arXiv:2607.12001v1 Announce Type: new
Abstract: Simulating inviscid, incompressible fluids on non-simply-connected curved surfaces requires careful treatment of the flow's local and global behavior. While recent theoretical advancements have established the critical dynamics of the harmonic component in such flows, practical applications remain computationally restricted by a lack of spatial and temporal adaptivity. Furthermore, simulations on poor-quality meshes often lead to numerical instability and a failure to preserve the flow's underlying harmonic component when using naive interpolation methods. In this paper, we introduce Adaptive Fluid Cohomology, a framework that integrates dynamic spatial and temporal refinement into the simulation of the Euler equations. We leverage a posteriori error estimation to adjust spatial resolution on the fly, alongside a standard Dormand-Prince 5(4) time-stepping scheme for temporal accuracy. To ensure stability during mesh mutations, we develop a novel method that robustly transfers the harmonic basis during remeshing. While our experimental evaluation focuses on 2D surface flows, the underlying theoretical formulation is presented to capture the 3D setting as well. Our evaluation demonstrates that this adaptive approach accurately recreates the dynamics of high-resolution simulations while reducing the memory footprint by up to 86% and maintaining numerical stability even on poor-quality triangulations where static methods fail.
Science Journals
arXiv:2607.12051v1 Announce Type: new
Abstract: Large language models (LLMs) are increasingly being explored for clinical decision support, but their reliability in complex oncology treatment planning remains unclear. We evaluated agentic LLM systems for breast cancer treatment recommendation generation using 72 real clinical cases across stages I to IV and 1,147 case-specific rubrics generated through Asymmetric Information Rubric Generation (AIRG), in which the rubric generator had access to real clinical decisions unavailable to the evaluated models. Seven pipelines were compared, including single-LLM baselines, tool-augmented systems, and multi-agent architectures with fact checking and autonomous subagent spawning. The best-performing configuration, Claude Opus 4.8 with the D&C+SA pipeline, achieved a global score of 0.594 $\pm$ 0.025. Tool use and increased agent autonomy had mixed effects, improving performance in some settings but degrading it in others. Performance varied by clinical domain and disease stage, and oncologist-led error analysis revealed persistent clinically relevant failures, including incorrect or missing recommendations, flawed justifications, citation errors, outdated claims, and overconfidence. These findings suggest that agentic LLM systems can generate clinically relevant breast cancer recommendations, but remain insufficient for unsupervised clinical use.
arXiv:2511.11470v3 Announce Type: replace
Abstract: 3D urban generation from satellite imagery is an important task for scalable digital twins and real-world simulation environments. Existing approaches primarily rely on scene-level generation paradigms, which often require large-scale 3D city assets and struggle with controllability, geographic alignment, and realistic appearance grounding in real-world urban environments. To address these limitations, we present Sat2RealCity, a grounded urban generation framework that leverages object-level 3D generative priors for scalable city synthesis from satellite imagery. Our framework decomposes cities into geographically grounded building entities, enabling the reuse of pretrained object-level 3D generative priors while preserving real-world spatial structures. Supported by our constructed BuildVerse3D dataset, (1) we introduce an OpenStreetMap (OSM)-guided spatial grounding strategy to inject geospatial constraints into the 3D generation process; (2) we design an appearance-guided controllable generation mechanism for realistic architectural appearance and regional style consistency; and (3) we construct an MLLM-powered semantic pipeline for regional appearance understanding and semantic-aware appearance synthesis. Extensive experiments demonstrate that Sat2RealCity achieves strong geographic alignment, regional stylistic consistency, and plausible urban asset synthesis compared with existing urban generation and 3D asset generation approaches.
arXiv:2607.12765v1 Announce Type: new
Abstract: Quantum key distribution (QKD) offers unconditional security but existing QKD networks remain difficult to scale across heterogeneous infrastructures and administrative domains due to vendor-specific interfaces, trusted-node constraints, and limited interoperability. This work presents a flexible multi-domain and multi-site quantum-secure network architecture integrating vendor-agnostic QKD, SDN orchestration, and cloud-managed trust services. Communication is based on Zero Trust Network Access protocols featuring multi-level authentication mechanisms building upon post-quantum cryptography (PQC) signature and key encapsulation algorithms. The system is deployed on a real-world testbed with domains incorporating QKD nodes from 3 vendors, as well as domains without QKD infrastructure elements. Experimental results show that PQC and SDN overhead remain relatively low even on constrained devices, with the main bottleneck being QKD key retrieval and vendor-specific key streaming limitations. The proposed framework extends quantum-safe key transport beyond native QKD boundaries while preserving flexibility, interoperability, and compatibility with existing infrastructures.
