Papers

Fine-tuning Vision-Language-Action (VLA) models commonly relies on human teleoperation demonstrations, while reinforcement learning (RL) with sparse binary rewards faces an exploration challenge when successful trajectories are rarely sampled. We propose SynthDemo-RL, a teacher-student framework in which an automated teacher converts simulator-privileged state into successful manipulation trajectories, a VLA student is distilled from them by supervised fine-tuning (SFT), and PPO with binary task-success rewards refines the student. We study reward coverage, the fraction of tasks for which at least one success is observed under the fixed evaluation protocol, as a complement to the average success rate. On LIBERO-PRO, a public benchmark of perturbed LIBERO tasks for which no demonstrations exist, 27 of 57 scored tasks are at exactly 0% success for a pi_0.5 policy fine-tuned on the original LIBERO tasks. Direct PPO from this policy, under the same PPO recipe and the same RL compute as SynthDemo-RL's refinement stage, rescues 10 of these 27 tasks and leaves 17 at 0%. SynthDemo-RL, with 50 synthesized trajectories per task and no new human demonstrations, rescues all 27 and reaches average success rates of 97.8% and 97.1% on the Position and Task axes of LIBERO-PRO, respectively. On standard LIBERO, the same pipeline reaches 96.0% with no human demonstrations, within 1.7 points of pi_0.5 trained on 50 human demonstrations per task. We further validate the pipeline on RoboTwin 2.0 and verify that trajectories from a policy trained in a MuJoCo twin execute open-loop on a physical robot.

Hamiltonian normalizing flows are attractive generative models because their phase-space maps are invertible and volume preserving, but most neural constructions are formulated in Euclidean space. We introduce Riemannian Neural Hamiltonian Flows, which combine the fixed kinetic energy of a Riemannian manifold, a learned scalar potential, and an explicit geodesic integrator. We then study the interpretability of the learned potential through a thermodynamic reading in which the momentum variance plays the role of a temperature. The matched potential, whose Boltzmann distribution is the data, is the potential whose force balances the pressure of the data. The output of one pass of the flow is the Boltzmann distribution of the learned potential multiplied by a memory factor, so that the learned potential is the matched potential whenever the pass forgets its starting point. Numerical experiments on Euclidean, hyperbolic, and spherical spaces show competitive sample quality and numerical cost against a Riemannian continuous normalizing flow, and confirm the interpretability of the learned potential. We then show that this reading has a computational use. When the energy of the data is known but expensive, the landscape learned from existing samples is a surrogate energy for Hamiltonian Monte Carlo, allowing exact sampling with significant acceleration. We test this approach on the localization of a neutrino source from public IceCube data.

Automated features and advanced driving support systems have enabled a range of life-saving and comfort applications in modern personal vehicles. Development from mostly mechanical entities towards truly software-defined devices also means that added sensor complexity, processing power, and networking capabilities provide a platform for running even more complex algorithms and applications. As vehicles become smarter, they require even more energy to enable advanced machine learning- and artificial intelligence-based functionalities. This paper discusses the implications for sustainability and the threats to environmental ethics posed by the development of software-defined vehicles. We discuss the balance between introducing more resource-hungry software and hardware components and their effect on society and sustainability. We underscore the need for novel tools to measure, manage, and optimise energy consumption and other sustainability indicators, with design guidance and actionable recommendations.

Debate adjudication requires tracking how arguments develop through interaction, yet existing datasets rarely combine fine-grained debate transcripts with professional judgments collected during real competitions under a shared rubric. We introduce a dataset and benchmark for evaluating large language models' understanding of competitive Chinese-language debate at the match, stage, and speaker levels. We organized 182 matches and recruited 120 professional judges, with each match independently adjudicated by three judges using a predefined rubric. After excluding matches with incomplete records, the dataset contains 148 matches, 2,698 stages, and 20,542 exchange units, with manually verified transcripts and segmentation. It preserves original stage scores, match votes, best-debater ballots, and adjudication rationales. We define three tasks: winner-tendency prediction, stage-score prediction, and best-debater prediction. Zero-shot evaluation of multiple large language models yields a highest winner-prediction accuracy of 66.2%, a highest Pearson correlation of 0.250 between model stage scores and mean human ratings, and a highest best-debater prediction accuracy of 56.8%. The dataset and benchmark provide a testbed for studying large language models' understanding of interactive argumentation and their agreement with professional judges.

