Papers

Let $F$ be a $p$-adic field, let $k,c\ge1$, and let $\rho$ be a $(k,c)$ representation of $\mathrm{GL}_{kc}(F)$: a complex representation of finite length whose wave front set has $(k^c)$ as its unique maximal orbit, and which admits a $(k,c)$ functional that is unique up to scaling. We prove that the functions in the $(k,c)$ model of $\rho$ are of moderate growth, by relating this model to the Zelevinsky model of $\rho$, and obtain as a consequence a bound for the restriction of these functions to the standard Levi subgroup associated with the composition $(c^k)$ of $kc$. As applications, we study convergence problems of several integrals involving $(k,c)$ models: integrals from the generalized doubling method, and a new family of integrals which extends the Rankin–Selberg integrals for Speh representations of Lapid and Mao.

We propose a Japanese handwriting generation framework that integrates human preference feedback using Direct Preference Optimization (DPO) to align generative models with human cognitive and stylistic preferences. While font generation technologies leveraging Generative Adversarial Networks and diffusion models have advanced significantly, applying these approaches to handwriting generation presents unique challenges. Font generation is a deterministic process yielding consistent styles. Handwriting generation, on the other hand, is inherently non-deterministic. In handwriting, glyph shapes never match exactly even when the same writer writes the same character. We define this shape difference as handwriting variability. Although this variability is essential for conveying the natural quality of human writing, it complicates model training and substantially increases data collection costs to capture diverse distribution patterns. To address these limitations, we incorporate human perceptual feedback into the training loop of handwriting generation. While collecting extensive handwritten data is costly, evaluating pairwise preferences between two generated samples remains very easy for humans. Applying DPO, an efficient RL-free preference optimization method, to our generation model enables high-quality handwriting generation without relying on large-scale dataset collection. Experimental results across four distinct Japanese handwriting styles demonstrate that DPO-based training enhances overall structural accuracy. However, in highly complex styles with significant variability, such as brush pen writing, the model fails to capture style-specific features. These findings highlight a critical challenge in extending DPO to complex handwriting styles.

In this paper, we first apply the implicit function theorem to find nontrivial solutions to the following Bernoulli's overdetermined problem \begin{equation} -\Delta u=\gamma\,\, \text{in}\,\,\Omega \setminus\overline{D}, \,\, u=1,\,\,\partial_\mathbf{n} u=Q\,\,\text{on}\,\,\partial D, \,\,u=0\,\,\text{on}\,\,\partial\Omega \nonumber \end{equation} for a fixed boundary $\partial\Omega$ and a free boundary $\partial D$ in $\mathbb{R}^{N}$ with $N\geq2$, where $\gamma$ is a sufficiently small constant, the Bernoulli value $Q$ is determined by the construction, and $\Omega \setminus\overline{D}$ is a bounded ring-shaped domain. In dimension $N=n+1$ with $n\geq2$, we further study the existence of a nontrivial unbounded domain $\Omega \setminus\overline{D}$ in $\mathbb{R}^N$ such that the Bernoulli's overdetermined problem admits a solution. This unbounded domain we construct is periodic in one variable and radial in the other variables by bifurcation analysis.

We prove that the full Dirichlet-to-Neumann map for static isotropic elasticity uniquely determines both smooth Lamé parameters in a bounded smooth domain of \(\mathbb R^3\) under strong convexity. The proof uses an exact four-component Laplace-type reduction, but its decisive identities are obtained from physical elastic solutions. The characteristic quotient of the reduced system has rank three, whereas the leading physical adjoint lift supplies only two directions. We construct the missing polarization from the first nonvanishing correction of a reduced amplitude in the kernel of that lift. Paired with fixed primal and adjoint quotient frames, the resulting complex geometrical optics solutions yield a complete family of holomorphically weighted \(3\times3\) matrix moments through the physical Betti identity. Cauchy–Plemelj factorization converts these moments into an invertible relative transport normalized in the common exterior. Holomorphic dependence on the projective null direction and transport rigidity then determine a material ratio and the shear modulus, hence both Lamé parameters. No smallness, symmetry, separability, or analyticity hypothesis is imposed.