arXiv:2607.12760v1 Announce Type: new
Abstract: Tree trunks contain living tissues whose functioning depends on their thermal environment, yet the processes governing trunk temperature under field conditions remain poorly understood. In particular, it is unclear whether hydraulic transport contributes to buffering trunk temperature against atmospheric variability. We investigated this question by continuously monitoring sapwood temperature, local air temperature, soil temperature at 1~m depth, and spontaneous electrical potential (SP) in six mature trees, comprising three oaks and three hornbeams, over multiple years in a temperate urban forest garden. trunk temperature exhibited a smoother seasonal cycle than air temperature and remained consistently closer to deep-soil temperature, indicating substantial thermal buffering. Seasonal components extracted from the time series revealed a coherent delayed relationship between SP and the trunk-soil temperature difference. Phase-space analysis showed reproducible hysteresis across individuals, and instantaneous phase estimates indicated a lag of approximately 100~days in five of the six trees. A minimal energy-balance model further showed that trunk heat storage alone was insufficient to reproduce the observed seasonal dynamics. Positive effective soil-coupling coefficients were obtained for most individuals, although their magnitude varied markedly among trees. These results are consistent with a contribution of vertically mediated, hydraulically linked heat transfer to seasonal trunk thermal regulation. They further suggest that spontaneous electrical potentials may provide a non-invasive, integrative indicator of slow hydraulic and thermal processes within trees. Such thermo-hydraulic coupling may influence the thermal environment of the cambium and phloem and could therefore contribute to tree responses to seasonal heat and climatic variability.
arXiv:2607.12117v1 Announce Type: new
Abstract: Published molecular docking scores depend on the receptor, ligand, software, search box, seed, and preparation choices; a paper reporting only the score has published a number with unknowable provenance. We ask whether such claims can be re-executed from their own published records. We introduce MERS-Dock, a 16-field Minimum Executable Reporting Set, and a deterministic E0-E4 executability ladder over audited field states. In 236 open-access SARS-CoV-2 main-protease docking papers, only 8.1% met the essential-field rule for direct re-execution (E3), 47.9% were blocked by a missing foundational field (E1), and none reached E4; mean field completeness was 49.1% and the search-box centre was reported by only 33.9%. We validated the audit against two independent human reviewers on a 65-paper stratified sample: inter-reviewer agreement was 92% (pooled Cohen kappa 0.87), and the automated agent matched humans on the execution-blocking fields while over-calling two non-blocking fields; the resulting E-class was 68% concordant with humans and, where it differed, human review lowered the executable count -- so the low-executability finding is confirmed, not inflated. Reporting did not improve over 2021-2026 (completeness vs year Spearman rho = -0.01). A bounded within-paper re-execution shows the reproduction gap is a box-coverage geometry effect, not box-size disclosure. We read E-class as an executability gate, not a reproducibility predictor, and release Mpro-DockExec as a traceable measurement layer for digital-library and evidence-synthesis systems deciding what is checkable in published computational claims.
arXiv:2607.12725v1 Announce Type: new
Abstract: Neural audio codecs were originally developed for high-fidelity compression; however, their latent token representations and expressive decoders also constitute a powerful substrate for controllable audio transformation. This work introduces Neural Morphing, a training-free token-domain audio effect that selects residual-vector-quantized (RVQ) token grains from a user palette and decodes the edited stream through a pretrained codec. The method combines an RVQ-group transfer policy that separates coarse, middle, and fine codebook groups with a continuity-constrained sequence matcher that replaces independent greedy selection with bounded beam search. The intended output is a controlled hybrid: the source preserves rhythmic organization while the palette contributes timbral color and residual detail. We focus on the implementation and realtime behavior of a deployable VST3/AU system, including chunked rendering, palette-size scaling, and backend health checks.