Recent work applies human psychometric questionnaires to large language models to elicit moral and value profiles, but it is not clear whether these instruments measure anything stable in models or whether the resulting profiles can be moved toward a target human population. We administer the Norwegian Moral Foundations Questionnaire (MFQ-30) to six open-weight LLMs and compare their foundation profiles to a sample of N = 1,282 Norwegian respondents. We test two steering interventions, prompt-level persona steering and activation-level ActAdd. Half the models engage with the questionnaire under our attention check. The other half default to flat or central-tendency outputs that look near-human on average without tracking item content. A neutral Nordic-respondent persona, written without any distributional information from the human sample, brings the engaging models 44-77% closer to the Norwegian mean in Mahalanobis $d^2$. One-pair ActAdd at a fixed mid-layer flattens the foundation profile rather than steering individual foundations. For at least one model the same persona that shifts the profile also induces engagement that was absent at baseline, a concrete instance of the cognitive phantoms that Peereboom et al. (2025) warn about.

We present TERRA-NG, a portable, GPU-accelerated, matrix-free mantle-convection code. A single Kokkos C++ implementation runs at scale on NVIDIA, AMD, and Intel GPU supercomputers. TERRA-NG has a deliberately narrow design: built on a radially extruded mesh of spherical wedges, tailored to the spherical shell geometry, which enables domain-specific optimizations like single quadrature-point integral-evaluations, radial coordinate storage compression and radial shared-memory tiling. The corresponding low-order $W_1$-iso-$W_2/W_1$ wedge-based Stokes–energy discretisation is verified against the Zhong et al.(2008) spherical-shell convection benchmark suite. We showcase TERRA-NG through strong- and weak-scaling on the JUWELS Booster (NVIDIA A100), MareNostrum 5 (NVIDIA H100), LUMI-G (AMD MI250X), Hunter (AMD MI300A APU), and SuperMUC-NG Phase 2 (Intel PVC) supercomputers. Coupled mantle convection simulations at $\sim\!11$ km and $\sim\!5.6$ km radial spacing ($\sim 2.8$ B and $\sim 22$ B DoFs) can be run routinely on standard node partitions of all considered systems. Global $\sim\!1$ km-per-gridpoint mantle convection ($\sim 1.4$ T DoFs) is feasible on an extreme-scale allocation, and a sub-km hero-run at $\sim\!0.7$ km grid spacing scaling up to $\sim 11,000$ GPUs of LUMI-G ($\sim 11$ T DoFs) shows the potential of the code on future, larger machines.

The rapid development of artificial intelligence is often discussed primarily as a technological breakthrough. Such an approach is insufficient because AI is also transforming the allocation of capital, energy, natural resources, labour, and political power. Investment in generative AI and computing infrastructure is increasing rapidly and is increasingly concentrated among a small number of corporations and countries. At the same time, the expansion of data centres creates new electricity and water demands, potentially producing local infrastructure bottlenecks and distributive conflicts. The AI value chain is dependent on geographically concentrated supplies of advanced semiconductors, manufacturing equipment, cloud services, energy, and specialised knowledge. Control over these bottlenecks may give states and corporations structural power that does not derive directly from military superiority. The environmental effects of AI are similarly ambivalent: AI systems consume energy, water, materials, and computing hardware, but may also improve energy efficiency, climate modelling, renewable-energy integration, and environmental monitoring. Finally, AI-enabled automation, including intelligent robotics, may reduce employment in some occupations while augmenting labour and creating new tasks in others. The social outcome of this transformation will therefore depend not only on the technology itself but also on competition policy, infrastructure planning, environmental regulation, social protection, collective bargaining, and the distribution of productivity gains.

Multi-Channel imaging (MCI) data differs fundamentally from natural images, as each channel records a semantically distinct signal rather than a colour band. To adapt vision encoders to MCI data, Multi-Channel Vision Transformers (MC-ViTs) tokenize each channel independently and concatenate the resulting tokens into one sequence, and the channel count is no longer fixed by the architecture. Self-attention is then computed across all channel-patch tokens with no restriction on which channels attend to which, which dilutes the features of individual channels. The Decoupled Vision Transformer (DC-ViT) regulates this by separating updates computed within a channel from updates computed across channels, and by forming a representation per channel before the channels are combined. Its formulation, however, pairs tokens by spatial position, and thus requires the same visible tokens in every channel. Correspondence under independent per-channel masking is recovered by solving a linear assignment between the retained patches of each channel, which allows decoupled attention to be combined with current masked multi-channel training in its standard configuration rather than a restricted one. Across three classification and three segmentation benchmarks spanning fluorescence microscopy, imaging mass cytometry and satellite imaging, including dense prediction at high channel counts, the resulting formulation outperforms the strongest MC-ViT baseline.

