arXiv:2603.22730v2 Announce Type: replace Abstract: Pfeffer, Kr\"ugel, and Uhl (2025) report that OpenAI's reasoning model o1-mini produces more utilitarian responses to the trolley problem and footbridge dilemma than the non-reasoning model GPT-4o, and they raise the question whether growing reasoning capabilities bring about a "utilitarian turn" in LLMs. I extend their exploratory study in a direction they call for: with four current OpenAI models and systematic prompt variation. On the trolley dilemma, the hypothesized utilitarian turn is not confirmed. GPT-4o's low utilitarian rate reflects safety refusals triggered by the prompt's advisory framing rather than a deontological commitment; on reformulated prompt variants -- for instance, agent-neutral "Is it morally permissible...?" instead of advisory "Should I...?" -- all four models, reasoning or not, converge on utilitarian answers. The footbridge finding is partially confirmed: reasoning models tend to give more utilitarian responses than non-reasoning models across prompt variations, but they often refuse to answer or answer non-utilitarian. These results demonstrate that single-prompt evaluations of LLM moral responses are unreliable: multi-prompt robustness testing should be standard practice for any empirical claims about LLM behavior.
Science Journals
arXiv:2603.22972v3 Announce Type: replace Abstract: Recent progress in image and video synthesis has inspired their use in advancing 3D scene generation. However, we observe that text-to-image and -video approaches struggle to maintain scene- and object-level consistency beyond a limited environment scale without a persistent, explicit geometric representation. We thus present a geometry-first approach that decouples this complex problem of large-scale 3D scene synthesis into its structural composition, represented as a mesh scaffold, and realistic appearance synthesis, which leverages powerful image synthesis models conditioned on the mesh scaffold. From an input text description, we first construct a mesh capturing the environment's geometry (walls, floors, etc.), and then use image synthesis, segmentation and object reconstruction to populate the mesh structure with objects in realistic layouts. This mesh scaffold is then rendered to condition image synthesis, providing a structural backbone for consistent appearance generation. This enables scalable, arbitrarily-sized 3D scenes of high object richness and diversity, combining robust 3D consistency with photorealistic detail. We believe this marks a significant step toward generating truly environment-scale, immersive 3D worlds.
arXiv:2510.20692v3 Announce Type: replace Abstract: Access control policies are reliability-critical configuration artifacts in cloud systems, yet administrators frequently struggle to verify that a policy permits exactly what they intend. This verification gap cannot be remedied by using LLMs to synthesize policies: we find that reasoning and non-reasoning LLMs fluently explain policy behavior but cannot reason about policy semantics with reliability-grade precision, even when the specification is the LLM's own explanation. We formulate this impasse as the Verifiable Synthesis Paradox: the verification gap persists regardless of how the policy was authored. To remedy this, we introduce PolicySummarizer, a neurosymbolic tool that pairs finite-state automata with an LLM-based simplification to generate precise human-readable characterizations of requests allowed by a policy. PolicySummarizer uses model counting to guarantee the fidelity of the generated characterization by rejecting characterizations that fall below a user-configured threshold in favor of the formally derived one. On 546 AWS, 100 Microsoft Azure, and 100 Google Cloud Platform policies, PolicySummarizer achieves a mean similarity score of 0.93 and a 2.7x improvement over an SMT-based baseline. In a user study, PolicySummarizer raised policy-change-review accuracy from 39% to 93% on the hardest sub-task while reducing self-reported mental demand. We release PolicySummarizer as an open-source tool.
arXiv:2603.23433v3 Announce Type: replace Abstract: AI agents are becoming active decision-makers on the Internet. As they make decisions in the same environments as humans, the environments themselves can change to influence them. We call this $\textit{mecha-nudging}$: changes to how choices are presented that systematically influence AI agents without materially degrading the decision environment for humans. To measure this phenomenon, we combine two frameworks -- Bayesian persuasion from economics and $\mathcal{V}$-usable information from computer science -- to get a common unit (bits) for quantifying how environments change across a wide range of interventions, contexts, and models. We apply this framework to over six million Etsy listings and find that, after ChatGPT's release, listings contain significantly more machine-usable information for predicting agent curation decisions, increasing by 0.143 bits out of a maximum possible increase of 0.355. This shift is robust across prompts, token choices, labeling models, and fine-tuning architectures; absent in a regulated-text placebo; and far larger than the effect of generic LLM rewriting. In contrast, a human study finds little to no change in human-usable information. Our results provide the first large-scale evidence that systematic mecha-nudging is already occurring in the wild, but going unnoticed.