Programming abstractions have traditionally been designed for humans, but code generation is increasingly performed by models and autonomous agents. We investigate abstractions for machine-driven programming through GPU and AI-accelerator kernel generation. High-level domain-specific languages (DSLs) simplify search by delegating implementation details to compilers, while lower-level representations provide finer-grained control at the cost of a larger and more complex search space. In this paper, we present TileCas, a unified, pluggable interface that enables coding agents to develop kernels across multiple editable abstraction levels. Rather than committing to a single representation, TileCas allows an agent to transfer and refine candidates across levels through Cross-Level Diagnosis, which maps correctness and performance feedback to actionable controls at different abstraction levels; Abstraction Scheduling, which selects both the candidate and representation from which subsequent refinement should proceed; and a Heterogeneous Search Tree, which preserves and revisits alternative search trajectories spanning multiple representations. We integrate TileCas with two code agents AKO and KDA, and evaluate both integrations on 50 end-to-end KernelBench Level 3 workloads across two accelerator platforms and two language models. Giving equal weight to the four accelerator-model configurations, AKO with TileCas achieves an arithmetic-mean speedup of $4.84\times$, compared with $3.75\times$ for the strongest evaluated baseline without TileCas, an improvement of 29.0%. Integrating the same interface with KDA improves its mean speedup by 69.2%, with gains in every configuration. These results show that allowing coding agents to navigate across programming abstractions can improve machine-driven software development and can be reused across different base workflows.

In this paper, we show how a recently introduced $\lambda$-calculus for randomized computation, called $\lambda\mathbf{BLL}$, can accurately capture the notions of first- and second-order polytime computations used in cryptographic definitions and proofs. These results, spelled out in the style of so-called implicit computational complexity, imply that $\lambda\mathbf{BLL}$ is sufficiently powerful to express experiments and adversaries as used in cryptography, and that reductions, when formulated within the calculus, are guaranteed to be complexity-preserving. Remarkably, while polynomial-time soundness holds for the full calculus, polynomial-time completeness already follows in presence of a minimal finite set of basic functions.

Grassroots platforms are designed to be owned, operated and governed by their participants, and thus are constituted from person-to-person smartphone apps that are run by the participants. A grassroots app is specified using a grassroots social contract, which is compiled through an abstraction cascade into a volition-guarded Grassroots Logic Program (vGLP), a multiagent concurrent logic programming language. A vGLP program in turn specifies a smartphone UI via which people express their volitions and thus control the app. Prudence dictates large-scale evaluation and testing of grassroots platforms prior to deployment, but using the many people needed to do that is a contradiction in terms. Hence another solution is needed. Here, we present stochastic GLP (sGLP), designed to allow grassroots platforms to be simulated at scale from the vGLP programs that deploy them, with stochastic agents in place of people. sGLP is GLP with a rate extension for specifying stochastic behaviour. A stochastic agent is an sGLP program replacing both the vGLP UI and the person behind it. It is typed by the dual of the interactive type of the UI and specifies a stochastic profile of a person's behaviour. We demonstrate sGLP on the grassroots social graph, with a population of men and women drawn from four stochastic profiles — homophile or indifferent, wary or sociable — and on grassroots currencies, with four profiles — spenders or savers, cautious or generous — and report what the mix of profiles does to the social graph that forms and to the currencies that circulate. We prove that every run of a vGLP program in which sGLP agents replace the UI and the people behind it is a vGLP run. We implemented sGLP in GLP by Gillespie's method, and ran reproducible simulations with it.

Training language agents on long-horizon interactive tasks is challenging because sparse terminal rewards provide weak credit signals for intermediate actions. Existing group-based methods either assign a shared trajectory-level advantage or estimate step-level advantages from state groups, which may conflate the current action with subsequent decisions. We propose Prefix Tree Group Policy Optimization (PTGPO), a critic-free method that constructs prefix trees over trajectory suffixes within anchor-state groups. PTGPO assigns each tree edge a credit given by the difference between its child and parent empirical mean returns, capturing the conditional return change revealed by each additional action. These prefix-tree credits provide fine-grained learning signals without requiring a critic or additional rollouts. Experiments on interactive benchmarks ALFWorld, WebShop and Sokoban show that PTGPO consistently improves over RL baselines, demonstrating its effectiveness.