arXiv:2601.08010v3 Announce Type: replace
Abstract: Vision-language models achieve strong performance across a wide range of multimodal understanding and reasoning tasks, yet their multi-step reasoning remains unstable. Repeated sampling over the same input often produces divergent reasoning trajectories and inconsistent final predictions. To address this, we introduce two complementary approaches inspired by test-time scaling: (1) CASHEW, an inference-time framework that stabilizes reasoning by iteratively aggregating multiple candidate trajectories into higher-quality reasoning traces, with explicit visual verification filtering hallucinated steps and grounding reasoning in visual evidence, and (2) CASHEW-RL, a learned variant that internalizes this aggregation behavior within a single model. CASHEW-RL is trained using Group Sequence Policy Optimization (GSPO) with a composite reward that encourages correct answers grounded in minimal yet sufficient visual evidence, while adaptively allocating reasoning effort based on task difficulty. This training objective enables robust self-aggregation at inference. Extensive experiments on 13 image understanding, video understanding, and video reasoning benchmarks show significant performance improvements, including gains of up to +26.2 percentage points on ScienceQA and +9.1 percentage points on EgoSchema.
arXiv:2607.12736v1 Announce Type: new
Abstract: Scientific discovery increasingly depends on interdisciplinary teams whose members contribute distinct expertise, conceptual frameworks, vocabularies, assumptions, and standards of evidence. Today's AI research assistants are largely designed to support individual researchers through literature review, writing assistance, coding, and data analysis. While these capabilities improve personal productivity, they provide little support for the collaborative reasoning required to integrate knowledge across disciplines. We argue that AI research assistants should evolve from tools that optimize individual workflows to systems designed for interdisciplinary teams. We introduce a\"ira, an AI research assistant built around this idea. Rather than focusing solely on summarization or question answering, a\"ira identifies disciplinary perspectives, translates terminology, highlights assumptions, and synthesizes collaborative research opportunities. We describe the design principles underlying a\"ira, present its system architecture, illustrate its outputs through interdisciplinary research meetings, and outline future research directions for AI systems that support collaborative scholarship.
Integrated 3D fully kinetic simulation of field-reversed-configuration formation with embedded coils
arXiv:2607.11908v1 Announce Type: new
Abstract: We present an integrated, three-dimensional, fully kinetic particle-in-cell simulation of field-reversed-configuration (FRC) formation at the device scale. To our knowledge, this is the first fully kinetic model of whole-device FRC formation. The model embeds the drive coils directly inside the computational domain as physical conductors, advancing them self-consistently with the plasma on a single explicit grid and coupling them in closed loop to an external circuit. We apply this unified framework to the Yingguang-1 $\theta$-pinch. Unlike the magnetohydrodynamic and hybrid models used previously, our framework advances the electrons as kinetic particles rather than a fluid, capturing fast magnetic reconnection and electron heating from first principles. The simulation reproduces the complete formation sequence, from reversed-bias lock-in through reconnection to the emergence of a closed-flux FRC, reaching a peak ion density ${\sim}2.2\times10^{22}\,\mathrm{m^{-3}}$ consistent with experiment. The compressed core is electron-dominated, with $T_e\approx1.7\,$keV exceeding $T_i\approx1.2\,$keV, and is pinched to a separatrix radius $r_s\approx1\,$cm, several times below the equilibrium-inferred value, indicating that the plasma never relaxes to a pressure-balanced equilibrium within the microsecond pulse. The model further reproduces a non-axisymmetric, four-fold ($m=4$) deformation of the compressed column, matching the square cross-section recorded by the experiment's end-on framing camera, a feature beyond the reach of the two-dimensional models previously applied to this device. Running on modest GPU hardware, this work brings integrated, first-principles kinetic modeling of fusion-relevant FRCs within reach.