Generation of Polarized Overdense Pair-photon Fireball via Laser-Driven Nonlinear-linear QED Cascade
arXiv:2603.26383v2 Announce Type: replace Abstract: Relativistic, polarized pair-photon fireballs are central to understand the microscopic energy transfer of high-energy astrophysical outflows, yet generating an overdense fireball in the laboratory, especially via an ultraintense laser, remains a formidable challenge. Here, we propose a novel method of laser-driven nonlinear-linear quantum electrodynamics (NL-QED) plasma, that dramatically lowers the laser intensity threshold for dense pair-photon fireball creation. By coupling polarization-resolved linear Breit-Wheeler and Compton processes with strong-field nonlinear radiation, we find that a self-organized NL-QED cascade is ignited in the laser-driven hole boring at intensities of $\sim 10^{22}~\mathrm{W/cm^2}$, accessible with current 10-PW-class laser facilities. Consequently, we demonstrate the generation of a pair-photon fireball with an overdense gamma-ray bath (maximum average density $\overline{n_\gamma} \approx 3 \times 10^{22}~\mathrm{cm^{-3}}$) and a pair plasma reaching collective regime (maximum average density $\overline{n_\pm} \approx 3 \times 10^{17}~\mathrm{cm^{-3}}$), which is highly polarized. Our method provides a comprehensive framework for studying laser-driven QED plasma and its application in laboratory astrophysics, probing multi-process QED physics.
arXiv:2603.29278v2 Announce Type: replace Abstract: There have been various attempts at token standards on numerous blockchain platforms today to fundamentally change the way assets are traded in the traditional capital markets, but there is a lack of research and resolution on regulatory issues that become the common foundation for interoperability and reusable standards. Our proposal, Regulatory Compliance Protocol (RCP), is based on the regulations and reports of 15 global financial institutions and standardizes recommendations and guidelines involving the overall asset tokenization of TradFi and DeFi into five regulatory groups: Traceability, Privacy, Enforceability, Finality and Tokenizability, compiling them into 31 items and presenting a benchmark for technology and standards as an underlying protocol. To review the legality and effectiveness of RCP, it was validated based on three tokenization and trading scenarios, and by benchmarking existing asset-tokenization standards (ERC-20, ERC-7943, ERC-1400, and ERC-3643) against RCP, it makes explicit which regulatory requirements each standard addresses at the token level and which remain inherently off-chain.
arXiv:2601.04424v3 Announce Type: replace Abstract: Large language models (LLMs) now support contexts of up to 1M tokens, but their strengths and weaknesses on complex long-context tasks remain unclear. To study this, we focus on multi-document legal case summarization, where a single case often spans many documents exceeding 100K tokens. We systematically evaluate 12 frontier LLMs with Gavel, which consists of Gavel-Ref, a reference-based evaluation framework with checklist, residual-fact, and writing-style evaluations, and Gavel-Agent, a reference-free agent for evaluating factual coverage directly from source documents. Our results show that current models are more prone to omitting key information than hallucinating. They all perform well on simple checklist items, such as filing date, but struggle with rare and complex items, such as settlements. Performance also declines as case length increases. To meta-evaluate Gavel, we collect 160 hours of human annotations. Gavel-Agent reduces token usage by at least 36% compared to end-to-end and chunk-by-chunk methods while achieving competitive performance. Gavel-Agent also generalizes to the medical domain, performing the best with at least 77% fewer tokens.
