Papers

We construct an explicit involution on Dyck paths satisfying $\defc\le\min(\area,\dinv)$ that interchanges area and dinv, where $\defc=\binom n2-\area-\dinv$ and $n$ is the semilength. In addition, we give an involution on Dyck paths with $\defc \le 2n-8$ that interchanges area and dinv. Finally, we give an explicit partition formula for the portion of the $q,t$-Catalan polynomial of total degree at least $\binom n2-2n+8$, thereby proving a conjecture of Lee and Li~\cite[Conjecture~4]{LeeLi11}. Altogether our results give a combinatorial explanation of $q,t$-Catalan symmetry in the region $\defc\le\max(2n-8,\min(\area,\dinv))$. Our constructions rely on a number of new combinatorial objects developed here and three main external tools: the dual Dyck insertion of \cite{Hawkes26}, the Garsia–Milne involution principle~\cite{GarsiaMilne81,Doyle19}, and the partition bijection of ~\cite{LoehrWarrington09}. We also explore rational analogs of these results, give an answer to Question~4.9 of Kiem and Lee~\cite{KiemLee24} and consider deficit up to $3n-15$ briefly.

Surgical instrument segmentation (SIS) plays a critical role in robotic assistance and surgical workflow analysis. However, most existing SIS methods formulate segmentation as a category-driven localization problem, limiting their ability to capture procedural context and task-dependent semantics in surgical workflows. We introduce Reasoning-Aware Surgical Instrument Segmentation (RA-SIS), a task formulation that frames segmentation as query-conditioned inference under surgical context. To benchmark this setting, we construct SurgRS, a surgical reasoning segmentation dataset consisting of 41,000 image-text pairs, which aligns instance-level masks with structured query-answer supervision to enable semantic grounding at the pixel level. Based on SurgRS, we propose Surgical Instrument Reasoning and Segmentation Assistant (SIRA), a multimodal framework that disentangles target-level and query-level semantics and integrates them with visual features through query-anchored dual alignment. By aligning query semantics with spatial features and segmentation prompts, SIRA enhances semantic-visual consistency in mask prediction. Extensive experiments on SurgRS demonstrate improvements over existing reasoning-aware baselines. Code is available at https://github.com/linxir226/SIRA.

Conversational AI systems produce fluent, socially appropriate responses, yet whether they participate in cooperative communication or merely simulate its surface forms remains unclear - a question central to how these systems are evaluated, trusted, and designed. This study investigates how morality, politeness, and alignment - three dimensions central to cooperative dialogue - function in human-AI interaction compared to human-human conversation. We analyze 15,881 human-ChatGPT and 10,784 human-human multi-turn dialogues, using mixed-effects models to identify which features predict turn-to-turn alignment. We observe a consistent dissociation: AI produces the surface features of cooperative communication without the underlying social architecture. Moral output appears preconfigured rather than negotiated; warmth is generated without face sensitivity; linguistic convergence declines persistently. Most strikingly, the cooperative mechanisms themselves reverse direction: hedging and softening associated with greater accommodation between humans are associated with reduced alignment when produced by AI, and purity framing associated with human divergence coincides with users converging toward the AI. Agency - giving users room to shape the exchange - is the most consistent predictor of alignment across both interaction types, while lower moral assertiveness in more recent models is not accompanied by better cooperation. Together these patterns suggest that AI reproduces the surface of cooperation without the mutual adaptation that grounds it between humans - and, more surprisingly, that mechanisms sustaining human accommodation can run in reverse with AI, suggesting a turn-level view may be insufficient for interaction-level success.