arXiv:2607.12742v1 Announce Type: new
Abstract: Encrypted control lets a cloud coordinate a fleet of agents on fully homomorphically encrypted state, keeping their positions and commands private. The approximate scheme for real-valued control, CKKS, returns decryptions that carry the encryption noise, a key-recovery leak; the loop must decrypt to actuate, so the leak is unavoidable. Yet the security of approximate FHE is studied statically, encrypted control assumes an honest-but-curious cloud, and persistent-threat games never reach inside the cryptosystem. We model the loop's security under an advanced persistent threat as a two-phase game, passive reconnaissance then active manipulation, separated by a measured residual detector that sees only the manipulation. The passive phase reduces to the known flooding tradeoff; the active defense is re-keying, not bootstrapping, since only re-keying resets accumulated leakage. The active phase is a detection-evasion timing game: overt manipulation is caught, so the rational adversary stays stealthy, and at its Stackelberg equilibrium the defender re-keys on the laziest cadence that denies it, set by the control-theoretic fragility of the graph topology. The marginally-stable graph must re-key far more often than the well-connected one. A three-way tension among FHE precision, control accuracy, and re-key cadence sets where this game lives, between a securability floor and a static-suffices ceiling. The efficient secure point is that window, where re-keying is the price of precision efficiency. More broadly, security for an approximate cryptosystem in a feedback loop is a dynamic game whose defender's move is the scheme's own refresh, applying beyond control to any system that must repeatedly decrypt to act.
arXiv:2604.04280v4 Announce Type: replace
Abstract: Maintaining situational awareness in high-stakes multi-robot applications requires balancing exploration of unobserved regions with sustained monitoring of changing Regions of Interest (ROIs), often under unknown and time-varying distributions, partial observability, and limited communication. We propose a decentralized multi-agent coverage framework that serves as a high-level planning strategy, in which each agent computes an adaptive ergodic policy, implemented via a Markov-chain, that tracks an updated belief over the underlying importance map. Beliefs are maintained online via Gaussian Process (GP) regression from local noisy observations exchanged with neighbors. The resulting policy drives agents to spend time in ROIs in proportion to their estimated importance, while preserving sufficient exploration to detect and adapt to time-varying environmental changes. Unlike existing approaches that assume known importance maps, centralized coordination, or a static environment, our framework addresses the combined challenges of unknown, time-varying distributions under a decentralized, partially observable setting. We further show that our framework is robust to communication and memory degradation, robot loss, and can scale up to hundreds of robots.
arXiv:2510.00506v2 Announce Type: replace
Abstract: How can we reconstruct 3D hand poses when large portions of the hand are heavily occluded by itself or by objects? Humans often resolve such ambiguities by leveraging contextual knowledge -- such as affordances, where an object's shape and function suggest how the object is typically grasped. Inspired by this observation, we propose a generative prior for hand pose refinement guided by affordance-aware textual descriptions of hand-object interactions (HOI). Our method employs a diffusion-based generative model that learns the distribution of plausible hand poses conditioned on affordance descriptions, which are inferred from a large vision-language model (VLM). This enables the refinement of occluded regions into more accurate and functionally coherent hand poses. Extensive experiments on HOGraspNet, a 3D hand-affordance dataset with severe occlusions, demonstrate that our affordance-guided refinement significantly improves hand pose estimation over both recent regression methods and diffusion-based refinement lacking contextual reasoning.
arXiv:2607.12616v1 Announce Type: new
Abstract: Continuous-time generative frameworks construct probability paths between base and target domains by optimizing time-dependent velocity fields. While theoretical targets favor straight trajectories, empirical networks develop complex path deformations. This paper presents the Finite-Time Spectral Sensitivity (FTSS) g(t), a gradient-free, forward-pass metric that exposes flow geometry by tracking the root-mean-square singular value of the state-transition matrix. Serving as a continuous proxy for stable rank, g(t) reveals a distinct geometric pathology under data scarcity: while generalizing models maintain stable effective dimensions, overfitting causes a spectral collapse. We leverage this structural phenomenon to develop an internal geometric audit based on g(t). Our framework detects generative memorization using purely internal trajectory dynamics, removing the need for external membership queries or baseline data comparison.
arXiv:2607.12203v1 Announce Type: new
Abstract: In this paper, we present EXP-SEC, a novel framework which can explain the intrusion detection decisions of DL-based NIDS (which lead to security alerts) in a way that is aligned with the domain knowledge of analysts working in Security Operations Center (SOC). We highlight the following features of our framework: (1) a forensic module that isolates the suspect packets/flow which likely caused an alert (2) an explanation module which can handle much more complex feature dependencies in network traffic than existing methods (features can be divided into overlapping groups and some groups are more important than others), and (3) a multi-stage mapping module which translates the feature/group-based explanations generated by explanation module to domain-specific explanations suitable for processing by security analysts. We evaluate EXP-SEC with state-of-the-art DL-based NIDS and our evaluation results show that EXP-SEC outperforms xNIDS (existing best performing explanation framework) in terms of group-level and overlap-aware explanation utility metrics while performing similarly in terms of conventional feature-level metrics such as descriptive accuracy, sparsity and stability. Moreover, taking the case of a state-of-the-art DL-based NIDS, we demonstrate the security analyst-friendly explanation format generated by EXP-SEC.