Optimization by PROmpting (OPRO) directly leverages Large Language Models (LLMs) to generate candidate solutions through natural-language prompting. However, OPRO's optimization process is difficult to track, which prevents LLMs from capturing useful information accumulated over previous iterations, thereby limiting solution quality. To address this limitation, we propose a Tool-Augmented Strategic Agent (TASA) which summarizes and exploits trajectory to improve solution quality. Specifically, TASA plans the strategy for each step based on the optimization trajectory, which decomposes the optimization process into traceable strategy selections. Moreover, we leverage LLM to construct problem-specific tools and introduce a strategy-guided adaptive tool planner, which efficiently utilizes the information of trajectory to update solutions. We establish local convergence under persistent exploration and evaluate the complete framework under a shared candidate-evaluation budget. The results support its solution quality and the traceability of strategy-guided search.

Fractions require learners to revise properties associated with whole numbers. This paper presents a design rationale and a critical analysis of an eight-level source progression for fractions (Grades 1-8, nominal ages 6-14), from which the French-curriculum and International Baccalaureate chapters published at https://commeunjeu.org are assembled. Its central hypothesis is a bridge from sharing to numerical measure: a share of one unit is represented by a bar, read as a length relative to a specified unit, and located on the number line. The reference unit is kept stable in the initial tasks to focus attention on fraction magnitude, equivalence and operations; changing the unit is deliberately postponed, without treating this as a mathematical requirement or an established reduction in cognitive load. Three principles organise the meanings: sharing one unit, coordinating bars and number lines, and establishing fractions as numbers before explicitly treating them as scalar operators. Two supporting principles concern valid local justifications before formal generalisation and chains of equalities, introduced in Grades 2-3, with written reasons in lessons from Grade 4. The analysis relates these choices to the research literature and selected curricula, examines task strategies and feedback, and distinguishes realised features from discrepancies and proposed revisions. The source documents and their curriculum projections do not establish the order actually experienced by pupils. No empirical evaluation is reported. Primary hypotheses address transfer between representations, coordination of units and the numerical status of fractions before operators; the audit also identifies limits to the reproducibility of its findings.

Set-based variational autoencoders (VecSet VAEs) represent 3D shapes as compact sets of latent tokens. Although each token originates from a query at a sampled surface location, whether these latents truly align with geometry remains unclear. We investigate this alignment and identify four issues: (i) attention-sink-like tokens that attract queries from across the shape; (ii) spatially periodic repetitions in decoder attention responses; (iii) mismatched normal information in encoder and decoder query features; and (iv) missing supervision of the intrinsic relation between signed distance fields and surface normals. To address these issues, we introduce Align-VAE with four corresponding modifications: (i) a shared bank of learnable decoder tokens provides a workspace for global computation; (ii) distance-biased cross-attention guides retrieval toward nearby predicted spatial anchors; (iii) normal-free encoder queries align the two stages' access to normal information while retaining normals in the encoder keys and values; and (iv) a screened Poisson-inspired regularizer supervises surface values and normal-direction derivatives. These changes align latent retrieval and field prediction with geometry without enlarging the per-shape code. On the ABO dataset, Align-VAE reduces Chamfer distance by 26.6% compared to the standard VecSet VAE.

B-Human won multiple competitions in the Middle Division of the new Humanoid Soccer League in 2026. In addition to regular matches with three or five players per team, we ventured into uncharted territory by playing demo matches with eleven humanoid robots, highlighting the tactical flexibility of our behavior and the overall robustness of our system. As B-Human formerly competed in the Standard Platform League with robots half the size and less than a third the weight, we begin by outlining the adjustments needed for this transition. The central section of this paper details our behavior architecture, drawing on concepts from RoboCup leagues that have used 11 players for much longer.