Open-vocabulary 3D scene mapping aims to build a persistent representation of the objects in an environment. Existing systems typically rely on engineered mapping pipelines to associate observations, merge information across views, and maintain a consistent scene representation over time. Many additionally store feature-rich object representations, such as embeddings or image crops, increasing the size and complexity of the persistent memory. We introduce SceneLM, a Scene-Language Model that directly maintains a textual scene map. The full scene is represented as a structured text list of objects, which serves as the model's only persistent memory. For each input image, the model reads the current scene state and updates the map by adding, editing, and removing objects. To learn this behavior, we introduce supervision tasks for iterative scene map maintenance together with an automatic annotation pipeline that generates training data from images without human labels. We evaluate SceneLM on both a language-grounded retrieval benchmark and a localization benchmark. Across both benchmarks, the model produces a scene map that achieves competitive performance with complete mapping systems built from dedicated perception and geometric modules while producing a scene representation that is 6-12x more compact. We further show that SceneLM can be run online on an edge device through experiments on a quadruped. These results show that a persistent open-vocabulary 3D scene map can be maintained directly by a single vision-language model using only a lightweight text representation. Training and inference code is available on https://goldengait.github.io/scenelm/.

We establish stability principles under compact gluing for the reverse square-root inequality and the Riesz transform. Below a common Sobolev dimension, the reverse estimate passes from finitely many model manifolds to their connected sum. A boundary-compatible Poisson parametrix also proves the Neumann reverse inequality for every $1<p<\infty$ for real symmetric uniformly elliptic divergence-form operators on exterior Lipschitz domains. For the forward transform, we glue a bounded-core model and an independent whole-space end model that agree with the coefficients on overlapping regions. A homogeneous Fredholm construction then resolves the form-valued error. The resulting range is controlled by the two model exponents, with the additional restriction $p<n$ for Dirichlet data. More broadly, our method illuminates the interplay between the bounded interior and exterior problems, and provides a flexible framework for studying other boundary value problems.

The bit thread formulation provides a geometric description of holographic entanglement entropy, while partial entanglement entropy (PEE) and PEE threads resolve this structure with respect to individual boundary points. Motivated by the fact that PEE thread flows can be superposed to reconstruct conventional bit thread flows, we introduce a differential form description of PEE thread flow. For an interval in the vacuum AdS$_3$/CFT$_2$ setup, we show that the flux of the resulting form reproduces the known entanglement contour, providing a consistency check of the proposed description. We then investigate whether the PEE form can be related, possibly up to an exact form improvement, to a current constructed from the covariant phase space (CPS) formalism. As a first test, we consider exact Killing vectors in Rindler-AdS$_3$. The Rindler parameter $a$ parametrizes a one-parameter family of backgrounds. At each value of $a$, the Killing vector of the corresponding background is substituted into the Iyer-Wald surface charge form, and the full $a$-dependence of the resulting form is retained. We define a finite candidate current by integrating this surface charge form over $a$. For the reflection-symmetric PEE flow sourced at the central boundary point $r_0=0$, we find that no choice of the Killing parameters reproduces both components of the known PEE flow after transforming to Poincaré-AdS$_3$. This result shows that the direct Killing vector CPS construction considered here is not sufficient to reproduce the central PEE flow. Possible extensions include an exact form improvement, a different CPS current, or a more general choice of generator.

Let $(a,b,c,d) = \left(2mn, 2mp, m^2 - n^2 - p^2, m^2 + n^2 + p^2\right)$ be a primitive Pythagorean quadruple and let $S=\langle a, b, c, d\rangle$ be the numerical semigroup generated by $a,b,c,$ and $d.$ For convenience, we also let $Q = n^2 + p^2, \delta = \gcd(n,p),$ and $n = \delta n_0$ for some $n_0 \in \mathbb{Z}$. In this paper, we determine the Frobenius number of $S$ and derive an explicit formula in terms of $m$, $n$, and $p$ with the assumption that $m\geq 2Q$ for $\delta = 1$ and $m \geq \frac{2Q}{\delta} - 1$ for the remaining cases. The proof is based on an explicit complete residue system modulo $2mn$, a normalization procedure for arbitrary semigroup elements, and a lift-orbit description of boundary representatives.

Underwater robot development is often hindered by the complexities of waterproofing and wiring, which significantly delay the rapid prototyping process. This paper presents MarineCraft, a modular toolkit designed to accelerate the development cycle through structural reconfiguration. The system features self-contained, waterproof propulsion modules that integrate power, wireless communication, and actuation. By eliminating centralized wiring and the need for repeated sealing, MarineCraft allows diverse robot geometries to be assembled and tested in minutes rather than days. Experimental results demonstrate that this reconfigurable architecture enables fast, iterative design cycles while maintaining reliable operation and leak-free performance at depths of up to 2.5 meters. Our toolkit effectively lowers the barrier to underwater robotics by transforming modularity into a vehicle for rapid physical prototyping.

