Papers

We describe the Nürnberg NLP system for ChildSafeAds 2026. The shared task asks what a monitoring system for commercial content in child-facing YouTube videos can achieve at a given level of data access. We answer with per-subtask ensembles of nine voters, organised into three branches that differ in backbone, adaptation method and class scope. Selection rests on channel-disjoint cross-validation, with the development set as a transfer check. The system wins two of the three subtasks. Its product-category score (ST2, 0.8243) and its compliance-flag score (ST3, 0.6530) are the best of the 22 final entries, and it places third on the task mean (0.7079). We further compare four access levels and report the cost at test-set scale.

Multi-reward policy optimization requires a joint update that reflects both the learning signals and the intended relationships among objectives. We introduce Objective-wise Reconciled Policy Gradient (ORPG), which constructs a separate clipped policy objective for each reward and reconciles the resulting gradients into one policy update. For compatible gradients, a cosine-dependent interpolation coordinates their contributions through a partially normalized reference while preserving the norm of their sum. We characterize this update as the unique solution of a spherical directional compromise. For conflicting gradients, projection follows the task's priorities. We evaluate the same compatible rule in helpfulness–safety alignment and correctness–cost optimization for mathematical reasoning. ORPG substantially improves average Useful and Harmless scores over the strongest external baseline on each axis. In mathematics, it achieves the highest average full-budget accuracy and three-budget hypervolume among the compared methods, with more accurate and shorter responses than the initial policy. Component comparisons and training dynamics show the larger contribution of compatible coordination and a complementary benefit from conflict handling. These results support gradient reconciliation for objectives with equal standing and for objectives with an explicit priority.

Super-leading logarithms from Glauber exchange complicate the resummation of non-global observables at hadron colliders. In this Letter, we extend Marzili, a Monte Carlo framework for non-global resummation based on effective field theory, to resum the leading nonzero colour contribution involving two Glauber exchanges. This is the first Monte Carlo resummation of super-leading logarithms. Together with Marzili's treatment of non-global logarithms, this opens a path towards studying their interplay and incorporating Glauber exchange into parton showers.

Smart contracts underpin decentralized finance, where growing demand for on-chain/off-chain communication(OFC) has driven diverse applications such as cross-chain bridges, real-world asset tokenization, and fiat-backed stablecoins. TheOFC-related security incidents in these applications are increasingly frequent, but prior studies address separate vulnerability categories within OFC applications rather than providing a unified view, causing vulnerabilities outside known patterns to be missed.In this paper, we identify OFC inconsistency (OFCI) as a root cause of OFC vulnerabilities, which arises from business-logic flaw and ultimately breaks the equivalence between the on-chain and off-chain asset representations to induce inconsistency.Automatically detecting OFCIs faces two challenges including (1)locating heterogeneous business logic, and (2) transferring existing vulnerability knowledge to identify unseen OFCI instances. To this end, we propose SmartMemory, the first framework to leverage a memory-based agent for OFCI detection. To address heterogeneity, SmartMemory maps diverse implementations ofOFC contracts into a canonical business-semantic representation to locate the business logic for OFCI inspection. For knowledge reuse, SmartMemory integrates a memory-based agent to distill vulnerability knowledge from features into patterns and detection rules, enabling knowledge transfer across cases to identify unseenOFCIs. Lastly, SmartMemory performs taint analysis to verify the reachability, type, and impact of each candidate OFCI. We construct the first real-world OFCI dataset comprising 48 DApps with 81 OFCIs for evaluation, on which SmartMemory achieves80.68% precision and 87.65% recall. In addition, through an analysis of 325 real-world OFC applications, SmartMemory detects 36 previously unknown OFCIs, all of which have been confirmed and fixed by corresponding parties.

Vision-Language-Action (VLA) models have shown great promise for robotic manipulation by mapping multi-modal semantics to physical actions. However, this mapping inherently struggles to align these coarse-grained semantics with fine-grained temporal execution. It leaves VLA models with limited generalization and insufficient robustness in cluttered environments. To overcome this issue, we propose ActionUNet, an efficient multi-scale fine-tuning framework that enhances pre-trained VLA models with minimal computational cost. ActionUNet first constructs a lightweight temporal U-Net within the temporal-aligned action feature space to fuse hierarchical structural priors, effectively bridging the scale gap between semantics and temporal executions. Recognizing that multi-scale modeling can disrupt microscopic temporal continuity and cause mechanical oscillations, ActionUNet then employs a conditional SIREN as a continuous action decoder. Equipped with explicit second-order smoothness constraints, this decoder guarantees temporal continuity and reduces high-frequency motion jitter. By smoothing temporal discontinuities from multi-scale fusion, this continuous formulation reduces mechanical execution failures while preserving the base VLA model's generalization and manipulation robustness. Extensive experiments on RoboTwin 2.0 and LIBERO-Plus benchmarks, together with real-world hard evaluations, demonstrate that ActionUNet significantly improves {\pi}0.5 success rates by absolute 9.8%, 6.1%, and 11.4%, respectively, while also generalizing to the regression-based OpenVLA-OFT backbone, highlighting its effectiveness and efficiency as a fine-tuning strategy. Code and implementation details are available at https://github.com/Di-Zhu123/ActionUNet.