Chronic wounds affect 1-2% of the global population and represent an unsustainable healthcare, financial and economic burden. Delayed healing is largely associated with bacterial infection and inadequate moisture balance, both of which remain difficult to control due to increasing antimicrobial resistance and the limited exudate management capacity and mechanical performance of current dressings. To address these challenges, we introduce a staining-free photodynamic material with work-light-triggered antibacterial activity, significant water-induced swelling (up to more than nine-fold increase in dry weight) and tunable wet state mean compression modulus (Ec: 76 - 289 kPa). The material was prepared by UV co-curing of an aqueous solution containing a poly(vinyl alcohol) derivative functionalised with both phenothiazine and methacrylate residues, 4-vinylbenzyl-functionalised type 1 rat tail collagen, and a water-soluble photoinitiator. Fourier Transform Infrared spectroscopy and Differential Scanning Calorimetry of the co-networks indicated the presence of characteristic chemical and thermal features of both components. 96-hour co-network incubation in vitro (pH 7.4, 37 degC) confirmed the significant retention of toluidine blue (> 93 wt.%), suppressed medium colouration, and an average mass retention higher than 98 wt.%. Exposure of the UV-cured samples to 10-min work light resulted in at least 1 log10 reduction in Staphylococcus aureus and Pseudomonas aeruginosa viabilities, in contrast to dark-control conditions, while the metabolic activity of L929 fibroblasts remained above 90%. The antibacterial properties, staining-free photodynamic behaviour, water-induced swelling, and wet-state mechanical properties achieved via network integration with functionalised collagen, make this material appealing for chronic wound healing applications and for on-demand delivery of localised therapies.

Long-term memory benchmarks for language agents primarily evaluate whether information from earlier interactions can be retrieved and used when relevant. Persistent memory systems face an additional problem before future use is known: deciding whether a memory should remain exact, be compressed, consolidated, or deleted. Existing benchmarks capture long-range retrieval, temporal reasoning, evidence reuse, and incremental interaction, but do not directly evaluate these earlier representation decisions. We formalize memory lifecycle evaluation and introduce VitaBench, a benchmark that makes lifecycle decisions measurable through repeated memory reviews and full-horizon counterfactual evaluation. VitaBench contains a controlled Diagnostic tier that isolates future-use patterns and a Naturalistic tier that composes them into generated customer-service-style interaction streams. We characterize lifecycle decision structure with three diagnostics: lifecycle headroom, measuring the advantage of adaptive choices over the best fixed strategy; horizon reversal, measuring when short-horizon decisions change under delayed consequences; and optimal-action diversity, measuring whether different memory states favor different lifecycle operations. Across both tiers, VitaBench exhibits substantial lifecycle structure: lifecycle headroom is 1.364 in Diagnostic and 1.405 in Naturalistic, and 60.2% and 56.5% of states, respectively, change their preferred action between short and full horizons. A learned delayed lifecycle policy further improves over the best fixed strategy on VitaBench-Naturalistic, reaching 101.5 versus 83.4 joint utility. These results show that retrieving past information and deciding how it should persist are distinct evaluation problems, and that lifecycle management requires evaluation of the delayed consequences of representation decisions.

AI agents require specialised retrieval across domains and tasks, but adapting embedding models is constrained by costly manual training-data design. Here we introduce SYNTRA, a self-improving agentic framework that turns a few retrieval examples into reusable synthesis instructions. Given an unlabelled corpus and a few relevant query-document pairs, SYNTRA infers implicit information needs and relevance criteria, then uses feedback on trial outputs to refine instructions for query generation and relevance assessment. Experiments across 24 datasets spanning 15 domains demonstrate that the same synthesis workflow produces effective training data, benefiting even strong embedding models with and without existing labelled training sets. Self-improvement strengthens this supervision on all six evaluated tasks, yielding better retrieval than human calibration on four. The training value persists at larger scales: on MS MARCO, retrieval gains continue across a 32-fold range of synthetic training-data sizes. SYNTRA thus enables scalable embedding-model adaptation with minimal human intervention.

Modern software systems are large and complex, demanding scalable and autonomous techniques for vulnerability discovery. Existing approaches still depend on human expertise for domain knowledge, and on human intervention throughout the process of vulnerability discovery. AI agents offer us an opportunity to make vulnerability discovery more adaptive, scalable, and autonomous. Agents bring the ability to make continuous decisions and adapt, and general knowledge that can be augmented. In this paper, we present our experience using AI agents to find vulnerabilities across diverse programs such as Windows drivers, the Linux kernel, XNU, firmware, and web applications. We use an iterative agentic workflow and find that it generalizes across these targets. We discuss tasks that agents help with, which were previously challenging, and some future research directions we consider important for advancing autonomous vulnerability discovery.