arXiv:2607.12206v1 Announce Type: new
Abstract: We present RegHead, a framework for constructing semantic blendshape sets for animatable non-humanoid head avatars. With a fixed expression vocabulary, semantic blendshapes provide a low-dimensional and interpretable animation interface and support cross-identity retargeting. Building such blendshape sets remains expensive because (i) expression-consistent supervision is scarce, (ii) generated 4D assets typically lack correspondence, and (iii) facial motion is highly localized. We propose (1) a large-scale dataset of non-humanoid identities paired with a shared expression vocabulary, obtained by expanding a small artist-rigged library via fine-tuned image editing; (2) a dense stochastic anchor motion representation tailored to localized facial deformations; and (3) a fast feed-forward registration model that converts unregistered expression meshes into a corresponded blendshape basis by predicting anchor-based deformations from the neutral shape. Experiments show that our approach produces higher-fidelity expression meshes than baselines, while running orders of magnitude faster than optimization. We further demonstrate real-time retargeting from human face tracking signals to non-humanoid characters, capturing both head pose and localized facial motions. Our project page is available at https://snap-research.github.io/RegHead/.
arXiv:2607.12216v1 Announce Type: new
Abstract: Multi-agent and memory-augmented LLM systems often place coordination content, shared state, prior discussion, tool outputs, summaries, and role instructions, inside the same finite prompt used for the current task. This creates a practical allocation problem: every token spent on coordination is unavailable to task instructions or evidence when a call is assembled under a fixed context budget. We introduce the Roundtable Context Window Test (RCWT), a controlled protocol for measuring this task-budget displacement effect. RCWT varies coordination content while controlling total budget, position order, task family, and scoring. In the main context-dependent recall task at $W=4096$, three commercial models remain near baseline through moderate overhead and then degrade sharply once residual reference evidence falls to a few hundred tokens. Window-scaling summaries are consistent with a task-specific residual-budget interpretation rather than a fixed percentage threshold, but we treat this as descriptive evidence rather than a universal law. To test whether the fixed-budget cliff persists when task evidence remains intact, we add an intact-task ablation: the full task/reference block is kept present while coordination tokens increase by expanding total prompt length. In that setting, all tested calls return every scored field correctly across GPT-4.1-mini, Claude Haiku 4.5, and Gemini 2.5 Flash up to a 95\% coordination ratio. This ablation narrows the claim: the main RCWT cliff is best read as task-budget displacement, not as proof that coordination volume alone causes semantic interference in the original open-ended task. RCWT is therefore a measurement primitive for context-allocation budgeting, not a complete theory of multi-agent benefit or session-level coordination.
arXiv:2607.12637v1 Announce Type: new
Abstract: Emerging knowledge can be envisioned as protruding magma. This paper describes how such a metaphoric view can be turned into a model to capture the 'continental drift' of concepts in an epistemic lithosphere. We call this new approach Knowledge Tectonics. We detail conceptual, mathematical and engineering operations to create such a scalable framework which allows us to interpret and manage knowledge evolution within Semantic Web environments. We use the WikiArt Emotions dataset which contains information on 4,105 paintings spanning 600 years to construct a proof--of--concept interface which enables visual analytics of where artworks are situated in a specific landscape of features. We demonstrate how fused semantic and pragmatic metadata can be modelled as evolving pressure zones. This way we are making 'forces' behind the evolution of artifacts of creativity visible. Core elements of our new workflow are gradient vector fields derived from Poisson potential surfaces applied onto dynamic knowledge graphs, eventually capturing stylistic and emotional shifts as directed intensity flows. By treating artefacts of creative processes as a dynamic manifold, we provide a novel methodology for quantifying the 'drift' of human inquiry. We argue that this approach is applicable also to other areas of creative human actions, including scientific knowledge production.