World action models (WAMs) predict the future alongside actions during training. Due to the heavy computation cost of video denoising, whether the future must still be generated during inference is disputed: Explicit WAMs denoise it into clean frames along with every action chunk, whereas Latent WAMs discard it entirely for acceleration. We find that latent WAMs, despite matching explicit ones on in-distribution tasks, fail to retain the generalization benefits that originally motivated WAMs. To demonstrate this, we evaluate generalization along three axes: environmental perturbation, data efficiency, and task generalization. Controlled comparisons with a matched backbone, training data, and budget reveal consistent degradation across all three axes when the action expert no longer conditions on future representations. Further analysis shows that the gap arises almost entirely from the first denoising step: the benefit comes from preparing the future, not generating it. We therefore propose Simple-WAM, which simplifies future modeling into a single forward pass of fully noised video tokens and adapts the training-time noise schedule to this inference behavior. Across simulation and real-world tasks, Simple-WAM achieves the best of both worlds, leading explicit WAMs in generalization performance with efficiency comparable to Latent WAMs. Project Page: https://zrporz.github.io/Simple-WAM-Web/

We construct and analyze a renormalized three-dimensional massive Lee model; its renormalized Hamiltonian is defined intrinsically through an abstract construction which explicitly provides the operator domain, action and a Kre\u\i n-like resolvent formula. Conservation of the total excitation number reduces the spectral analysis of the full Hamiltonian to a sector-wise study of the spectral profile of an explicit operator pencil, playing the role of an operator-valued Weyl function. We construct essential spectrum components in a sector generated by lower excitation sectors and obtain the inclusion in the expected sector-wise HVZ type formula; the full formula is proved in the first two sectors and, under explicit conditions, in every sector. Below the free or one-excitation threshold, the point spectrum in every higher sector is described by a bounded self-adjoint nonnegative Birman–Schwinger operator. In every sector with at least two excitations we prove absence of sub-threshold spectrum for arbitrary coupling when the renormalized gap is not smaller than the boson mass, and under an explicit weak-coupling condition when it is smaller. In the latter case we also derive a sufficient condition for the existence of a simple bound state in the two-excitation sector.

Bound states in the continuum (BICs) enable complete wave confinement within open systems through destructive interference or symmetry protection, giving rise to ideally infinite quality factors and extreme field enhancement. Their practical exploitation, however, is hindered by a fundamental paradox: the same mechanism that suppresses radiation also prevents efficient excitation from the far field. Existing experimental approaches typically overcome this limitation by introducing external coupling channels that inevitably perturb the protected state and reduce its confinement. Here we demonstrate the non-invasive excitation of a symmetry-protected acoustic BIC through nonlinear virtual sources generated directly at the surface of the structure. By focusing time-modulated ultrasonic beams onto a metamirror, nonlinear demodulation produces localized low-frequency sources whose spatial phase matches the symmetry of the target mode while avoiding any physical coupling to the resonator. This approach injects energy into the BIC without opening additional radiative channels, preserving its intrinsic non-radiative character. We develop a theoretical framework describing the formation and symmetry protection of the mode, realize a compact experimental implementation through symmetry reduction, and directly measure the confined pressure field and its quality factor. Our results establish nonlinear virtual sources as a route to accessing symmetry-protected states in open systems while maintaining their exceptional confinement properties, providing a general strategy for wave manipulation and high-sensitivity resonant devices across physical platforms.