Automatic tooth segmentation and FDI numbering on panoramic radiographs underpins computer-assisted dental diagnosis, yet which factors govern performance remains unclear. We assemble a corpus of 1,422 panoramic radiographs containing 42,142 expert-delineated tooth polygons across the 32-class FDI taxonomy, annotated by 30 dental practitioners and independently reviewed by two others, and use it to isolate input resolution, architecture and anatomical priors under a single evaluation protocol. First, resolution dominates: across a controlled 640/1024/1280 ablation, mask mAP50-95 rises 0.656 -> 0.710 -> 0.717 while mAP50 stays flat at ~0.982. Both gains are significant under a paired bootstrap over images (p < 0.001, p = 0.024); neither mAP50 change is distinguishable from zero. Added resolution buys boundary precision, not detection. Second, architecture is nearly irrelevant in-domain: a query-based transformer with 2.1x the parameters is statistically equivalent to a one-stage detector (95% CI [-0.0064, +0.0064]), only marginally better under domain shift, 5.5x slower on CPU and not executable under standard ONNX runtimes. Third, three targeted interventions fail: a LoRA-adapted self-supervised encoder underperforms, a promptable foundation segmenter degrades masks by 39%, and globally optimal anatomical label assignment yields +0.0007 despite correcting a constraint violated in 40% of out-of-domain predictions. Zero-shot transfer to an independent multi-centre cohort, verified overlap-free, costs 62% of mask mAP50-95 but only 18% of mAP50, reproducing the dissociation. Decomposing masks along the tooth axis localises the residual error to the apical third. Boundary precision is therefore the binding constraint, and effort is better directed at resolution and acquisition diversity than at architectural novelty.

Consensus protocols ensure the robustness and scalability of blockchains and decentralized applications built on them. However, existing consensus mechanisms often impose high computational cost or require heavy communication overhead. To address these challenges, we propose proof of verifiable delay (PoVD), a lightweight consensus protocol based on the verifiable delay function (VDF). We present the detailed protocol of PoVD, including the block mining and verification rules, and illustrate how PoVD can flexibly adjust the block time distribution according to the network condition. Through mathematical proof, we point out that, under a relatively weak network assumption, PoVD can achieve a lower fork rate than PoW while maintaining the same throughput. Our experiments verify that PoVD exhibits low computation and communication complexity and can also improve the blockchain consistency, making it particularly suitable for resource-constrained nodes and bandwidth-limited networks.

Large language models (LLMs) are increasingly deployed for enterprise information extraction (IE), where the same document must be reorganized differently for each user. Existing prompt optimization methods, however, rely on a single prompt optimized against a global objective, which is misaligned with the inherent user heterogeneity of real workplaces. We formulate enterprise IE as per-user prompt adaptation under interaction feedback and propose Self-Meta-Evolve, a hierarchical framework that maintains a dedicated prompt for each user and continuously refines it through a dual-loop process: an inner loop that edits structured prompts based on persona-conditioned feedback, and an outer loop that evolves the meta-prompt itself by distilling successful editing patterns. To enable scalable training and evaluation, we release a persona-driven IE benchmark of 292 simulated enterprise users, paired with a reproducible persona-generation pipeline grounded in O*NET occupational taxonomies. On this benchmark, Self-Meta-Evolve achieves a 74.58% success rate, outperforming the strongest prompt-optimization baseline by 13.56 absolute points, and reaches 52.54% within only two iterations. A double-blind human study with twenty real professionals further confirms that prompts adapted by our framework win against static baselines in 71% of pairwise comparisons.

Problematic internet use affects a growing share of the population, yet common interventions, e.g., time limits, blocking, forced breaks, are coercive and easily circumvented. We explore a less restrictive alternative: adapting the emotional intensity of visual content. Prior work has shown that optimization can steer an image's affective content, but its per-image optimization cost makes it impractical for real-time deployment. We instead learn a model that predicts this transformation in a single forward pass: a MobileNetV4 backbone with FiLM-based emotion conditioning outputs parameters for differentiable global transformations. This replaces prior iterative optimization (80 s per image) with a single 3.7 ms forward pass. In a user study (N = 54), the model reduced viewer-reported arousal relative to unedited images, comparably to the grayscale well-being filter, while being rated higher in perceived quality. We integrate the model into an Android app that adapts Instagram video in real time, sustaining 60 fps on a Samsung Galaxy S23.