We establish improved asymptotic rate bounds for adversarial insertion–deletion codes. For binary unique-deletion correction in the linear-deletion regime, we strictly improve Levenshtein's classical lower-bound expression. At deletion fraction $\delta=0.1$, our achievable rate increases from $0.162009$ to $0.194890$, a relative gain of approximately $20.30\%$. For list decoding with any prescribed finite list size $L$, we obtain new achievable-rate bounds for binary deletion and insertion codes. Our approach combines the Lovász local lemma with refined counting of common subsequences and supersequences, using generating functions and run-restricted sampling. We also establish a combinatorial reduction that trades $L$ units of insertion budget for one unit of deletion budget without changing the code or increasing the list size. It transfers insertion lower bounds at $\gamma+L\delta$ and deletion upper bounds at $\delta+\gamma/L$ to mixed errors. We also derive a run-counting Levenshtein-type list-decoding upper bound and a higher-order Elias-type upper bound. The latter recovers Yasunaga's asymptotic unique-decoding bound at $L=1$ and strictly improves the Haeupler–Shahrasbi–Sudan insertion upper bound for every fixed $L$ and $0<\gamma<q-1$. Numerical comparisons quantify the gains and the remaining gaps.

Muscle-synergy studies use factorisation methods such as nonnegative matrix factorisation (NMF) to extract muscle synergies from surface electromyography (sEMG). Reconstruction quality and restart repeatability are commonly used to evaluate these estimates, but whether they indicate recovery of underlying synergies remains unclear. Here, we defined synergy recovery as selection of the correct ground-truth rank and a minimum matched spatial cosine similarity of at least 0.80. We generated 4320 independent 16-muscle sEMG sequences across five simulation factors: ground-truth rank, signal-to-noise ratio, sequence length, spatial mixing and activation mixing. Rank selection used a variance accounted for (VAF) threshold of 0.90. At the selected ranks, NMF achieved VAF of at least 0.90 for every sequence, with median restart repeatability of 0.985. However, synergy recovery was achieved in only 7.92% of sequences. Supplying the ground-truth rank increased recovery to 60.79%. Reconstruction quality and restart repeatability had modest associations with spatial cosine similarity (Spearman's rho = 0.253 and 0.281, respectively). Recovery varied with all five factors, with nine significant two-factor interactions in exploratory analyses. In a supporting analysis, we assessed recovery using reference signals constructed from spatial weights and activation coefficients estimated from measured sEMG during five movement conditions. Selected ranks were lower than reference ranks in all five conditions despite high reconstruction quality. Accurate reconstruction and repeatable estimates therefore did not establish recovery of the underlying synergy structure. Physiological interpretations require additional evidence of functional relevance and assessment of sensitivity to the selected rank.

This paper studies the practice of combining multiple outcomes into a summary index to estimate a causal effect. For common estimators and index constructions, the estimate equals a weighted sum of the estimated effects on the components, with weights that are implicit and rarely reported. The paper derives the weights and shows that, for inverse-covariance-weighted indices, they can be negative and unrestricted in magnitude, so the index effect can have the opposite sign to every component effect. The paper proposes two procedures for valid inference on the index effect: a variance estimator that accounts for the data-dependent weights, and a shifted t-test that requires no such correction. Conventional t-tests of the null of no effect remain valid. Contrary to common claims, summary indices do not generally improve power. Three published studies illustrate the results.