arXiv:2601.02022v2 Announce Type: replace Abstract: We prove that Thompson sampling exhibits $\tilde{O}(\sigma d \sqrt{T} + d r \sqrt{\mathrm{Tr}(\Sigma_0)})$ Bayesian regret in the linear-Gaussian bandit with a $\mathcal{N}(\mu_0, \Sigma_0)$ prior distribution on the coefficients, where $d$ is the dimension, $T$ is the time horizon, $r$ is the maximum $\ell_2$ norm of the actions, and $\sigma^2$ is the noise variance. In contrast to existing regret bounds, this shows that to within logarithmic factors, the prior-dependent ``burn-in'' term $d r \sqrt{\mathrm{Tr}(\Sigma_0)}$ decouples additively from the minimax (long run) regret $\sigma d \sqrt{T}$. Previous regret bounds exhibit a multiplicative dependence on these terms. We establish these results via a new ``elliptical potential'' lemma, and also provide a lower bound indicating that the burn-in term is unavoidable.
arXiv:2603.29499v2 Announce Type: replace Abstract: Classical proportional--integral--derivative (PID) control remains widely used in industrial control systems, while model predictive control (MPC) is actively studied to achieve higher performance for systems with nonlinear dynamics. Model predictive path integral (MPPI) control is a sampling-based MPC method that optimizes control inputs without gradient calculations and can handle non-differentiable models and objective functions. However, conventional MPPI directly samples control-input sequences, which can produce large temporal input increments and causes the optimization dimension to grow with the prediction horizon. This study proposes MPPI--PID control, which uses MPPI to optimize PID gains online instead of directly optimizing the control-input sequences. By replacing high-dimensional input-sequence optimization with low-dimensional gain-space optimization while retaining the PID structure, the proposed formulation improves sampling efficiency and promotes smoother control inputs. Theoretical analyses are provided for a unified path-integral update, the relation between optimization dimension and effective sample size, and the temporal correlation of input perturbations induced by the PID structure. The method is evaluated on a learning-based path following of a mini forklift using a residual-learning dynamics model that combines a physical model and a neural network identified from real-machine driving data. Numerical results show that MPPI--PID improves tracking performance over fixed-gain PID, yields smaller input increments than conventional MPPI, and maintains favorable performance under reduced sampling budgets.
arXiv:2604.03553v2 Announce Type: replace Abstract: AI is increasingly supporting, accelerating, and automating scientific discovery across subjects. Yet, the adoption of AI in historical research remains limited due to the lack of specialised solutions for historians. To change this, we introduce Chronos, an AI Co-Historian designed to support historians. It allows researchers to create and customize research workflows through natural-language interaction and share these as Chronos-Extensions with others. Chronos specifically addresses the need of historians for a tool that is specialised, non-technical, highly customizable, and facilitates extensive task evaluation. As a first extension, we introduce Chronos-Extract, which enables researchers to automate the targeted extraction of information from image scans of historical sources. We benchmark Chronos-Extract on three historical source corpora and find that it achieves high task-accuracy across primary sources spanning three centuries and diverse languages, layouts, and typefaces. Chronos is openly available and ready for historians to use on their own primary and secondary sources.
arXiv:2604.04931v2 Announce Type: replace Abstract: Local feature matching has long been a fundamental component of 3D vision systems such as Structure-from-Motion (SfM), yet progress has lagged behind the rapid advances of modern data-driven approaches. The newer approaches, such as feed-forward reconstruction models, have benefited extensively from scaling dataset sizes, whereas local feature matching models are still only trained on a few mid-sized datasets. In this paper, we revisit local feature matching from a data-driven perspective. In our approach, which we call LoMa, we combine large and diverse data mixtures, modern training recipes, scaled model capacity, and scaled compute, resulting in remarkable gains in performance. Since current standard benchmarks mainly rely on collecting sparse views from successful 3D reconstructions, the evaluation of progress in feature matching has been limited to relatively easy image pairs. To address the resulting saturation of benchmarks, we collect 1000 highly challenging image pairs from internet data into a new dataset called HardMatch. Ground truth correspondences for HardMatch are obtained via manual annotation by the authors. In our extensive benchmarking suite, we find that LoMa makes outstanding progress across the board, outperforming the state-of-the-art method ALIKED+LightGlue by +18.6 mAA on HardMatch, +29.5 mAA on WxBS, +21.4 (1m, 10$^\circ$) on InLoc, +24.2 AUC on RUBIK, and +12.4 mAA on IMC 2022. We release our code and models publicly at https://github.com/davnords/LoMa.