Fine-grained object (FO) manipulation requires robots to distinguish a specified FO from visually similar objects and execute actions reliably despite scene distractors. However, scene-level visual conditioning lacks explicit object selection, while category-level guidance cannot reliably distinguish FOs within the same category. We present a SAM3-guided visuomotor policy framework that addresses these challenges through persistent object memory and focused visual conditioning. First, we introduce FO Memory-driven SAM3 (FOM-SAM3), which learns reusable FO memory tokens from limited multi-view registration images, while keeping SAM3 fully frozen. Through one-vs-rest learning, these tokens encode persistent memories for localizing target FOs and rejecting similar alternatives, which can be stored in a memory bank for reuse. Second, we propose Focused Spatial-Appearance Encoding (FSAE), which combines in-FO appearance features with explicit bounding boxes to condition action policies, including Diffusion Policy (DP) and Action Chunking with Transformers (ACT). The effectiveness of the proposed FOM-SAM3 is validated on the FO-30 dataset comprising 30 physical objects across four coarse categories. Across three real-robot FO manipulation tasks, our FOM-SAM3-guided policies demonstrated robustness against distractors, discriminative ability of similar FOs, and extendibility to new FOs.

On-policy distillation (OPD) improves reasoning models by learning the token-level discrepancy between a stronger teacher and an on-policy student. However, this discrepancy does not purely reflect the capability gap between the teacher and the student: it also contains deviations arising from the teacher itself, which are consequently mixed into the observed teacher–student discrepancy and indiscriminately learned by standard OPD during training. This issue is further exacerbated by privileged OPD, where privileged information induces larger teacher-side likelihood shifts, thereby encouraging the student to learn more of the teacher's own deviation. We introduce Calibrated On-Policy Distillation (Cal-OPD), which estimates the teacher's self-deviation region through positive and negative privileged interventions and calibrates the original teacher–student discrepancy by retaining only the component that lies beyond this region. Experiments on mathematical reasoning benchmarks show that, while retaining only about 52–65% of the original teacher–student discrepancy as the optimization signal, Cal-OPD consistently outperforms standard OPD and its variants across model scales.

Self-play in high-throughput simulators yields driving policies with robust closed-loop performance, but improvement per unit of simulation diminishes as training scales. Policies learn to handle common situations early, while further rollouts repeatedly encounter unresolved failures. Post-training offers an opportunity to target these failures, but existing methods primarily evaluate alternative actions or continuations at visited states, although successful recovery may require changing driving style earlier. We propose CounterPlay, a counterfactual self-play post-training approach that backtracks from failed tasks and retries them under alternate driving styles. CounterPlay rests on three key components. First, failure-driven backtracking uses the policy's value estimates to select an earlier stored state from which to retry the task. Second, reward conditioning enables a single policy to retry the task from this state using candidate styles ranging from cautious to aggressive. Third, CounterPlay retains task-completing retries only if no other vehicle incurs a new or earlier collision or off-road event relative to the factual branch. Retries that pass verification with fresh randomness are then distilled into the policy under its deployment condition. On BehaviorBench, CounterPlay achieves state-of-the-art scores on both the Interactive and Random splits across all eight traffic regimes using 1B post-training transitions, which is just 1% of the anchor's 100B self-play training budget. Improvements over the anchor hold across all three evaluated driving styles. CounterPlay resolves a substantial fraction of the anchor's timeout cases on BehaviorBench and achieves a balance between task completion and safety that neither continued self-play nor adopting a more aggressive driving style attains.

Defect prediction depends on knowing which commits fix bugs, yet the labels that encode this are produced by routes that each introduce noise. Reused benchmarks carry documented data-quality problems, issue-tracker links are biased and the underlying reports are frequently mistyped, and matching keywords in commit messages is a coarse heuristic. This paper examines whether commits can be labelled as bug fixes from their content alone, using open-weight language models that run locally and therefore keep the process reproducible, inexpensive at corpus scale, usable on proprietary code, and independent of any issue tracker. Against datasets of manually validated and curated bug fixes spanning Java, Python, and JavaScript, we compare a keyword baseline with a set of open-weight models of varying size, prompting each with the commit message and the code diff. On the manually validated corpus the keyword baseline recovers fewer than half of the fixes, whereas the open-weight models recover the large majority and outperform it repository by repository with statistical significance, and larger models do not consistently outperform smaller ones. We further show that evaluation corpora without negative examples cannot support a precision-aware comparison of such classifiers. We release the labelling pipeline together with a labelled, multi-language corpus produced by the recommended configuration, as a reproducible silver-standard resource for building current, project-specific datasets.