Nonreciprocal charge transport, characterized by direction-dependent electrical response, provides a powerful probe of electronic structure and a platform for rectification. However, its realization is typically governed by intrinsic crystal symmetry, restricting the range of suitable materials. Here, we demonstrate nonreciprocal charge transport in an otherwise centrosymmetric superconductor, FeTe0.7Se0.3 (FTS), through kink boundary-induced strain engineering. The stress relaxation during growth generates a cusp-like kink-band boundary with asymmetric curvature, producing a spatially nonuniform strain field and an effective in-plane polarity that breaks inversion symmetry. Pronounced nonreciprocal transport is observed under out-of-plane magnetic fields, with strong dependence on current and temperature across the superconducting transition, in contrast to the negligible response in flat nanosheets. Our results identify kink boundary-induced symmetry breaking as a general route to engineer nonreciprocal charge transport in centrosymmetric systems.

Autoregressive world models have emerged as a powerful paradigm for sample efficient decision making, yet their open loop rollouts often exhibit increasing drift that leads to error accumulation even within a relatively small planning horizon. Hierarchical temporal abstraction is a natural candidate to mitigate such drift by reducing the effective prediction horizon; however, existing hierarchical designs can introduce added structural and conceptual complexity, sometimes making them difficult to control and limiting gains over flat baselines. To this end, we introduce MHWA (Multi-timescale Hierarchical World-Action model), which employs a strided high level context channel to reduce global context update frequency, effectively curbing the propagation of compounding drift. Crucially, via context aware routing, the gating network dynamically reconfigures a low level mixture of experts by using both global context and current observations, thereby enabling flexible high-low coordination, helping alleviate interference commonly observed in static hierarchies. Empirically, using RLiable's seed-by-game aggregation, MHWA (77M parameters) achieves an IQM human normalized score (HNS) of 0.884 across three training seeds on a 14-game Atari subset using 10% subsampled offline data, outperforming a matched JOWA-70M baseline (0.529) and achieving numerically comparable performance to JOWA-150M while using nearly half the parameters and 70.88% fewer training FLOPs per sequence.

Latent diffusion models enable high-quality video generation, but video VAE encoding and decoding remain computationally expensive. Existing methods use stage-specific lightweight operators, yet either constrain spatial and temporal computation jointly or remove temporal computation altogether, without distinguishing their compression sensitivities in pretrained Conv3D. To address these limitations, we present LightVAE, a compression framework that exploits redundancy in temporal modeling and network depth. LightVAE exactly reparameterizes eligible causal Conv3D layers into an aggregate spatial response and a difference-driven temporal correction. Based on their distinct compression sensitivities, temporal low-rank compression (TLR) keeps the aggregate spatial kernel dense and applies layer-specific low-rank approximation only to the temporal correction. An algebraically equivalent folded implementation converts the arithmetic savings into practical acceleration. LightVAE further uses width-preserving block pruning to reduce depth and probe-guided feature alignment, guided by feedback from a frozen teacher suffix, to recover quality after compression. The framework also extends to video encoders. Across three video VAE backbones and two GPU platforms, LightVAE consistently accelerates decoding. For the Wan VAE family, it achieves up to $2.24\times$ and $1.11\times$ the decoding throughput of state-of-the-art efficient VAE decoders on NVIDIA H100 and RTX~5090, respectively, with higher DAVIS PSNR and lower LPIPS. On MiniMax-H3, it achieves up to $2.97\times$ the original decoding throughput while retaining 99.51% of the original model's overall VBench-2.0 score. Our code is available at https://github.com/ModelTC/LightVAE.