Natural audio-visual interaction is emerging as an important interface for AI assistants, allowing users to communicate through speech and vision rather than carefully composed text prompts. However, existing benchmarks of interactive capabilities still focus primarily on response quality, leaving a more fundamental question underexplored: can a model correctly infer the user's underlying demand from complex multimodal interaction? Real-world user demands are often underspecified in speech and must be inferred from multimodal cues and dialogue history. This inference is further complicated by ambiguous or disfluent expression and noisy acoustic environments. Conversely, request-like speech may not constitute a demand to the assistant, leading to false triggers. We establish Omni Demand Understanding (ODU) as a distinct multimodal contextual inference problem: given an interaction stream, a model must detect whether a user demand is present and infer intent from multimodal and conversational context. ODU evaluates this capability along five dimensions, covering both single-turn and multi-turn interactions. We construct ODU-Bench using a challenge-driven taxonomy, taxonomy-guided agentic video generation, and human-recorded interactions, followed by media-grounded annotation and human verification. We evaluate 14 native MLLMs. Even the strongest, Gemini 3.1 Pro, recovers only 44.7% of key information that must be inferred from visual, acoustic, or conversational context. Moreover, 11 of the 14 models exhibit false-trigger rates above 50% on non-demand scenarios. These results reveal a systematic capability gap in current MLLMs' ability to infer contextual user demands. We hope ODU can establish the evaluation of a previously underexplored yet essential capability in multimodal interaction: correctly understanding user demands before generating an appropriate response.

Neural Radiance Fields (NeRF) have attracted extensive attention in recent years due to their strong capability for high-quality 3D reconstruction and novel view synthesis from multi-view images. Existing methods usually rely on high-quality sharp inputs, while real-world image acquisition is highly susceptible to blur degradation, which severely affects the reconstruction quality of NeRF. In this paper, we propose a novel Mamba-driven world-state-aware adaptive deblurring neural radiance field, termed WS-NeRF, to address image degradation and 3D inconsistency. We formulate the alternating optimization of radiance fields as a dynamic evolution process with temporal memory, and jointly exploit comprehensive multi-dimensional world states and a mixture-of-experts mechanism to dynamically adjust the confidence of deblurring priors. Experimental results show that WS-NeRF significantly improves blurry radiance field reconstruction quality, achieving better performance on PSNR, SSIM, and LPIPS, while exhibiting more stable iterative recovery behavior.

Air operations rely on complex rules, established procedures, and time-critical analysis under limited connectivity and strict security constraints. In such environments, analysts must combine written doctrine with images, often without access to external computing resources. This paper studies offline large language models as decision support tools, deployed in isolated and restricted environments to give analysts access to doctrinal knowledge that remains traceable to its original sources through natural language interaction. We describe a modular retrieval-augmented architecture suitable for operation without Internet connectivity, supporting both text and image input from technical manuals. As a first step toward evaluating this architecture, we report a pilot study with four image analysts of the Brazilian Air Force, combining (i) a doctrinal knowledge assessment based on their electronic-target identification doctrine, comparing human and proposed system performance on the same test, and (ii) a measurement of the cognitive workload involved in manually producing a reconnaissance target report (Relatório de Missão de Reconhecimento - REMIR) without AI assistance. The results show a demanding manual task, especially in terms of mental demand (6.0/7) and effort (5.0/7), while the proposed system matches the human score (8/10) and completes the assessment in 7.1 minutes (compared to a human average of 26.5 minutes), establishing a baseline for future AI-assisted evaluation. Finally, we describe a future evaluation protocol to systematically compare manual and AI-assisted workflows.

Families of plateaued Boolean functions with pairwise disjoint Walsh supports are useful in secondary constructions of cryptographic Boolean functions. Of particular interest are maximum-cardinality families whose members admit no nonzero linear structures. To the best of our knowledge, the previously known general construction attaining both properties is spectral (Hodžić et al., IEEE Trans. Inf. Theory 65(9): 5865–5879, 2019). In that work, explicit algebraic normal forms are not generally provided, and no general method is established for prescribing a common algebraic degree for all family members. In this paper, we present two new explicit algebraic constructions within a unified framework, one based on linear functions and the other on partially linear functions with bent components. Let $p\geq 2$ and $q\geq 0$ satisfy $q<2^p-p-1$, and set $m=p+q$. Both constructions yield maximum-cardinality families of $2^{q+1}$ $(q+1)$-plateaued Boolean functions with pairwise disjoint Walsh supports. No member admits a nonzero linear structure, and every member has an explicit generalized Maiorana–McFarland representation. The first construction produces functions in $m+p+1$ variables and realizes any prescribed common algebraic degree $3\leq d\leq p+1$, provided that $q<\sum_{i=2}^{d-1}\binom{p}{i}$; its maximum attainable degree $p+1$ is optimal. The second construction produces functions in $n+p+1$ variables, where $n>m$ and $n-m$ is even, and realizes any prescribed common algebraic degree $3\leq d\leq p+(n-m)/2$, provided that $q<\sum_{i=2}^{\min\{d-1,p\}}\binom{p}{i}$; its maximum attainable degree $p+(n-m)/2$ is next-to-optimal.