The notion of induced packing treewidth aims to unify classes defined by forbidden induced subgraphs or induced minors with classes defined by the existence of certain structured tree decompositions. For a graph $H$, \emph{induced $H$-packing treewidth}, denoted by $\treepi_{H}$, is a tree-decomposition-based graph parameter that, for each bag, measures the maximum number of pairwise anticomplete induced copies of $H$ intersecting that bag. This notion generalizes some previously studied parameters: when $H=P_1$, it is equivalent to tree-independence number, and when $H=P_2$, it is equivalent to induced matching treewidth. We prove the following: \begin{itemize}[itemsep=2mm,leftmargin=6mm] \item For all $a,t\in \mathbb{N}$, $K_{a,a}$-free graphs of bounded induced $P_t$-packing treewidth have bounded tree-independence number. This extends the previous result of Abrishami et al. [SIAM J. Discrete Math., 2025] for $t=2$, and a result of Hajebi and Spirkl who showed that $(P_t,K_{a,a})$-free graphs have bounded tree-independence number. \item If $H$ is any fixed path or a star, then the class of graphs of bounded induced $H$-packing treewidth is $\chi$-bounded. Again, this extends the previous result of Abrishami et al. [SIAM J. Discrete Math., 2025] for $H=P_2$. \item Finally, we study the relationship between induced packing treewidth and sim-width, a width parameter based on branch decompositions. We show that, although sim-width and induced $P_3$-packing treewidth are incomparable, graphs of bounded sim-width that exclude all \emph{$H$-obstructions}—certain graphs that force large induced $H$-packing treewidth—have bounded induced $H$-packing treewidth. This simultaneously generalizes and resolves questions posed by Abrishami et al. [SIAM J. Discrete Math., 2025] and Brettell et al. [European J. Comb., 2025]. \end{itemize}

Kannan's algorithm, as analyzed by Hanrot and Stehlé in 2007, solves the exact Euclidean shortest vector problem in polynomial space and $n^{\frac{n}{2e}+o(n)}$ time. In the classical setting with polynomial space, we obtain the first improvement on this bound via a randomized algorithm that runs in $n^{\frac{n}{4e}+o(n)}$ time. The main idea is to represent a fixed shortest vector in many ways as a difference of samples, thereby enabling the low-space collision search of Lyu and Zhu (SODA 2023) to replace exhaustive enumeration in the original analysis.

Model-free reinforcement learning can acquire contact-rich robotic manipulation skills through trial-and-error interaction, but it often requires the policy to learn both task strategy and low-level motion generation. In this setting, the action representation is critical because it determines how policy outputs are converted into robot motion, shaping both exploration and physical execution. Direct Cartesian command interfaces require the policy to generate motion at every decision step, coupling task-level adaptation with continuous low-level control and increasing the learning burden. We propose PA-RL, a reinforcement-learning framework that uses artificial potential fields as the action representation. Instead of commanding motion directly, the policy adapts the parameters of an energy-like potential field, which generates a state-dependent guidance direction executed through a Cartesian impedance controller. We evaluate PA-RL on peg-in-hole insertion, a representative contact-rich task with nonlinear dynamics and discontinuous contact transitions. In simulation, PA-RL is compared with Cartesian velocity, Cartesian pose, and variable-impedance action spaces using the same RL algorithm. PA-RL is the only method to reach a 100% evaluation success rate within the allotted training time, while the best baseline reaches 92.6%. It also reduces joint-torque variation by 55.4% and Cartesian acceleration variation by 70.8% relative to the best baseline, without explicit motion-quality penalties in the reward. The simulation-trained policy further completes 9/9 real-robot insertions without fine-tuning, demonstrating the deployment feasibility of the learned potential-field interface.

Despite a growing demand for causal evidence about social media, independent researchers remain severely constrained in their ability to conduct experiments directly on online platforms. To cope, multiple methodological workarounds have emerged - from controlled surveys and simulations to client-side overlays and platform partnerships - each requiring distinct trade-offs between desirable experimental properties. The recent emergence of open social media platforms offers a qualitatively different methodological opportunity. Here we propose a design space of social media experimentation and discuss Open Platform Field Experiments (OPFEs). OPFEs represent a distinct class of experimental approaches that enable independent researchers to directly intervene on functional platform components - such as clients, recommendation systems, and moderation services - within live social media environments. Through a comparative analysis of experimental archetypes, we show that OPFEs occupy a previously unexplored region of the design space. We then bridge theory and practice by characterizing the architectural and governance elements that enable OPFEs, mapping them onto Bluesky and the AT Protocol, and illustrating the end-to-end lifecycle of a complete OPFE design. Overall, this work establishes OPFEs as a practical methodological paradigm for independent, transparent, and ecologically grounded experimentation on open social media.