In the present work, we investigate the production of $\chi_{c0}$ in the $\bar{\Lambda} n \rightarrow \chi_{c0} K^{0}_{S}$ process by utilizing an effective Lagrangian approach. Based on the $\chi_{c0} \rightarrow n K^{0}_{S} \bar{\Lambda}$ decay observed by the BESIII Collaboration, the coupling constants are determined by fitting the $M_{K^{0}_{S} \bar{\Lambda}}$ and $M_{n K^{0}_{S}}$ invariant mass spectra, where eight $N$ and $\Lambda$ state are considered. Moreover, the parameters $\beta$ is taken to be 1, 2, 3 as special cases to check the parameter dependence. Our results show that two clear structures can be observed in both $M_{K^{0}_{S} \bar{\Lambda}}$ and $M_{n K^{0}_{S}}$ invariant mass spectra, which should correspond to $N(1720)/N(1895)$ and $\Lambda(1600)/\Lambda(1890)$, respectively. In addition, the contributions from $N(938)$ and $\Lambda(1116)$ intermediate states are much smaller than those of the above-threshold resonances. With the fitted parameters for $\beta=1$, 2, 3, the cross sections for $\bar{\Lambda} n \rightarrow \chi_{c0} K^{0}_{S}$ are estimated to be 1.9 nb, 4.2 nb, and 8.0 nb at $\sqrt{s}=6$ GeV, respectively. Furthermore, our estimations indicate that the differential cross sections for $\bar{\Lambda} n \rightarrow \chi_{c0} K^{0}_{S}$ reach a maximum in the forward angle region.

In the classical occupancy problem, $B$ balls are placed independently and uniformly into $d$ boxes. We study a graph analogue of this model. Let $G$ be a directed $r$-regular graph on $d$ vertices and fix a vertex $i$. Consider all paths of length $L$ starting from $i$, and let $N_L(i,j)$ denote the number of such paths ending at a vertex $j$. Thus the vertices play the role of boxes and the $r^L$ paths play the role of balls. In general, these occupancies depend strongly on the geometry of $G$. We show that inserting independent uniform permutations between successive layers restores the classical Poisson behavior at the empirical level. Although the path endpoints remain strongly dependent, their empirical occupancy distribution has the same limiting Poisson law as independent balls thrown into boxes. More precisely, at the critical scale $r^L/d\to\lambda\in(0,\infty)$, provided $rL/d \to 0$, the empirical path multiplicity distribution converges in Wasserstein distance to $\operatorname{Pois}(\lambda)$, independently of the geometry of the underlying graphs. We also prove a local Poisson limit theorem and a Berry-Esseen bound for the empirical distribution functions. These estimates determine the full total-variation transition of the normalized path multiplicities toward uniformity. For arbitrary doubly stochastic kernels, we further establish exact covariance and Frobenius norm identities, together with an exact contraction formula for the chi-square divergence. As applications, we study synchronous opinion averaging, obtaining an explicit cutoff profile for equal-size groups and quantitative convergence to consensus for heterogeneous group sizes.

A method integrating policy gradient with fuzzy inference has been proposed. This method, hereafter referred to as the base model, has high flexibility with respect to the configuration of fuzzy inference rules and fuzzy representation. It can also automatically adjust the rule weights and the membership functions in the antecedent and consequent parts using the same framework for learning. In methods that integrate reinforcement learning with fuzzy inference, however, the reward function must be designed in advance. It is therefore useful to add learning-based reward estimation to the base model. This paper proposes a fuzzy control method that integrates a reward estimation mechanism based on the adversarial inverse reinforcement learning (AIRL) framework into the base model. The generative adversarial nets (GAN) framework adopted in AIRL, however, tends to suffer from unstable learning. To address this, the proposed method treats the policy of the base model as the generator and also uses the set of membership functions in the policy as the cost function of the discriminator. This configuration allows the discriminator update for reward estimation and the generator update for policy estimation to be performed simultaneously on the parameters shared by both, without estimating a value function. In simulations of an illustrative vehicle speed control task, the proposed method acquired policies with high generalization performance through stable and efficient learning.

We investigate signal transmission in a one-dimensional chain of $n$ locally coupled plasma oscillators with asymmetric bidirectional coupling $(\epsilon_1,\epsilon_2)$. Our results reveal a clear contrast between symmetric and asymmetric interactions: while symmetric coupling suppresses signal propagation, coupling asymmetry enables efficient nonlinear transmission across the network. In particular, sustained long range transmission is achieved when the forward coupling strength exceeds the backward coupling strength ($\epsilon_1 > \epsilon_2$), as demonstrated through a comprehensive phase diagram in the $(\epsilon_1,\epsilon_2)$ parameter space. Furthermore, under asymmetric diffusive coupling, signal transmission can persist even when the first few oscillators in the chain undergo oscillation death. A central finding of this study is the intrinsic generation of higher harmonics induced by nonlinear interactions, leading to a pronounced reversal of vibrational resonance. This nonlinear redistribution of energy facilitates selective amplification and robust propagation of specific frequency components.