Consumer earbuds already stream inertial motion data from the head, one of the most widely worn sensor locations on the body. We ask how much of the 3D body pose a single such head IMU can recover, and whether adding more consumer sensors actually helps. We build a multimodal capture pipeline that records four-view RGB-D video together with an AirPods head IMU and two Striv insole IMUs, synchronize the streams post-hoc, and generate pseudo-ground-truth with SAM 3D Body, yielding a 35-take single-subject benchmark spanning gait, turning, vertical, everyday, and clinically inspired motions. Adapting two recurrent model families (IMUPoser and MobilePoser), we show that one head IMU recovers lower-body pose at 79.0 mm rigid-MPJPE and per-foot ground contact at 0.809 macro-F1, and that a causal variant retains most of this accuracy at streaming latency. In paired per-take significance tests across both families, adding the consumer foot IMUs never significantly improves pose and significantly degrades it in two of four model-split combinations; a mounting-bias probe and feet-only ablation identify insole orientation quality, not foot placement, as the mechanism. Extending the output to a 20-joint full-body skeleton maps the boundary: gross distal-arm motion is partially recoverable from the head alone, proximal upper-body pose is not, and staged fine-tuning recovers the leg accuracy that naive joint training sacrifices to multi-task dilution. For learned pose from consumer wearables, sensor reliability, not sensor count, is the binding constraint here. For the devices tested, the earbud is its sweet spot. Code is available at https://github.com/ZhilinGuo/one-sensor-whole-body.

Multimodal Large Language Models face significant efficiency challenges that stem from two distinct yet coupled sources: data redundancy and computational redundancy. While most methods focus on data redundancy by pruning visual tokens from the output of the visual encoder or computing redundancy in LLM decoders using blockwise importance, the finer-grained inter-layer representation shifts and the distribution differences within the layers themselves have not been fully explored. In this work, we comprehensively investigate this dual-level inefficiency. We posit that intermediate layer tokens from vision encoders should be considered for effective visual token pruning, as semantic focus shifts across layers, with middle-layer tokens capturing more detailed object-centric information that deeper layers may abstract away. Furthermore, we reveal the differential contributions of Attention and FFNs across distinct LLM decoder layers. Building upon these discoveries, we propose SPIDER, a training-free framework that integrates multi-layer \underline{S}emantic visual token \underline{P}run\underline{I}ng with an a\underline{D}aptive sub-lay\underline{ER} skipping mechanism. Experimental evaluations demonstrate that SPIDER consistently maintains strong performance across various MLLM architectures and reduction ratios. For instance, on LLaVA-NeXT-7B, SPIDER reduces FLOPs by $79\%$ while maintaining 96$\%$ of the baseline performance.

The integration of artificial intelligence into computer-aided design frameworks has sparked a shift in the design of analog integrated circuits (ICs), transitioning the field from using manual and algorithmic-based solutions to adopting automated and intelligent paradigms. In this scenario, the GDSII file represents the industry-standard database containing the ultimate and most accurate source of information of the analog circuit, encapsulating the complex physical geometries and parasitic realities that define tape out performance. This paper proposes a novel framework that combines fine-tuned LLMs and CNNs to analyze GDSII files of analog circuits, enabling a conversational interface between the tool and the designers. Experimental results using thousands of analog designs across four realistic tasks demonstrate that the proposed solution outperforms state-of-the-art general-purpose massive VLMs by a significant margin (up to 81%), thus providing a lightweight solution to the problem of GDSII analysis.

Group Relative Policy Optimization (GRPO) improves language-model reasoning by comparing verified rewards among multiple solution rollouts for each query. However, difficult training queries can yield only incorrect rollouts, leaving GRPO with no reward contrast or learning signal. Prior hint-based methods construct auxiliary hints from solution evidence and use them to re-solve failed queries, recovering learning signal. Yet the resulting trajectories are typically treated as ordinary solution trajectories despite being generated under an assisted condition unavailable at evaluation. We discover hinted reward shift: recovered reward contrast can concentrate policy updates on hinted trajectories, limiting improvement without hints. This also creates a trade-off: increasing hinted trajectories can accelerate early learning but intensify reward shift later. To address this problem, we propose HATCH (Hint-Annealed Self-Teaching), an online single-policy framework that learns from both generating and using its own hints to improve reasoning without assistance. To mitigate hinted reward shift, we introduce online weighting to anneal the contribution of hinted trajectories. However, learning to generate hints can conflict with improving query solving. We therefore use gradient projection to remove the opposing component of hint-generation updates. Together, these designs support self-improvement by enabling the policy to create learning opportunities for itself and turn them into stronger reasoning without hints. We evaluate our method on mathematical reasoning benchmarks and outperform state-of-the-art methods by 1.02 pp on Llama-3.2-1B-Instruct, 2.84 pp on Qwen3-1.7B, and 4.32 pp on Qwen3-8B.