Recurrent Transformers increase computational depth through temporal recurrence, feeding each token's high-level hidden state into the computation of the next. This raises a natural question: is additional computation better spent on more temporal steps or greater physical depth? We investigate this question using Latent Recurrent Transformers (LRT), which retain one backbone forward pass per vocabulary token during decoding and provide a controlled setting for comparing these two ways of adding computation. We insert a latent thought token between consecutive vocabulary tokens, adding a hidden-state refinement step through the shared backbone before predicting the next token. On 16- and 20-layer NanoChat MoE backbones, we compare this model with double-depth vanilla Transformers and LRT, two-loop Transformers, and PonderLM-2 with one thought token. All configurations execute 2L blocks per vocabulary token during decoding. Under matched training FLOPs, LRT with one thought token outperforms PonderLM-2 and overtakes the two-loop Transformer as compute increases. Its gap to vanilla double-depth narrows from 0.014 to 0.001 bits per byte, despite approximately 46% fewer parameters and approximately half as many unique training tokens. At the training budget used for our main comparison, one thought token recovers an average of 89% of the gains from doubling LRT depth across the two backbone sizes, with approximately 48% fewer parameters. These results highlight temporal thinking as a parameter-efficient way to extend computation in recurrent Transformers.

Let $\mathcal A$ be a complete deterministic finite automaton on a state set $Q$ of size $n$ with $k$ letters, and for a proper nonempty subset $S$ of $Q$ let $\mathrm{minext}(S)$ be the length of a shortest word $u$ with $|Su^{-1}|>|S|$, where $Su^{-1}=\{q: q\cdot u\in S\}$. To each state $q$ attach the integer $\beta^{\ast}_q=\sum_{t=1}^{n-1}k^{\,n-1-t}(\mathrm{indeg}_t(q)-k^{t})$, where $\mathrm{indeg}_t(q)$ counts the pairs $(p,u)$ with $|u|=t$ and $p\cdot u=q$, and let $B(S)=\sum_{q\in S}\beta^{\ast}_q$. On every synchronizing automaton, $B(S)\ge0$ implies $\mathrm{minext}(S)\le n-1$, so, as $B(Q)=0$, one of $S$ and $Q\setminus S$ extends within $n-1$; when $B(S)>0$ no hypothesis is needed. Kari's Eulerian extension lemma is the case $\beta^{\ast}=0$, and $\beta^{\ast}$, like every member of the family $\sum_{t=1}^{n-1}c_t\sigma_t$, $c_t>0$, vanishes identically if and only if the automaton is Eulerian, where $\sigma_t(S)=\sum_{q\in S}(\mathrm{indeg}_t(q)-k^{t})$. On strongly connected automata $\sigma_t(S)/k^{t}$ has Cesàro limit $n\,e(S)/e(Q)-|S|$ for Friedman's weight $e$; that limit certifies singletons but no larger subset in general. The hypothesis $B(S)\ge0$ cannot be relaxed by one integer unit, nor can the constant $n-1$ be improved. A second-moment test on the sizes $|Su^{-1}|$ certifies 60 to 95 percent of the subsets with $B(S)<0$ at $n\le7$. Along non-Eulerian automata whose words of length $n-1$ merge a fraction of the state pairs bounded below, with $\max_q\mathrm{indeg}_{n-1}(q)=o(nk^{n-1})$, it certifies all but a vanishing share of them. The functional $B$ certifies half of the subsets outside $\{B=0\}$. At each subset size coprime to $n$ ($n\ge4$) some synchronizing Eulerian binary automaton attains the constant $n-1$; whether only there is open. No reset bound follows: Černý's automata have subsets not extending within $n-1$.

Given a cyclic code $C_g = \langle g(x) \rangle$ of odd length $m$ over $\mathbb{Z}_4$, one common way to build a quasi-cyclic (QC) code is to pick $f_1, \dots, f_\ell \in \mathbb{Z}_4[x]$ and let $C = \langle (f_1 g, \dots, f_\ell g) \rangle$. Because $\mathbb{Z}_4$ is not a field, the type of $C_g$ ($4^{k_1}2^{k_2}$) is not necessarily inherited by $C$. Determining conditions under which the type of $C_g$ is inherited by $C$ was posed as an open problem recently in \cite{AydinLuOnta2023}. In this paper, we settle this problem. We first give two sufficient conditions for the type to be preserved: one requires a single $f_i$ to be coprime to $x^m-1$ over $\mathbb{F}_2$, the other only requires the $f_i$ to be jointly coprime to it. Neither condition is necessary in general. Using the fact that $x^m-1$ is squarefree for odd $m$, we decompose $\mathbb{Z}_4[x]/\langle x^m-1\rangle$ into a product of finite chain rings using Chinese remainder theorem (CRT) and derive a condition on $f_1, \dots, f_\ell$ that is both necessary and sufficient for $C$ to match the type of $C_g$. This condition depends only on the irreducible factors of $x^m-1$ where $g$ does not already vanish. This also yields a simple test for when $C$ is a free $\mathbb{Z}_4$-module. Finally, we report many new QC codes over $\mathbb{Z}_4$, found by computer searches using Magma software~\cite{Magma1997} guided by this criterion, with Lee distances greater than previously known codes of the same type.