A graph $G$ is deletion-induced-saturated for $H$ if $G$ has an edge, contains no induced copy of $H$, and deleting any edge of $G$ creates an induced copy of $H$. We prove, with finite certificate verification, that a finite graph $H$ admits such a finite graph $G$ if and only if $H$ is not complete. This resolves the deletion conjecture of Fan, Hajebi, Hajebi and Spirkl. The main step transfers suitable free amalgamations to finite extensions using a local lifting theorem of Auinger, Bitterlich and Otto. A second criterion treats edge addition by protecting specified nonedges and then taking a maximal induced-$H$-free completion. Structural results of Bonamy, Groenland, Johnston, Morrison and Scott reduce the remaining targets to dense templates and a finite hereditary class. Two uniform constructions in halved cubes handle the dense templates. The finite part is supported by exhaustive coverage certificates, structural certificates and explicit hosts, including a circulant graph on $30$ vertices.

Relexicalization is a pivotal technique in clinical NLP, as it facilitates robust masking of sensitive information while synthesizing datasets that retain high-fidelity, real-world characteristics. However, preserving structural integrity, relational coherence, and temporal consistency during transformation remains a significant challenge. Existing approaches frequently rely on independent entity replacement, which results in clinical inconsistencies across longitudinal records. This reduces the value of such relexicalized datasets for downstream scientific analysis. To address these limitations, we introduce G-RELIC (Graph Based Contextual Relexicalization with Improved Consistency) which combines the power of LLMs with graphs. G-RELIC implements a graph-based mapping mechanism which optimizes for one-to-one correspondence between original and surrogate entities. It also introduces a deterministic temporal repositioning algorithm to preserve temporal consistency. Empirical evaluations on diverse, real-world clinical datasets validate that G-RELIC significantly outperforms state-of-the-art baselines. G-RELIC yields a 30.4 percentage point improvement in relational integrity (62.1% to 92.5%) and 45.9 percentage point improvement in temporal coherence (46% to 91.9%) without compromising on the recognized privacy benchmarks for clinical datasets. This maximizes the analytical utility of relexicalized datasets while minimizing re-identification risk.

Comprehensive video understanding is crucial for advancing artificial intelligence toward the intricate dynamics of the physical world. While recent advances in Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in video understanding, existing benchmarks remain confined to simple scene-level queries or global summaries that require only single-step inference. Real-world video understanding involves more challenging tasks that require multi-hop multimodal reasoning, and there is a critical absence of video benchmarks equipped to rigorously evaluate these agentic capabilities. To bridge this gap, we introduce AgentVidBench, a multi-hop video question answering benchmark focused on evaluating the spatial, temporal, and causal reasoning capabilities of MLLM agents. Beyond standard question-answer pairs, AgentVidBench provides step-by-step solution traces to support trajectory evaluation that assesses whether agents explicitly acquire the evidence needed to justify their answers. Experiments with 12 proprietary and open-source MLLMs show that single-turn performance remains limited on AgentVidBench, while integrating these models into state-of-the-art agentic workflows generally improves performance with respect to both accuracy and trajectory scores. We further present a simple yet effective agentic strategy that serves as a competitive baseline on AgentVidBench, establishing our benchmark as a holistic testbed for future research on agentic video understanding. Code and datasets are available at https://github.com/krafton-ai/agentvidbench and https://huggingface.co/datasets/agentvidbench/agentvidbench.