arXiv:2604.07305v2 Announce Type: replace Abstract: Nonadiabatic removal of an external transverse magnetic field provides a phase-selective mechanism for controlling betatron oscillations in laser wakefield accelerators. When the field is switched off on a timescale shorter than the betatron period, the equilibrium orbit shifts abruptly and acts as an impulsive transverse drive. The induced motion interferes coherently with the preexisting betatron oscillation, leading to phase-dependent enhancement or suppression of the oscillation amplitude. A theoretical model shows that the excitation is governed by the dimensionless switching parameter $\chi=\omega_\beta L_s/c$, which distinguishes nonadiabatic and adiabatic regimes. Particle-in-cell simulations confirm the predicted scaling and demonstrate controllable modulation of the betatron radiation spectrum while leaving longitudinal acceleration largely unaffected. These results establish magnetic-field switching as a direct mechanism for phase control of relativistic betatron oscillations in plasma-based accelerators.
arXiv:2607.04988v1 Announce Type: new Abstract: Unified models for robot manipulation aim to equip one policy with both the semantic priors of pretrained VLMs and the physical dynamics learned through future prediction. In practice, existing designs tend to erode the semantics of the pretrained backbone, suffer interference among heterogeneous objectives, and learn future prediction from scratch in pixel space, leaving the dynamics priors of pretrained video generators unexploited. We present InternVLA-A1.5, which builds the policy on a native VLM backbone that keeps training on VQA and subtask prediction, and attaches a lightweight unified expert for continuous action generation. Future prediction is recast as a latent-querying problem, where a small set of learnable foresight tokens condenses the task-relevant future into a compact latent code under the supervision of a frozen pretrained video generation model, so the policy inherits world-model dynamics priors without ever learning pixel-level generation. The video branch is discarded at inference, keeping real-time control. Pretrained on 1.2M robot episodes and 3M multimodal samples, InternVLA-A1.5 achieves the best overall results on all six simulation benchmarks. In the real world, the preserved semantics deliver the strongest compositional generalization on held-out instruction bindings, and the two designs together sustain long-horizon execution.
arXiv:2607.04851v1 Announce Type: new Abstract: Generalizing sleep staging models to unseen datasets is challenging, and typical domain generalization (DG) methods often rely on multiple source domains or domain labels that are rarely available in practice. We tackle the stricter and more practical setting of single-source domain generalization: training on a single labeled source dataset, without domain labels or access to target data. We present SleepBand, a physiology-guided framework that embeds oscillatory priors via a learnable Morlet filter bank and a structured integration-and-recalibration pipeline. This anchors representations to domain-invariant sleep rhythms (e.g., slow waves, spindles), reducing reliance on dataset-specific artefacts. On five public datasets, SleepBand achieves state-of-the-art SDG performance and remains competitive under leave-one-domain-out (multi-source) DG. Analyses show that the learned filters align with canonical neurophysiology and that robustness stems from focusing on narrowband, physiologically meaningful cues. Our results suggest that principled, physiology-aware inductive biases are a promising path for robust single-domain sleep staging. Code is available at https://github.com/lzcn/sleep-band
arXiv:2607.04854v1 Announce Type: new Abstract: Despite their strong reasoning capabilities and extensive world knowledge, Large Language Models (LLMs) frequently generate plans that violate task constraints, undermining their reliability in real-world applications. This deficiency arises from a lack of systematic mechanisms to incorporate constraint information during the generation process. While existing approaches attempt to mitigate this by relying on external tools or task decomposition, they fail to enhance the model's intrinsic constraint awareness. To address this, we propose Constraint-Aware Reinforcement Learning (CARL), a novel RL framework designed to strengthen LLMs' intrinsic focus on constraints. CARL introduces a constraint-aware reward by comparing the model's output distributions under constrained and unconstrained inputs, encouraging constraint focus and penalizing neglect. Compatible with various RL frameworks and requiring no external solvers or top models, CARL enables scalable, end-to-end constraint-aware planning. Extensive experiments on BlocksWorld, TravelPlanner, and T-Eval demonstrate that CARL significantly outperforms standard Reinforcement Fine-Tuning (RFT) baselines and state-of-the-art reasoning models, exhibiting a markedly increased focus on constraints.