Access to academic support is a key determinant of student success, yet students experience it unequally: some readily seek help from lecturers or tutors, while others hesitate due to anxiety, fear of judgement, uncertainty about expectations, or low confidence in their understanding. This may be especially evident in computing education, where programming tasks are cumulative and cognitively demanding. Although students increasingly turn to general-purpose generative AI tools, these can produce responses that are inaccurate, insufficiently contextualised, or misaligned with module expectations. This study presents and evaluates Beacon, a course-specific Retrieval-Augmented Generation (RAG) system providing private, immediate, module-aligned academic support. Grounding responses in approved teaching materials, Beacon was designed to lower barriers to help-seeking while encouraging independent learning. Using a design-based research approach, Beacon was developed iteratively and evaluated via mixed methods, combining questionnaires and semi-structured interviews with students and staff at a Higher Education institution. Students described Beacon's responses as closely aligned with module content and more trustworthy than unrestricted generative AI tools, valuing its use of pseudocode and scaffolded explanations over direct solutions. Although participants remained cautious about trusting AI-generated responses without verification, they viewed the system as a valuable first point of support before consulting lecturers or official resources. The findings suggest that carefully designed course-specific AI systems may reduce barriers to academic support by occupying an intermediary space between independent study and formal support. Rather than replacing educators, educational AI may be most valuable when it broadens access to guidance while preserving the pedagogical role of lecturers.

Fraudulent job posting detection aims to identify job advertisements that are corrupted either through fake content, misleading information, or negative intent, disrupting the online eco-system of job-seekers and employers. Existing studies in this domain lack effective methods to simultaneously achieve high accuracy and meaningful structure of latent-space representations that capture subtleties among fake posts. To this end, we propose Centroid-Guided Contrastive Loss (CGCL), a loss function which unifies classification with densely formulated clustering to consistently reshape latent-space through a centroid-driven top-$k$ push-and-pull mechanism. The complementary nature of CGCL enables the model to enforce accurate decision boundaries and maintain high clustering compactness, effectively capturing both class separability and latent structure. Extensive experiments demonstrate the state-of-the-art (SOTA) performance of our method on EMSCAD, a public benchmark dataset. The code associated with this work is available at: https://github.com/ali-ahmed925/CGCL_code

Monocular 6-DoF pose estimation of non-cooperative targets is important for on-orbit servicing and debris removal. A single-image estimator can confuse near-symmetric spacecraft orientations, and tracking can preserve an incorrect pose. We present Hypothesis-Anchored Tracking (HAT), a causal framework that uses inter-frame motion to select among competing CAD-based pose hypotheses before alignment and fusion. Rather than independently choosing the highest-scoring hypothesis in each image, HAT retains competing orientation histories and selects a pose to anchor the relative trajectory estimated by monocular SLAM. Sparse anchors and pose fusion provide per-frame estimates after initialization without revising past outputs. The method requires only a calibrated RGB sequence, a metric CAD model, and target image regions, which can be supplied by detection or segmentation. The pretrained pose and SLAM networks require no target-specific training or fine-tuning. We evaluate two versions, Mega-HAT and Pico-HAT, using MegaPose and PicoPose, on SPARK-2024, SwissCube and SHIRT, with YCB-Video assessing performance outside the space domain. Using one temporal configuration per method, the arithmetic means of the four dataset-wise comparisons show 9.4% lower mean pose error and 3.76 times the sustained input FPS for Mega-HAT relative to independent MegaPose, and 23.9% lower mean pose error and 2.42 times the FPS for Pico-HAT relative to independent PicoPose. Mega-HAT ablations on SPARK and an offline reference examine component contributions and the effect of revising past estimates.