In the first part of this work we showed that, within a Hilbert plane deprived of the Side-Angle-Side axiom, the Side-Angle-Angle criterion, together with a ray correspondence principle [RCT], the existence of angle bisectors [AB], the congruence of supplements of congruent angles [SA], and the Pons Asinorum [PA], suffices to reconstruct SAS. We left open the question of whether [PA] is genuinely required alongside the other three principles, noting only a qualitative asymmetry in the nature of the principles involved. In this second part we answer this question: we show that \begin{equation*} SAA,\ [RCT],\ [AB] \;\vdash\; [PA], \end{equation*} so that [PA] is redundant among the hypotheses of our main theorem, which improves to \begin{equation*} SAA,\ [RCT],\ [AB],\ [SA] \;\vdash\; SAS. \end{equation*} The proof adapts an argument recently given by Donnelly, who reconstructs SAS from SAA together with an angle addition axiom and the existence of angle bisectors.

Mitra, Offen, and Sayag introduced distinguished multisegments and multisegments of Speh type, and proposed that every distinguished multisegment is of Speh type, together with a related duality hypothesis (see A. Mitra, O. Offen, and E. Sayag, Klyachko Models for Ladder Representations, Documenta Math. 22 (2017), 611-657). They proved these statements for sets of segments and when at most two segments share an endpoint. We develop a combinatorial theory of relevant decompositions and prove that relevance is preserved under the involution $\Delta\mapsto\Delta^\vee$ together with reversal of standard order. Hence, $\mathfrak m$ is distinguished if and only if $\mathfrak m^\vee$ is distinguished, so the two hypotheses are equivalent. We define $S_{\mathfrak m}(\Delta)=\sum_{t\ge0}(-1)^t\mathfrak m(\nu^t\Delta)$ and show that $\mathfrak m$ is of Speh type if and only if $S_{\mathfrak m}(\Delta)\ge0$ for every segment $\Delta$. Using rigidity properties, we obtain a peeling theorem and a numerical tameness condition, and prove both hypotheses for every tame multisegment, extending the previously known classes. Finally, we construct a five-segment multisegment that is not distinguished, although every standard order with non-increasing endpoints admits a non-trivial relevant decomposition. Thus, endpoint-ordered witnesses alone cannot prove the hypothesis in general. The remaining case reduces to multisegments for which both $\mathfrak m$ and $\mathfrak m^\vee$ are non-tame.

Depth-recurrent language models refine predictions through repeated latent updates. Why can intermediate answers agree with the endpoint while their scores continue to change? We derive a sharp margin characterization that decomposes the conservatism of a magnitude bound into common translation, direction relative to the winner, and the pairing of each competitor's update with its score gap. Across Huginn-3.5B and Ouro-1.4B, accounting for update direction and competitor pairing reduces the mean earliest qualifying depth by a further 22.5-34.4% of the total depth beyond translation removal under full answer-text scoring. This retrospective comparison uses completed trajectories. Substantial contributions also occur under label scoring. For shared predictive distributions, we separate common and contrast motion orthogonally and express the common component through candidate-set mass and within-set concentration. Common and contrast energies can attenuate at different rates, allowing a growing preference-change share to coexist with shrinking absolute updates. These findings explain finite-depth answer preservation through the geometry and composition of observed score changes.

Operators of service-based systems act on forecasts of how a running execution will continue, and such a forecast is actionable only if its reliability is known. Mainstream deep-learning models for this task are discriminative and deterministic: they emit a single next activity and a single remaining-time estimate, without a distribution to reason over. We instead cast the problem as generative sequence modelling with marked temporal point processes, which define a joint density over the next mark and its inter-event time and therefore deliver predictive distributions by construction. Real event logs violate the simple-point-process assumption these models rest on, since consecutive events frequently carry identical timestamps; we handle such ties explicitly and combine a transformer encoder with a mixture decoder over inter-event times, trained by exact log-likelihood. On ten public logs, the resulting model matches discriminative baselines on point accuracy, dominates them on the calibration and sharpness of remaining-time distributions, and is the cheapest at inference, since a full predictive distribution is obtained in a single forward pass without sampling.