LLM-based Web agents can autonomously complete user tasks, yet deceptive interfaces can steer them toward outcomes that conflict with users' interests. Existing defenses primarily intervene on agent behavior through blocking, guidance, or replanning. We identify a distinct failure mode: a task-valid action can still realize an unauthorized consequence because of the current Web state. This motivates treating task-relevant Web state itself as a runtime control target. We introduce Veer, an agent-side runtime defense that leaves task planning to the base agent and intervenes on Web state when a proposed action would produce an unauthorized consequence. Before modifying the live environment, Veer constructs a prospective intervention trajectory toward a safe task-relevant state and executes it with runtime grounding and verification. Across TrickyArena and WebDecept, Veer achieves the highest safe task completion in all three evaluation settings, exceeding the next-best defense by 15.9 and 25.0 percentage points on TrickyArena-Single and TrickyArena-Multi, respectively, while reducing dark-pattern success on WebDecept to 0.3%. These gains persist across dark-pattern types and all 12 agent, model, and benchmark configurations. Ablations show that active state intervention provides the largest gain, while prospective rollout and temporal evidence contribute additional improvements. These results establish task-relevant Web state as an effective runtime control target for protecting Web agents from deceptive outcomes.

Full-duplex voice agents can now listen, speak, use tools, and act during spoken interactions, but fluent dialogue does not guarantee correct completion of delegated professional workflows. We introduce APEX-Voice, a benchmark of 120 interactive professional workflows spanning ten work archetypes such as form completion, corporate negotiation, coordination, consulting, and interviewing. Each workflow executes in a stateful Voice Workbench environment with task-specific knowledge, typed tools, gold-annotated final work artifact, authorization constraints, and a user simulation policy backed by validated, pre-compiled speech realizations. We evaluate both artifact field accuracy and end-to-end workflow success, which requires the correct terminal state, valid process, completed actions, and a valid final artifact. Across five frontier real-time voice agents-GPT-Live-1, Gemini-3.8-Live, Grok-Voice-Think-2.0, Step-Audio3, and GPT-realtime-2.1, none exceeds 25% Pass@1, and the best Reliable@3 is only 10.8%. Moreover, stateful coordination is the dominant failure point across systems, while success decreases further on workflows requiring greater knowledge retrieval and mid-speech corrections. Overall, APEX-Voice is the first benchmark for evaluating whether voice agents can translate conversational competence into dependable professional work.

Long visual token sequences often account for a substantial fraction of the computational overhead in multimodal large language models~(MLLMs). Existing approaches reduce this cost by pruning redundant visual tokens, but permanently discard visual evidence that may become useful in subsequent layers. We instead ask whether all visual tokens can be preserved while reducing the cost of repeatedly evolving the representations through the Transformer. To answer this question, we perform low-rank interventions on visual-to-text information flow. We find that, after visual-to-text attention is blocked, restoring only a few directions recovers most of the lost accuracy, suggesting the relevant visual influence is concentrated in a low-dimensional subspace. We further observe strong predictability in layer-specific visual states: lightweight MLPs approximate them with high cosine similarity and low reconstruction error. Motivated by these findings, we propose $\delta$-Vision, which replaces repeated Transformer evolution of visual tokens with lightweight low-rank adapters that construct layer-wise visual memories while preserving all visual tokens for text retrieval. Across image and video benchmarks, $\delta$-Vision achieves higher accuracy than visual token pruning baselines at comparable or lower computation, while delivering competitive inference efficiency without discarding visual tokens.

Large Language Models (LLMs) have shown strong performance on tool-use agentic tasks when given a fixed tool schema. Yet a tool schema is not the action space of an agent; it is merely one interface representation of it. The same executable action can be exposed through many different, functionally equivalent tool definitions, and an agent that has truly learned a task should behave consistently across them. We show that current agents often do not, a phenomenon we term schema bias. To study this systematically, we introduce an executable transformation framework that rewrites a native tool schema using nine operators, including merging and splitting tools, altering how a single tool is expressed, and distributing one action across several dependent calls. The tasks, executable actions, and reachable states remain fixed, so any change in success is attributable to the interface alone. Evaluating eleven LLMs, including two closed models, on up to 32 schema variants, we ask how large schema bias is, how it manifests, whether the difficulty of a schema variant can be predicted without a full evaluation, and whether training removes it. We find that schema bias is substantial even for the newest models: success rates range from complete failure to 97% depending solely on the schema. To reliably estimate schema difficulty, it requires running a small sample of the target queries. Training repairs a schema variant only when that variant appears in the training data.

