Moreover, the spinning resonator can induce opposite frequency shifts in two counter-propagating optical modes. The logarithmic negativity of entanglement reaches the maximum at EPs, leading to significant enhancement of optomechanical entanglement. In contrast, the entanglement is suppressed at non-EPs but can be recovered by the Sagnac-Fizeau shift. Specially, by driving the resonator at EPs while spinning it at the optimal rotation speed, we achieve switching between classical and quantum nonreciprocity.

Understanding how humans reason – and how reasoning responses vary across tasks and individuals – remains a core challenge for modeling and explanation in cognitive science. We investigate the stability of response patterns within reasoners and whether variation in these patterns can be used to predict learning effects. We introduce a formal, geometry-based method to quantify distances between individual reasoning patterns and their internal variability, grounded in heuristic theories. The proposed framework is tested against experimental data via generalized linear mixed-effects models and clustering, where we find that our proposed variation measure interacts with correctness to predict performance gains. Moreover, we find that reasoning patterns are stable over time within the same reasoner. The method is general enough to be applied to other reasoning domains.