arXiv:2607.12645v1 Announce Type: new
Abstract: Generative modeling of longitudinal Electronic Health Records is increasingly important for privacy-preserving research, yet standard autoregressive models tend to underrepresent the co-occurrence structure of tail events (i.e., diseases, symptoms), reducing the fidelity and faithfulness of generated data for rare subpopulations. To this end, we propose AdaPCLA framework, which enables generative models to adaptively fit and generate EHR data through a data distribution-aware training strategy; this is achieved by internalizing data knowledge parameters by simulated annealing training. It also supports training-free adaptation to a diverse clinical population for generation through zero-shot distribution control. Moreover, our theoretical analysis characterizes rare-code logit updates through the label-wise empirical NTK and derives a prior-internalization bound for how annealing speed and NTK conditioning affect retained prior signals. Experiments on real-world data show that AdaPCLA achieves consistent gains in tail plausibility, downstream utility, and zero-shot control; in particular, it improves TailPairSeen over HALO by 114.2% on MIMIC-III and 65.1% on MIMIC-IV, outperforms GPT-style generation by 3.5% F1 for zero-shot cross-population adaptation.
arXiv:2607.12649v1 Announce Type: new
Abstract: Recent work on extractable memorization in LLMs suffers from two contrasting validity problems. Some studies overstate extraction, e.g., relying on sequences too short to distinguish memorization from predictability. Others imply that extraction is unreliable evidence of memorization, since models can also reproduce real-world text they weren't explicitly trained on. In different ways, both overlook what makes a valid extraction claim: the model must generate a training sequence with high enough probability to indicate memorization. To determine what's high enough, one has to perform a matched comparison: measuring the generation probabilities of both the training sequences of interest and comparable non-training sequences. Because non-training sequences cannot have been memorized, their probabilities provide a baseline for predictability; a training sequence exceeding this baseline provides evidence of memorization. We formalize matched comparisons in two ways: (1) a conformal test that calibrates a threshold to a chosen FPR when training and non-training sequences are sampled from populations, and (2) a census that calibrates against a matched non-training document when the object is a single document (e.g., a book). We show that matched comparisons enable rigorous, calibrated memorization claims, and reveal where prior setups have validity issues. For instance, on Wikipedia OLMo 2 32B reproduces non-training 10-token suffixes roughly 24% as often as training ones: that share of the training generation rate reflects false positives, not memorization. For Llama 3.1 70B on books, the thresholds we calibrate are as low as 1e-27, supporting memorization claims for sequences that no feasible sampling budget would extract. Based on these results, we refine "extractable memorization" to require a valid memorization claim and near-certain generation within a realistic budget.
arXiv:2607.12896v1 Announce Type: new
Abstract: Medical image segmentation foundation models are expected to generalize across diverse clinical scenarios, yet existing universal methods remain fragmented by prompt paradigms and spatial dimensions. Visual in-context learning, interactive segmentation, and language-guided segmentation are typically handled by paradigm-specific models, while 2D and 3D images are also modeled separately. Such isolation prevents heterogeneous annotations and data from being jointly absorbed by a single scalable model and limits cross-paradigm knowledge transfer. To address this bottleneck, we propose UniMedSeg, a Transformer-centric universal segmentation framework that maps visual examples, geometric interactions, language instructions, and 2D/3D images into a shared sequence space, enabling heterogeneous medical supervision to be jointly learned through a unified in-context interface without prompt- or dimension-specific branches. To overcome the long-sequence memory bottleneck caused by visual contexts, we introduce Decoupled Split Attention, which reduces attention complexity to linear while preserving hardware-friendly computation and focused context-target interaction. Extensively trained and evaluated on a large corpus curated from 27 public datasets, UniMedSeg achieves state-of-the-art performance across visual in-context, interactive, and language-guided segmentation without task-specific fine-tuning, demonstrating strong generalization on diverse held-out tasks. The code and model weights are publicly available at https://github.com/Lii1228/UniMedSeg
arXiv:2607.12875v1 Announce Type: new
Abstract: As LLM technology advances, the space of model families, compute hardware, quantization schemes, parallelization strategies, and specialized optimization kernels continues to expand, sharply increasing the code complexity and maintenance cost of general-purpose inference frameworks. Conventional software engineering uses multiple layers of abstraction to support diverse application scenarios, but these abstractions also increase system complexity and may introduce additional performance overhead. This paper presents metainfer, an 'LLM-as-Compiler' approach in which users specify only the runtime constraints of an inference program. An LLM-driven multi-agent collaboration system, coupled with a contract knowledge base, then automatically generates a compact customized inference framework that satisfies these constraints. We evaluate metainfer from three perspectives: the effect of source-code reference, the runtime behavior and performance profile of engines generated under the zero-reference constraint on CKB-covered targets, and knowledge-base evolution for new model and platform scenarios. The results show that metainfer organizes generation constraints, validation feedback, and knowledge consolidation into a continuous closed loop, enabling runnable customized inference solutions to be generated from explicit knowledge. The code is publicly available at https://github.com/MetaInfer/MetaInfer.