Safety-aligned LLMs can exhibit emergent misalignment (EM): narrow domain adaptation unexpectedly triggers catastrophic safety failures across unrelated domains. Prior static analyses leave training dynamics unmapped, while existing defenses rely on heuristics that degrade utility. We present a dynamic, second-order geometric study of EM. Tracking training trajectories reveals that directional Hessian curvature concentrates sharply on semantic pivot tokens. Grassmannian projections show that, in most settings, harmful-safe gap widens mainly because safe-gradient overlap declines. Leveraging these insights, we introduce a parameter-level Geometric Mitigation Framework that orthogonally projects empirical harmful gradient subspace out of parameter updates. On Qwen2.5-14B-IT, our defense suppresses free-generation EM by up to 80.0%; across the other three of four open-weight instruction-based model families (3B–20B), where single-layer behavioral EM is already near zero, teacher-forced evaluation shows same harmful subspace controls the conditional support of frozen EM responses. Crucially, these diagnostics unmask the illusion of behavioral safety: the same subspace remains measurable and steerable in models where behavioral EM is near zero. Code: https://github.com/WeiqiaoQUE/mechanistic-emergent-misalignment.

Recent zero-shot Vision-and-Language Navigation (VLN) methods increasingly rely on multimodal large language models (MLLMs) to reason over visual observations, navigation instructions, and candidate actions. Although effective, repeatedly invoking autoregressive multimodal reasoning at every navigation step introduces substantial inference latency, limiting the responsiveness of embodied agents. We propose NavJev, an efficient VLN framework that reformulates online navigation from repeated multimodal generation into compact visual compression followed by lightweight typed action selection. Specifically, Action-Centric Visual Compression (ACVC) integrates waypoint geometry, BLIP captions, and RAM semantic tags into compact representations of candidate actions, while Discriminative Action-Semantic Memory (DASM) filters shared semantics and maintains discriminative action-specific evidence across navigation steps. Based on these representations, Jev directly performs structured probabilistic decisions over the available action set. Experiments on R2R-CE show that NavJev achieves 27.0% SR and 22.4% SPL with only 0.65 s per navigation step, while substantially reducing inference latency and cost compared with MLLM-based VLN methods. The project page is available at https://kai-sheng-caesar.github.io/NavJev/.

Vision-language-action and world-action models are increasingly popular, yet remain bottlenecked by physical interaction data that is scarce, institutionally siloed, and task-heterogeneous. A natural federated solution is to let each client adapt a shared foundation model through parameter-efficient fine-tuning, avoiding the exchange of full-model updates. However, federating these adapters is nontrivial, as naive aggregation can entangle incompatible updates, while incorporating MoE-style routing into federated aggregation may dilute specialization and destabilize expert selection. We present RoboFL, which instantiates MoSAIC (Mixture of Slotted Adapters) for federated world-action learning. MoSAIC directly installs locally trained LoRA adapters as the expert branches of a server MoE. Server-side routers learn token assignments over these prior-informed branches while jointly refining routing and expert parameters. Foresight-to-Action Routing Distillation (FARD) aligns routing across the model's three paths, while Path-Consensus Expert Aggregation (PCEA) converts complete expert updates into a compact global adapter for personalized redistribution. Experiments on RoboTwin 2.0, RLBench, and a real-world Franka robot arm show the superiority of RoboFL with structured expert assembly, as it outperforms centralized PEFT InternVLA-A1 by 12.23% on the Franka arm, while reducing per-round client communication by up to 86.81% relative to MoE-based federated VLA baselines.

Semantic uncertainty quantification for large language models rests on a common template: sample several answers, measure how much they agree, and treat disagreement as uncertainty. We first formalize this template as two separate roles: an operator that compares two answers, and an aggregator that combines all pairwise comparisons into a scalar. Existing methods differ almost entirely in how they aggregate, while taking the operator off the shelf, typically an NLI model or a generic sentence encoder. We show that this reliance on off-the-shelf operators is the primary bottleneck of semantic UQ: they do not accurately measure factual equivalence of multiple answers to the same question. We resolve this with a deliberately simple recipe: a single encoder trained contrastively to isolate the targeted fact, utilizing synthetic data generated by an LLM and dataset both disjoint from all evaluation settings. Integrating the resulting operator into existing methods improves performance on 120 of 126 evaluation settings (95%) spanning 18 model dataset combinations across language and vision-language models. The best variant reaches 0.76 mean AUROC against 0.68 for the strongest baseline, while replacing the quadratic cross-encoder comparisons of entailment-based operators with one encoder pass per answer. The uniformity of the improvement supports the view that the operator, not the aggregator, is the limiting factor. The same operator also improves single generation token-level estimators: the norm it assigns to each token measures how much that token bears on the answer, and reweighting token log-likelihoods accordingly sharpens the estimate.