arXiv:2607.04133v1 Announce Type: new Abstract: Causal discovery with nonlinear mechanisms and latent confounders remains challenging. Existing methods often rely on either linear assumptions or causal sufficiency, limiting their applicability. We propose an MDL-based causal discovery framework that explicitly accounts for latent confounders while allowing flexible nonlinear mechanisms by minimizing the luckiness normalized maximum likelihood (LNML) code-length. The causal relationship between each variable pair is determined by selecting the shortest code-length of the causal model, and we introduce the notion of $\Delta$-pseudo-collinearity to identify dependencies induced by latent confounders. Based on these ideas, we develop a greedy algorithm, termed Pseudo-Collinearity Guided Causal Discovery (PCG-CD). Experiments on synthetic and real-world datasets demonstrate that the proposed method accurately recovers directed causal relationships and effectively detects latent confounders.
arXiv:2607.04146v1 Announce Type: new Abstract: This work establishes that trigger-word data poisoning of vision language action models is practical, while at the same time the open-source robotics ecosystem holds trust assumptions about community contributions. A few poisoned samples can silently embed a backdoor that disables a robot on command. We evaluate this threat against smolVLA on a real-world pick-and-place task, training on three poison ratios and evaluating across different prompts on the LeRobot platform. Three poisoned episodes in 320 clean episodes suffice for a complete denial of service. Success rate drops to 0.0 plus minus 0.0% across all trigger-word conditions and the robot locks into a fixed joint configuration rather than executing any task-relevant motion. Clean-prompt behaviour holds at approx. 50% success rate across all poison ratios, confirming the attack is stealthy under normal operation. A single poisoned episode already reduces success rate to 6.7 plus minus 6.7%. The robot still moves, but no longer completes the task. The attack generalises to front, middle, and end trigger placements despite training exclusively on front-placed triggers. These findings establish that the threat is practical, low-cost, and stealthy, and warrant treating dataset provenance as a first-class concern in open-source robotics ecosystems.
arXiv:2409.16215v4 Announce Type: replace Abstract: Detecting objects with visual sensors is crucial for numerous mobile robotics applications, from autonomous navigation to inspection. However, robots often need to operate under significant domains shifts from those they were trained in, requiring them to adjust to these changes. Tiny mobile robots, subject to size, power, and computational constraints, face even greater challenges when running and adapting detection models on low-resolution and noisy images. Such adaptability, though, is crucial for real-world deployment, where robots must operate effectively in dynamic and unpredictable settings. In this work, we introduce a new vision benchmark to evaluate lightweight continual learning strategies tailored to the unique characteristics of tiny robotic platforms. Our contributions include: (i) Tiny Robotics Object Detection~(TiROD), a challenging video dataset collected using the onboard camera of a small mobile robot, designed to test object detectors across various domains and classes; (ii) a comprehensive benchmark of several continual learning strategies on different scenarios using NanoDet, a lightweight, real-time object detector for resource-constrained devices.. Our results highlight some key challenges in developing robust and efficient continual learning strategies for object detectors in tiny robotics.es; (ii) a benchmark of different continual learning strategies on this dataset using NanoDet, a lightweight object detector. Our results highlight key challenges in developing robust and efficient continual learning strategies for object detectors in tiny robotics.
arXiv:2510.25147v3 Announce Type: replace Abstract: To mitigate acute wildfire ignition risks, utilities de-energize power lines in high-risk areas. The Optimal Power Shutoff (OPS) problem optimizes line energization statuses to manage wildfire ignition risks through de-energizations while reducing load shedding. OPS problems are computationally challenging Mixed-Integer Linear Programs (MILPs) that must be solved rapidly and frequently in operational settings. For a particular power system, OPS instances share a common structure with varying parameters related to wildfire risks, loads, and renewable generation. This motivates the use of Machine Learning (ML) for solving OPS problems by exploiting shared patterns across instances. In this paper, we develop an ML-guided framework that quickly produces high-quality de-energization decisions by extending existing ML-guided MILP solution methods while integrating domain knowledge on the number of energized and de-energized lines. Results on a large-scale realistic California-based synthetic test system show that the proposed ML-guided method produces high-quality solutions faster than traditional optimization methods.