We investigate whether coupling a time-series foundation model to hydraulic project knowledge improves surrogate forecasting of HEC-RAS water-surface elevation (WSE). We present KG-Chronos-2, which combines a frozen Chronos-2 predictor with exact-state residual decoding, graph-conditioned historical retrieval, and input-aligned correction. We compare the method with persistence, a residual LSTM, project-conditioned recurrent GeoFNO, a hydraulic DCRNN-style model, and frozen Chronos-2. Task-specific fitting uses the 2008 simulation. Evaluation covers 64 fixed 24-hour windows from the 2011 and 2002 simulations at 4,675 cross sections in 71 reaches on a shared geometry. KG-Chronos-2 achieves event-balanced root-mean-square error 0.246970 in native WSE units. It reduces RMSE by 14.13% relative to frozen Chronos-2, 29.38% relative to the hydraulic DCRNN-style model, and 39.54% relative to recurrent GeoFNO. The 95% hierarchical-bootstrap interval for its event-balanced RMSE difference from frozen Chronos-2 is [-0.075177, -0.016317]. KG-Chronos-2 also achieves the lowest active-window and final-lead RMSE among the six completed systems. These results support coupling a frozen temporal predictor to project knowledge for warm-start HEC-RAS forecasting on the fixed benchmark.

Environmental noise is a major source of decoherence and errors in quantum systems. In a non-Markovian environment, the noise has a finite correlation time, and the system can retain information about its previous interactions with the environment. Here, we investigate how such environmental memory can be observed through a finite-duration coherent quantum operation. Coherent control dynamically reshapes the stochastic coupling operator and thereby changes how two-time noise correlations affect the quantum process. For Ornstein-Uhlenbeck noise, we show analytically that two different limits reduce the information on the correlation time. In the short-memory limit, the leading dynamics depend on the integrated-noise combination {\sigma}/{\tau}c, where {\sigma} is the noise variance and {\tau}c is the correlation time, making these parameters locally degenerate. In the long-memory limit, the stochastic field becomes quasi-static during the operation, and sensitivity to the correlation time is suppressed. Exact stochastic simulations for a driven single-qubit X{\pi} rotation and a two-qubit exchange gate show that independent correlation-time information becomes largest between these limits, when the environmental and control timescales are comparable. Noncommuting controls enhance this information, whereas commuting controls do not. Simulations at different gate durations tg further show that the information window approximately follows the dimensionless correlation ratio {\tau}c/tg. These results provide a timescale-matching principle for probing finite-correlation non-Markovian noise with coherent quantum control.

Accurate traffic forecasting requires both understanding scene dynamics and synthesizing realistic future observations. Recent diffusion-based video generation models produce visually plausible predictions but require expensive end-to-end training and often entangle scene understanding with image synthesis. In this work, we propose a decoupled forecasting framework that separates future representation learning from video generation. A frozen V-JEPA encoder first extracts predictive latent representations from the observed traffic videos, capturing the underlying scene dynamics in a semantic latent space. A lightweight latent alignment module then projects these representations into the conditioning space of a frozen Cosmos diffusion module, enabling future video synthesis without retraining the large generative model. By freezing all foundation models and training only the lightweight alignment module, the proposed framework substantially reduces optimization complexity while preserving forecasting capability. Experimental results on the AI City Challenge 2026 Track 5 benchmark demonstrate that the proposed method achieved a score of 75.1297, ranking third in the competition. These results suggest that predictive world representations learned by V-JEPA can effectively guide downstream video generation, providing a practical and efficient alternative to end-to-end diffusion-based forecasting.

Reinforcement learning has substantially improved large language model (LLM) agents in verifiable domains, but remains difficult to apply to open-ended agent tasks, where solutions are diverse and reliable scalar rewards are hard to obtain. Recent pairwise evaluation methods alleviate reward discrimination collapse by replacing pointwise scoring with relative preferences. However, they still compress rich comparative feedback into a single trajectory-level reward, obscuring decisive intermediate steps and preventing successful behaviors from being consolidated into reusable skills. We propose ArenaFlow, a hierarchical credit propagation framework for open-ended agent reinforcement learning. ArenaFlow leverages tournament-based relative ranking to derive trajectory-level reward signals. Each comparison is further equipped with structured reflective evaluation, which reveals three types of supervision: pivotal success steps, reusable strategy skills, and usage attribution of retrieved skills. At the step level, ArenaFlow propagates trajectory-level advantages to high-confidence pivotal steps according to tournament survival depth, enabling more targeted optimization of local reasoning behaviors. At the skill level, ArenaFlow estimates skill utility from group-level usage attribution and maintains a global skill memory through utility-aware updating, pruning, and retrieval. The resulting high-utility skills further serve as policy priors for future exploration. Extensive experiments validate ArenaFlow's effectiveness on open-ended agent tasks.