Greenhouse climate control balances economic return with maintaining temperature, humidity and CO2 within crop-adapted growth ranges. Conventional reinforcement learning (RL) greenhouse controllers use fixed reward penalties to limit climate constraint violations, yet such heuristic penalties cannot explicitly constrain long-term cumulative violations. Poorly tuned weights either lead to overly conservative policies and lower yields, or fail to suppress persistent climate deviations that harm photosynthesis and induce crop diseases. To address this issue, we formulate greenhouse climate regulation as a Constrained Markov Decision Process (CMDP) and use a Lagrangian safe RL framework RCPO-PPO to separate economic optimization and cumulative safety constraints, enabling adaptive penalty adjustment without manual tuning. To handle strong nonlinear, time-varying coupling between greenhouse microclimate and crop growth, Kolmogorov-Arnold Networks (KANs) replace Multi-Layer Perceptrons (MLPs) as policy and value approximators for improved nonlinear representation. Sinusoidal cyclic time features are embedded in observations to capture diurnal environmental periodicity. Simulations use a classic winter lettuce greenhouse model driven by 40-day real weather disturbances. Compared with vanilla penalty-based PPO, our method cuts cumulative climate violations by 18.65% and raises lettuce economic profit by 2.91%, keeping violations stable near the safety threshold. This decoupled CMDP optimization with KAN-based policy representation mitigates long-term climate risks and boosts planting profits, offering a constraint-aware control strategy for precision greenhouse cultivation.

Many physiological time series, such as cardiac and brain recordings, exhibit cyclostationarity: their statistics vary periodically with an underlying cycle phase. Corruption from motion, poor contact, and physiological interference obscures morphology needed for diagnosis, making signal restoration essential. Existing diffusion approaches condition on corrupted observations alone and must learn cyclic structure implicitly. We instead propose two inductive biases which encode cyclostationarity: a shift-covariant wavelet representation and dense per-sample phase conditioning inferred from the corrupted input. We further introduce a training-free cyclostationarity index that quantifies phase structure and predicts when phase conditioning will help. Finally, we propose antithetic coupling of reverse trajectories to reduce sampling variance while achieving comparable performance with fivefold fewer network evaluations. Across modalities, our results show that explicitly encoding measurable cyclic structure improves physiological time-series restoration.

Preparation contextuality is often manifested operationally through a quantitative advantage over preparation-noncontextual models in information-processing tasks, yet quantum theory typically falls short of perfect success. Here we introduce a parity-oblivious Hidden Matching task that instead yields an all-vs-nothing manifestation of preparation contextuality. Alice encodes an $n$-bit string so that her message reveals no information about any input parity other than the two-bit parities. Bob, given a perfect matching on the input positions, must output an edge of the matching together with the parity of its two bits. A quantum protocol using a single $\lceil\log_2 n\rceil$-qubit message satisfies the parity-obliviousness constraint and succeeds with certainty. We prove that no preparation-noncontextual model achieves perfect success for any even $n\ge6$. The result holds for arbitrary ontic state spaces and assumes neither outcome determinism nor measurement noncontextuality. For $n=6$ and $n=8$, the optimal noncontextual success probabilities are $4/5$ and $3/4$, respectively, while asymptotically the optimal noncontextual success probability is $1/2+\Theta(1/\sqrt n)$. Our asymptotic upper bound follows from a Fourier-analytic sum rule derived via hypercontractivity on the Boolean hypercube. The resulting separation is qualitative rather than merely quantitative: quantum theory achieves perfect success, whereas preparation-noncontextual models cannot. We aslo show that at $n=8$ the noncontextual bound remains $3/4$ even when Bob is restricted to just suitable $4$ of the $105$ possible perfect matchings, bringing the effect within experimental reach.

Large language models can generate functionally correct code that still contains security weaknesses, motivating repair pipelines that first diagnose a weakness type before deciding how to fix it. The Common Weakness Enumeration (CWE) provides a standardized vocabulary for such diagnoses, but asking an autoregressive language model to generate a CWE label and extracting it from the response raises questions about output validity, speed, and cost, as well as accuracy. We evaluate Jev, a decision model that instead selects directly from a declared set of candidates and returns a probability for each, against six open-weight autoregressive models and a frontier proprietary model, GPT-5.6-Sol, on a controlled 50-way CWE classification task over 1,916 CyberSecEval benchmark examples. Jev outperforms all six open-weight baselines on every classification and ranking metric, while its comparison with GPT-5.6-Sol depends on the metric: GPT-5.6-Sol achieves higher Top-1 accuracy and Macro-F1, whereas Jev achieves higher Top-3 and Top-5 accuracy and a nearly identical MRR, at $6.27\times$ lower median API latency and $55.9\times$ lower estimated API cost. We further build JevVibe, a diagnosis-guided repair agent that uses predicted CWE labels to repair code generated by Qwen2.5-Coder-32B-Instruct. With Jev providing the diagnosis, the agent increases the detector-measured security pass rate from 63.5% before repair to 70.7%, compared with 66.1% for LLM-guided repair. These results show that JevVibe is effective at improving the security of generated code, with Jev providing reliable and efficient CWE classification.