Test-Time Discovery (TTD) aims to recognize known classes and detect novel ones from a continuous unlabeled test stream after deployment. Existing TTD methods mainly rely on visual clustering or prototype comparison, and can only assign anonymous labels to discovered classes, leaving them difficult to interpret in open-world applications. In this paper, we propose SEGA, a Semantic Grounding with Attributes framework for TTD. SEGA leverages large language models to generate attribute-rich descriptions for known classes and constructs an attribute-anchored visual-semantic space during training. At test time, we introduce Relational Semantic Grounding (RSG), which grounds unknown samples with similar and dissimilar known-class anchors, prompts the LLM to generate candidate concepts, and reranks them in the learned semantic space. The selected concept is further used for confidence-guided test-time semantic stabilization. By coupling attribute-level semantic alignment with relational concept grounding, SEGA transforms anonymous test-time clusters into interpretable semantic concepts. Experiments show that SEGA improves TTD baselines with the promise of semantic grounding for open-world.

World Action Models (WAMs) combine action generation with future prediction, allowing predicted futures to support action evaluation and robot decision making. However, the future associated with a candidate action may contain action-induced and environment-driven changes. Directly evaluating the predicted future may not fully characterize the future change associated with the candidate action. For example, a candidate action that worsens a favorable environmental future may still yield a favorable outcome, while one that improves an unfavorable environmental future may still yield an unfavorable outcome. To analyze this issue, we formulate a Structural Causal Model (SCM) for action evaluation in WAMs. The SCM formalizes how action-induced and environment-driven changes jointly shape future evaluation, and reveals a backdoor path through which environmental evolution can also affect the future evaluation associated with a candidate action. Intervention and backdoor adjustment account for this path, providing a causal basis for characterizing the future change associated with each candidate action. Based on this analysis, we propose Causal Action-Gain World Action Model (CAG-WAM). For each candidate action, CAG-WAM predicts an action-conditioned future and an environmental future under the same current observation, and defines their evaluation difference as causal action gain (CAG) for characterizing the future change associated with the candidate action. During training, we incorporate CAG into Group Relative Policy Optimization (GRPO) to guide robot policy learning. During inference, we evaluate the initial action with CAG and, when needed, revise it using predicted future information. Experiments on multiple benchmarks support the effectiveness of CAG-WAM.

Pharmacokinetic and pharmacokinetic-pharmacodynamic models are used in simulated clinical trials to compare dosing protocols and patient groups before a trial is run. Distinct virtual patient cohorts, representing renal impairment, obesity and overdosage, can impose different input distributions on the same model. Global sensitivity analysis (GSA) identifies which uncertain inputs drive the cohort's predictions, but a direct analysis incurs its full computational cost for every cohort. An alternative accounting is proposed here: a single surrogate is trained on an input envelope wide enough to contain every cohort, each cohort is then analysed using that surrogate, and the reported quantity is the number of cohorts at which the surrogate route has cost fewer mechanistic evaluations than the direct route. The study tests three GSA methods: variance-based GSA, Morris screening and optimal transport. It covers four PK and PK/PD models with seven to twenty-three free inputs, 21 cohorts and five surrogates. Every estimate is scored against a converged mechanistic solution obtained with its own method, and all costs are expressed in mechanistic model evaluations. The parameter ranking is recovered at low cost: gradient-boosted trees reconstruct it on 20 of the 21 cohorts and cost fewer evaluations to train than one direct cohort analysis, on all four models, at break-even counts between 0.02 and 0.14 cohorts. Amortisation is fastest at the highest dimensions, where each direct variance-based sample of a $D$-input model requires $D+2$ evaluations while the training budget does not scale accordingly. Recovering absolute index values is harder, and fewer surrogates manage it as inputs grow: two of the five surrogates fail at fifteen inputs, where the gradient-boosted trees and the Gaussian process still succeed, and at twenty-three inputs no surrogate succeeds on a majority of cohorts.