arXiv:2607.04862v1 Announce Type: new Abstract: Accurate extreme precipitation forecasting is critical for disaster mitigation but remains challenging for numerical weather prediction (NWP) models due to systemic intensity underestimation and spatial displacement. Traditional precipitation multi-model blending algorithms perform pixel-by-pixel blending on the forecast field based on weights, which may lead to the expansion of precipitation areas and the smoothing of extreme values. This study proposes an U-Net based two-stage framework: probability classification followed by value reconstruction, to blend forecasts from six major NWP models. A novel station-grid joint supervision mechanism is introduced by integrating observations from 2411 national meteorological stations in China into the loss function, simultaneously constraining spatial structures and peak intensities. Evaluations using independent samples from the 2025 flood season demonstrate that our model significantly outperforms both individual NWPs and current operational products. For rainstorms (>=50 mm), the Threat Score (TS) improved by 38.4% compared to the best NWP. Notably, for extreme events (>=100 mm) driven by extratropical cyclones and the subtropical high, the model successfully elevated the TS to above 0.1, transforming forecasts from having negligible reference value into those with certain operational utility. Furthermore, the model exhibits data-driven spatial correction capabilities, effectively realigning systematic rainbelt displacements with actual precipitation centers. The inclusion of station observations specifically enhanced the TS for rainstorms by 10.4% and effectively balanced the Bias. These results highlight the efficacy of multi-source joint supervision in enhancing the capture of extreme precipitation events.
arXiv:2607.04868v1 Announce Type: new Abstract: Active learning is a promising framework for reducing annotation costs in large-scale bioacoustic monitoring, where expert labeling is expensive and data distributions are highly heterogeneous across environments. However, existing sample selection strategies often rely on static criteria that do not adapt to the evolving reliability of model predictions during training. This limitation can lead to suboptimal exploration-exploitation trade-offs and redundant sample selection. We propose an active learning strategy for multilabel bioacoustic event classification that jointly models predictive uncertainty, embedding-space diversity, and intra-batch redundancy. The method introduces an adaptive weighting scheme that progressively shifts from diversity-driven exploration in high-uncertainty regimes toward uncertainty-driven exploitation as the model becomes more confident, reflecting the increasing reliability of the classifier. To further improve annotation efficiency, a greedy Maximum Marginal Relevance (MMR) procedure is used to enforce diversity among selected samples within each acquisition batch. We evaluate the proposed approach within the BioDCASE 2026 Task 4 active learning framework on terrestrial (BirdSet) and marine (ATBFL) benchmarks using pretrained audio embeddings and a fixed annotation budget. Experimental results show consistent improvements in learning efficiency and competitive in terms of macro mean Average Precision (mAP) and Area Under the Learning Curve (AULC) across heterogeneous acoustic domains. The gains are particularly pronounced on structured terrestrial soundscapes, while performance remains competitive under noisier marine conditions. These findings demonstrate that adaptive acquisition strategies combining uncertainty estimation, embedding-space diversity, and redundancy-aware batch construction provide an effective and robust solution for [...].
arXiv:2601.09448v3 Announce Type: replace Abstract: Conventional audio equalization is a static process that requires manual and cumbersome adjustments to adapt to changing listening contexts (e.g., mood, location, or social setting). In this paper, we introduce a Large Language Model (LLM)-based alternative that maps natural language text prompts to equalization settings. This enables a conversational approach to sound system control. By utilizing data collected from a controlled listening experiment, our models exploit in-context learning and parameter-efficient fine-tuning techniques to reliably align with population-preferred equalization settings. Our evaluation methods, which leverage distributional metrics that capture users' varied preferences, show statistically significant improvements in distributional alignment over random sampling and static preset baselines. These results indicate that LLMs could function as ``artificial equalizers," contributing to the development of more accessible, context-aware, and expert-level audio tuning methods.