Estimating mutual information (MI) from samples is a central objective in a variety of scientific fields. Modern neural estimators are accurate in the large-data regime, but they fall short when data is scarce, and each must be fit anew for every distribution under study. Current estimators are moreover tied to specific data types. These constraints limit their adoption in many applications where per-distribution training is impractical and sample sizes are small. We present ALICE, a foundation model that removes per-distribution training, while achieving competitive estimation accuracy. Trained exclusively on a broad family of synthetic distributions, ALICE acts as an in-context estimator of rectified-flow velocity fields: conditioned on samples of an unseen distribution, it estimates that distribution's velocity field without any explicit training. MI is then obtained through a fixed identity that integrates the squared difference between the joint and conditional fields. We validate ALICE on a standard, challenging benchmark and apply it in three domains, biology, genetics, and neuroscience, whose data the model has never seen. For the first time, we show that a single model closes the gap with neural estimators trained separately for each distribution, while natively supporting different data dimensionality and sample cardinality, enabling zero-shot MI analysis across scientific domains.

We build a Floer theory on the space of braids of periodic orbits of a Hamiltonian flow on a closed symplectic surface. The differential and continuation maps are defined by counting Floer isotopies of braids: tuples of Floer cylinders with pairwise disjoint graphs. With this new perspective, we prove a quantitative braid stability result that shows the persistence of braids under possibly large Hamiltonian perturbations. As an application, we show that for every $\alpha\geq 0$ there is a sequence of Hamiltonian diffeomorphisms $\phi_k$ on the two-torus $T^2$ such that $$d_H(\operatorname{Ent}_{\leq \alpha}(T^2,\omega),\phi_k)\to \infty \quad (k \to \infty),$$ where $\text{Ent}_{\leq \alpha}(T^2,\omega)\subset \text{Ham}(T^2,\omega)$ denotes the set of Hamiltonian diffeomorphisms with topological entropy at most $\alpha$ and $d_H$ the Hofer metric. We prove this result by studying only contractible periodic orbits, whereas analogous higher-genus statements were previously obtained using non-contractible orbits.

Formalizing research-level stochastic optimization in Lean requires both an algorithm model and domain theory connecting foundational libraries to convergence proofs. Revising a model to restore provability can change the mathematical claim. We introduce ProofLoom, a fully automated LLM-agent system for Proof-Obligation-Driven Theory Construction. Given a published algorithm, target theorem, and source proof, ProofLoom autonomously constructs the Lean model and supporting theory. Open proof obligations drive the development of definitions, interfaces, lemmas, and proof plans. Signature contracts record evidence and obligations for model revisions; an independent Judge rejects unsupported assumptions and weakened conclusions. Planner expands the published argument into intermediate claims, and Audit checks whether the Lean proof follows it. Across tasks, SOptLib accumulates verified mathematics and construction experience: reusable results are extracted, generalized, and verified, while modeling decisions and failed proof routes are recorded. Later tasks retrieve these results and records and contribute new developments, forming a cycle of construction, accumulation, and reuse. On fifteen textbook and research-paper tasks, ProofLoom obtains mean human ratings of 6.3/7 and 6.4/7, compared with 4.9/7 and 5.0/7 for the strongest of six baselines. Across 33 developments, it produces 490,693 lines of algorithm-local Lean code with no sorry. The formalizations also expose 28 incorrect formulas, proof gaps, and algorithm-analysis mismatches in published sources across 22 developments, each with checked evidence.

We propose a novel joint sampling design for surveys involving two interrelated populations of parents and children in the presence of complex family structures, including shared parental responsibility and blended families. The design combines a graph-theoretic representation of co-parenting relationships with balanced sampling. This framework ensures that, in the first sampling phase, exactly one parent is selected from each co-responsibility pair while maintaining accurate inference for both the parent and child populations. A second sampling phase selects one child per sampled parent using unequal probabilities that produce consistent weighting and unbiased estimation. The proposed framework accommodates overlapping family relationships while balancing auxiliary information for both populations. Results based on the Luxembourg population registers show that the estimated totals are virtually identical to the known population totals for all balancing variables, and other auxiliary variables, demonstrating the excellent precision and efficiency of the proposed sampling design.