arXiv:2607.12650v1 Announce Type: new
Abstract: Tool access alone does not make LLM empirical reasoning governable: accepted outputs need not descend from attested evidence, and accepted deductions need not hold up under formal scrutiny. We present EG-VAR (Evidence-Grounded Verified Agentic Reasoning), a Lean 4-based tool-calling architecture in which the Lean kernel is the sole minter of Verified claims via tool-attestation axioms and declared source lifts. Every verified output structurally descends from an attested tool call (Thm. 3.1) and a kernel-checked chain of valid inference (Thm. 3.2); residual outputs are honest Abstain with a replayable audit trail. On a subcollection of TableBench numerical reasoning (n=120), EG-VAR attains 120/120 versus a 95% same-tool baseline; on counterfactual stress tests (5 domains x 2 models), EG-VAR stays 100% source-faithful while same-tool drops to 80-90% (no-tool 50-80%). With the LLM as deployment-time formalizer, residual semantic-formalization error is 3.3% on Sonnet and 1.7% on Opus. We position EG-VAR as a technical-governance interface for high-stakes empirical claims: a formal sidecar makes the target proposition, source scope, evidence boundary, proof obligation, and abstention condition auditable, eliminating unsupported Verified outputs today while turning formalization errors, lift and source-authority disputes, ambiguities, and abstentions into explicit audit targets. Over time, typed sidecars in datasets, APIs, public records, and AI-generated documents can amortize this formalization burden into reusable infrastructure.
arXiv:2607.12654v1 Announce Type: new
Abstract: This paper examines the foundational distinctions between proof theory and dependent type theory (DTT) in the design of interactive theorem provers. While several implemented systems are designed using the dependently typed {\lambda}-calculus to represent proofs, no major proof assistant is designed using modern structural proof theory, even though, as I will argue here, the sequent calculus offers a compelling alternative framework. Six specific topics are proposed where the proof-theoretic perspective is arguably superior to the DTT perspective. These topics include the separation of logic from proof structure, the strategic use of non-determinism in proof reconstruction, and the avoidance of complex typing-discipline issues such as universe levels and proof irrelevance. The final topic -- the treatment of bindings -- is further developed to demonstrate how a natural, intensional approach is achieved through the mobility of binders. This methodology is illustrated via the Abella theorem prover, which leverages lambda-tree syntax and the nabla-quantifier to provide an elegant environment for reasoning about the meta-theory of languages and logics involving complex binding.
arXiv:2607.12876v1 Announce Type: new
Abstract: Conventional statistical models often struggle to fully capture the complex spatio-temporal dynamics, intermittent fluctuations, and heavy-tailed distributions characteristic of real-world air pollution data. Furthermore, existing literature frequently focuses on extreme events, overlooking the persistence of low-pollution states and temporal memory effects. To address these gaps, we apply superstatistical frameworks from non-equilibrium statistical physics to analyse a comprehensive five-year dataset (2020-2025) of hourly air pollutant concentrations across the United Kingdom. Excellent fits of experimentally measured distributions are obtained from our theoretical models. We observe large heterogeneities of the best fitting parameters depending on the locations where the measurements are performed. These parameters form characteristic patterns in the 3-dimensional parameter space and depend on the type of pollutant considered, as well as on the environmental conditions (high traffic, industrial, or rural surroundings). We also investigate autocorrelation functions and provide evidence for differences in day-time and night-time decays of the autocorrelation function. Our investigation mainly focuses onto the dynamics of NO, NO2, PM2.5, PM10, but we also report on some anomalous distributions observed for O3.