Robotic fabric comparison needs to actively combine visual appearance and tactile cues. Here, we present AVT-Fabric, an RGB-first framework that allocates tactile evidence according to the difficulty of each comparison. A dual-scale gate evaluates answer-token confidence and raw logit separation to determine whether another force-tagged GelSight observation is needed. Compact textual memory preserves the executed history, and majority voting consolidates the selected predictions. On 400 held-out comparisons, AVT-Fabric achieves 98.0% accuracy with a compact 7B Multimodal Large Language Model (MLLM), surpassing the 94.0% reported by the 90B MLLM-Fabric baseline by 4.0 percentage points while processing only 1.60 of five available stages on average. It improves on matched passive inference by 9.25 percentage points and reduces model-side latency by 61.8%, while also improving on RGB-only accuracy. Four additional MLLM backbones support the generalizability, accuracy, and efficiency of the framework. This framework is also deployed on a real robotic system, achieving 78.1% pairwise ranking accuracy and correct fabric selection in seven of eight application scenarios. AVT-Fabric demonstrates that adaptive evidence selection can improve both the accuracy and efficiency of robotic visuo-tactile reasoning.

We consider the 4.8.8 color code in a brickwork layout on the square lattice with nearest-neighbor gates. We identify periodic superdense circuits with twelve CNOT layers that appear to preserve the full distance of these codes. Including ancillas, the family uses $q=(2d^2+5d-5)/2$ physical qubits for distance $d$: asymptotically $2/3$ as many as the triangular 6.6.6 color code and $1/2$ as many as the rotated surface code at the same code distance. Circuit-level simulations targeting superconducting architectures demonstrate favorable scaling of 4.8.8 color codes at plausible physical noise rates, including both for planar codes and their toroidal cousins under periodic boundary conditions.

Optical frequency combs (OFCs) are revolutionary light sources characterized by discrete and equally spaced spectral lines, and they have found widespread applications in metrology, spectroscopy, and communications. In the early stages, OFCs were realized using mode-locked lasers. With advancements in the fabrication of high-quality factor ($Q$) microresonators and the increasing demand for miniaturized and integrable photonic chips, microresonator-based OFCs, commonly referred to as microcombs, have been developed. Although early studies of microcombs primarily focused on single microresonators or a few resonators, a significant breakthrough occurred in 2021 when it was theoretically predicted that light propagating in the topological edge channel of an array of ring resonators could generate nested frequency combs known as topological OFCs. Since then, the field of topological OFCs has progressed rapidly, with experimental observations made in 2024. This Perspective will introduce the history of OFCs, placing particular emphasis on the emergence and development of topological OFCs, as well as exploring the research challenges and opportunities associated with them.

Baryons in galaxy ecosystems are believed to undergo continuous cycles of inflow and outflow, forming fountain-like patterns that encode key information about how galaxies acquire matter from their environments and respond through feedback. The presence of such baryon cycles has been inferred from pieces of observational evidence, but a concrete understanding remains elusive because individual galaxy ecosystems are diverse and dynamic. Here we introduce a stacking method that combines baryonic fields across ensembles of individual galaxy ecosystems to suppress irregularities and reveal the underlying pattern. Applied to a cosmological hydrodynamic simulation, this approach unveils strikingly regular patterns in gas properties across the full spatial extent of galaxy ecosystems, in close agreement with those inferred from observations. This method is straightforward to implement, allowing the processes shaping the gas-cycling pattern to be fully understood within the structure-formation paradigm, and a solid framework to be constructed for linking simulated galaxy ecosystems with observations.