We resolve an open problem posed by Frank–Loss. Let $n\ge 3$ and let $\varphi\in L^p(\mathbb{S}^n)$, with $\frac{n}{n-1}<p<\infty$, be a nontrivial Dirac zero mode on $\mathbb{S}^n$, i.e. a nonzero spinor satisfying \begin{equation*} D\varphi = iA\cdot\varphi, \end{equation*} where $D$ is the Dirac operator and $A$ is a vector field with $\mathrm{d} A^\flat\in L^{n/2}$. We prove the sharp lower bound \begin{equation*} \|\mathrm{d} A^\flat\|_{\frac{n}{2}} \ge 2\left[\frac{n}{2}\right]^{-\frac12}\frac{n-1}{n-2}S_n = \left[\frac{n}{2}\right]^{-\frac12}\frac{n(n-1)}{2}\omega_n^{\frac{2}{n}}. \end{equation*} Equality is attainable if and only if $n$ is odd; in that case, modulo conformal and gauge transformations, $\varphi$ is a Killing spinor and $A$ is a real multiple of the Reeb field associated with $\varphi$. The case $n=3$ was proved in our recent paper using a different method. In the paper we divide the remaining cases into 3 cases: i) $n\ge 5$ is odd, ii) $n\ge 5$ is even and iii) $n=4$. All these cases need to use different methods. For $n\ge 5$, the argument crucially reduces to an improved Sobolev inequality on $\mathbb{S}^n$ under a barycenter constraint. Specifically, for $u\in W^{1,2}(\mathbb{S}^n)$ we consider \begin{equation*} \mathfrak{a}_n := \inf\Bigg\{ \frac{ \int \Big(|\nabla u|^2 + \frac{n(n-2)}{4}u^2\Big) - \frac{n(n-2)}{4}\omega_n^{\frac{2}{n}}\|u\|_{\frac{2n}{n-2}}^2 }{ \omega_n^{\frac{2}{n}}\|u\|_{\frac{2n}{n-2}}^2 - \int u^2 } \,\Bigg|\, \int x|u|^{\frac{2n}{n-2}}=0,\ \omega_n^{\frac{2}{n}}\|u\|_{\frac{2n}{n-2}}^2 - \int u^2>0 \Bigg\}, \end{equation*} and we obtain the following universal estimate \begin{equation*} \mathfrak{a}_n > \frac{n(n-2)}{4(n^2-3n+1)}, \end{equation*} which is enough for our aim. Determining the exact value of $\mathfrak{a}_n$ remains open.

We investigate static, charged, asymptotically flat black holes in logarithmic nonlinear electrodynamics (LNED) within the $(Q,\epsilon)$ parameter space. We distinguish three phase structures: the horizon geometry, the background circular null orbits, and the circular photon orbits of the effective optical geometry. All share a zero-radius boundary $Q_{nc}(\epsilon;M)=\gamma M^{2/3}\epsilon^{1/6}$, but possess distinct critical structures: the extremal curve $Q_{ext}(\epsilon;M)$ and the background merger curve $Q_{m}(\epsilon;M)$ both terminate at the bifurcation point $(Q_{b},\epsilon_{b})$, located at $r=0$. In contrast, the effective photon-orbit merger curve $Q_{m}^{ph}(\epsilon;M)$ does not pass through this point. For $\epsilon\ge\epsilon_{b}$, approaching $Q_{nc}$ shrinks the horizon to zero radius, resulting in black points with non-zero, quantum-corrected entropy. Their Hawking temperature vanishes at the bifurcation point and diverges for $\epsilon>\epsilon_{b}$, except when $\eta\delta=1$, where the temperature at the bifurcation point also diverges. The background potential $V_{eff}^{bg}(r)$ develops an infinite barrier at $r=0$ without local extrema, acting as a perfect repeller. In contrast, the photon potential $V_{eff}^{ph}(r)$ is regularized with a maximum at a finite radius, acting as a partial reflector. Consequently, the black points retain photon spheres of finite radius. Away from $Q_{nc}$, for $\epsilon<\epsilon_{b}$, both geometries support horizonless structures above $Q_{ext}$; for $\epsilon\ge\epsilon_{b}$, only the optical geometry does so in the region $Q_{nc}<Q<Q_{m}^{ph}$. Non-radial geodesics are blocked from the singularity and can form metastable bound states, while radial geodesics either reach $r=0$ or escape to infinity. The critical impact parameter encodes both gravitational and optical-geometry effects, probing the distinct phase structures.