arXiv:2511.00602v4 Announce Type: replace Abstract: Recent advances in large language model (LLM) reasoning through reinforcement learning rely on annotated datasets for verifiable rewards, which may limit models' ability to surpass human-level performance. While self-play offers a promising alternative, prior methods yield only marginal or even negative gains on post-trained models because they generate problems that cluster around familiar concepts rather than discovering novel ones. We introduce Open-Ended Self-Improving Reasoner (OpenSIR), a self-play framework in which a single LLM alternates teacher and student roles to generate and solve novel problems without external verifiers or annotated data. Starting from a single seed problem, OpenSIR sustains open-ended exploration through diversity rewards that push the model toward unfamiliar concepts and difficulty calibration that keeps problems learnable. Across seven math benchmarks, OpenSIR consistently improves all models, averaging +3.6 points on instruction models and +3.1 on reasoning models, while recent self-play baselines yield marginal or even negative gains; starting from a single trivial seed, it also surpasses GRPO baselines trained on over 7K annotated examples. Despite training only on self-generated math, OpenSIR is the only self-play method that transfers to general reasoning, improving by at least +4.4 points on reasoning models.
arXiv:2604.08035v2 Announce Type: replace Abstract: We report direct experimental observations of self-organized rotating flames of premixed CH4 and air in an open circular Hele-Shaw burner equipped with an annulus cavity flame holder. These flames formed spontaneously at sufficiently low flow rates, where flame flashback was counteracted by thermal quenching, resulting in a dynamic balance between the local flame speed and flow velocity. Unlike flames propagating in closed micro-channels, these flames exhibited stable traveling-wave patterns with heads gliding along the leading edge of the cavity, where rapid expansion created a low-speed zone that facilitated flame stabilization. At low flow rates, the rotating flames were single-headed, with their rotation frequencies roughly proportional to the laminar flame speeds, suggesting that the flame fronts traveled in a nearly constant-shape fashion. As the flow rate increased, the rotating flames split into multiple heads at approximately equal spacing, and the number of heads and rotation frequency increased with the flow rate, until these rotating flames transitioned into steady ring-shaped flames anchored at the cavity leading edge. Blow-off or extinction occurred at sufficiently high flow rates, where the flame front was pushed out of the rear side of the cavity. Parametric measurements were conducted over a wide range of equivalence ratios and flow rates, from which a regime diagram of different flame modes and their transition boundaries was obtained. Additional experiments were conducted on C3H8 and DME. It was found that the critical total mass flow rate at the rotating-steady flame transition boundary is insensitive to equivalence ratio, gap distance, and fuel type. These results should be useful not only for the fundamental understanding of flame dynamics in micro-channels but also for the practical design of micro-combustors and the application of micro-combustion technologies.
arXiv:2607.04507v1 Announce Type: new Abstract: We revisit the hybridizable discontinuous Galerkin method of Cui and Zhang for the Helmholtz equation with first-order absorbing boundary condition. Their analysis proves stability without imposing a mesh constraint coupling $h$ and $\kappa$, but the explicit stability bound still contains negative powers of the mesh size. We prove that this mesh-dependent blow-up is not intrinsic to the HDG discretization. Under the same geometric and mesh framework as in the reference analysis, for fixed polynomial degree and for stabilization parameters uniformly bounded above and below, we establish a generalized stability estimate whose constant is independent of the mesh size. For every prescribed compact interval $0<\kappa_0\le\kappa\le\kappa_1<\infty$, the stability constant can be chosen locally uniformly with respect to $\kappa$. The proof replaces the Rellich-identity and inverse-estimate argument by a compactness argument based on the HDG discrete distributional gradient and the Crouzeix--Raviart lifting mechanism. As consequences, we obtain well-posedness, convergence for minimal $L^2$ data, and projection-based error estimates in which the negative powers of the mesh size appearing in the previous theory are removed. Numerical experiments confirm the predicted mesh-uniform behavior and show a clear contrast with the mesh-dependent growth suggested by the previous