This article presents an electronic experimental validation of discontinuity-induced bifurcations, including period-adding cascades, in a wake-induced dynamical system subjected to soft impacts. A cylindrical structure immersed in a cross-flow undergoes vortex-induced vibrations and interacts with a movable barrier during the upper and lower branches of the response. The structural motion is modelled as a harmonic oscillator, while the fluctuating lift is represented by a Van der Pol oscillator. The soft contact interaction between the structure and the movable barrier is modelled using a bilinear restoring force, resulting in a piecewise-smooth wake-oscillator system. A systematic numerical investigation reveals period-adding cascades of periodic orbits interspersed with chaotic responses as the barrier distance and fluid-flow velocity are varied. The results are experimentally validated using an inductorless electronic analogue of the piecewise-smooth wake model constructed from operational amplifiers and implemented on a printed-circuit board to improve experimental robustness. The route to instability of the periodic and aperiodic responses is further characterized using Floquet multipliers and Lyapunov exponents. A zero-time transverse discontinuity mapping is implemented to account for the evolution of perturbations in the local neighborhood of the discontinuity boundary during stability analysis. The combined numerical, experimental, and stability analyses demonstrate the role of soft impacts in generating complex nonlinear responses in the wake-induced system.

Geometrically exact beams naturally arise in physical phenomena and engineering practice for modeling flexible structures, such as soft robotic manipulators and DNA structures. To the best of our knowledge, there appear to be no geometric integrators for stochastic geometrically exact beams in a variational formulation. In this paper, we introduce variational principles for the dynamics of stochastic geometrically exact beams, in both continuous- and discrete-time settings. By introducing perturbations via stochastic potentials and stochastic kinematics, we derive the equations of motion from a variational viewpoint. The resulting stochastic geometric integrators simultaneously discretize time and space in a parallel manner, which provides a distributed computational framework for solving the stochastic dynamics of complex beam systems. In particular, the proposed integrators are shown to be a stochastic extension of the well-known deterministic staggered leapfrog scheme for the wave equation, and also a stochastic extension of Yee's scheme for spatially one-dimensional computational electromagnetism. Numerical results are shown to demonstrate the effectiveness of the proposed approaches.

Random walk (RW)-based algorithms are widely used in decentralized learning due to their low communication overhead and scalability. While most existing methods rely on a single token, multiple tokens can explore the network in parallel and exchange information upon meeting. However, the fundamental benefits of such meeting-triggered interactions remain unclear. In this work, we investigate multi-token RW learning under two information exchange policies: Average-Gradient and Average-Model, which average stochastic gradients and updated models, respectively, when tokens meet. We establish iteration complexity lower bounds for non-convex, convex, and strongly convex objectives on complete graphs, which apply to both exchange policies. Compared with single-token RW learning, the scaling factor in these bounds is reduced by a factor of $m^2$ for non-convex and convex objectives, and by an exponentially in $m$ for strongly convex objectives, where $m$ denotes the number of tokens. Numerical experiments reveal distinct behaviors of the two policies under two performance criteria: mean-token loss (the average loss across all tokens) and best-token loss (the minimum loss across all tokens). Average-Model achieves lower mean-token loss and is relatively insensitive to the network-to-token ratio $N/m$, where $N$ is the number of nodes. In contrast, Average-Gradient achieves lower best-token loss but exhibits a trade-off between parallel exploration and gradient interference as $N/m$ decreases.

Early-stage design of heterogeneous UAV computing systems requires evaluating how algorithm, hardware, communication, and system topology affect closed-loop flight before the complete system is available. However, design choices interact through the robotic feedback loop, making task performance difficult to predict from isolated component metrics, while existing execution-driven approaches postpone system-level design feedback until the computing system is integrated. This work presents C2Flight, a closed-loop co-simulation framework that represents target systems as parameterized dataflow graphs and integrates models of hardware under design with independently accessible physical hardware through a unified node interface. Its graph runtime captures the combined effects of buffering, activation, resource sharing, communication, and local timing on closed-loop flight behavior. Validation against onboard and wireless offboard physical computing pipelines confirms its timing fidelity and closed-loop prediction accuracy. Two case studies demonstrate early-stage exploration of deployment and accelerator bit-width choices. These results show that C2Flight exposes system-level design trade-offs before full-system integration, enabling rapid early-stage design refinement.