In-context learning using Large Language Models (LLMs) offers a compelling path to training-free post-OCR correction, yet its effectiveness for Devanagari script remains entirely unexplored. We present the first systematic evaluation of LLMs (3B-32B) for post-OCR correction in Hindi and Marathi, comparing three in-context example retrieval strategies: domain-random selection, dense semantic retrieval, and our proposed CharBM25, which retrieves examples by character n-gram BM25 similarity over OCR inputs to target shared error patterns with the test sentence. Across a 20,000-sentence benchmark spanning five news domains, retrieval strategy is the decisive factor in correction quality: CharBM25 outperforms domain-random selection by 2.8-4.0pp absolute WER on Hindi and 2.9-3.8pp on Marathi, using character trigrams, which consistently outperform bigrams and unigrams. Scale dominates performance: Gemma-3-27B achieves WER reductions of 55.0% for Hindi and 33.3% for Marathi under CharBM25-5. Few-shot gains are capacity-gated: models below 8B do not reliably improve over the OCR baseline, and on Marathi the smallest models (3B) degrade more sentences than they improve. Marathi is persistently harder to correct than Hindi across all scales, reflecting its greater morphological complexity. These findings establish CharBM25 as an effective, GPU-free retrieval strategy that matches or exceeds dense retrieval at negligible computational cost, and show that combining it with a general-purpose LLM of 12B+ parameters delivers reliable, training-free Devanagari post-OCR correction without task-specific fine-tuning. Dataset: https://huggingface.co/datasets/AbhishekBhandari/Devanagari-OCR-ICL-Benchmark

Declarative SPMD programming uses tensor sharding descriptions to drive distributed execution, separating parallelization from model code. However, the evaluated PyTorch DTensor stack dispatches every operator below autograd, incurring repeated dispatch and metadata costs, while lacking an inexpensive end-to-end validation path. Existing FSDP and distributed Muon implementations also mismatch two-tier supernode topologies: FSDP relies on explicit parameter packing and unpacking, and Muon's whole-matrix orthogonalization conflicts with parameter sharding. We observe that distributed tensors need only express sharding semantics at the tensor API boundary above autograd, allowing differentiation and kernels to operate on plain tensors. Based on this insight, we present HyperParallel-FSDP, featuring: (1) dual-mode DTensor execution, using one sharding plan for both a production mode with one-time layout resolution and no steady-state dispatch overhead, and a validation mode with end-to-end metadata propagation, fail-fast checks, and gradient-equivalence testing; (2) topology-aware FSDP, with zero-copy intra-supernode collectives, fused inter-supernode reduction, and a cross-layer backward pipeline that avoids waits on slow links; and (3) layout-driven distributed Muon, with sharding-derived communication groups, deduplicated orthogonalization, and shape-fused Newton-Schulz iterations. On Atlas 900 A3 SuperPoD, HyperParallel-FSDP scales from 16 dies to 384 cards (768 ranks), sustaining 421k tokens/s for a 505B-parameter MoE while FSDP communication uses 2.9% of step time. It reduces mean step time by 29.7% versus PyTorch FSDP2 and 25.5% versus Megatron DDP, with Pearson correlation above 0.999997 over 1,000 steps. Distributed Muon improves profiler step time by 5.4-16.0% over competing systems. Source code is available at https://atomgit.com/mindspore/hyper-parallel.

Four-dimensional scanning transmission electron microscopy (4D-STEM) records a two-dimensional diffraction pattern at each electron-probe position, yielding spatially resolved reciprocal-space information but large, heterogeneous data volumes. Here we describe 4D-ImageNet, a collection of 174,000 diffraction patterns comprising 145,000 experimental patterns selected from 29 acquisitions and 29,000 multislice simulations. The experimental data cover acquisition-level labels for Ag, Au, mixed Au-Ag, CoO, Pd and ZnO specimens across multiple fields of view, scan dimensions, camera lengths and exposure times. Each acquisition contributes 5,000 quality-ranked patterns with source scan coordinates and acquisition metadata. A set-prediction detector provides model-derived pseudo-labels for the direct-beam position and Bragg-disk centres, with a confidence score for each disk. The simulation data cover 13 crystal structures and include Euler rotations, reciprocal-space sampling and approximate low-index beam directions. A grouped mixed-domain masked-reconstruction benchmark is provided to assess leakage-resistant loading and evaluation across experimental and simulated data. The dataset is intended for representation learning, disk detection, diffraction-pattern retrieval, orientation analysis and simulation-to-experiment studies.

The kissing number in dimension $d$ is the largest number of non-overlapping congruent spheres that can simultaneously touch a central sphere of the same size. We study dimensions $25$-$31$, where the best previous constructions are based on Leech lifting from the optimal kissing configuration in dimension $24$. Our method exploits the absence of contacts between the unlifted bulk and the block consisting of lifted and auxiliary vectors. Rotating this block while keeping the bulk fixed creates room for two antipodal points in dimensions $26$, $27$, and $28$, and one point in dimension $29$. Three further modifications yield improvements in dimensions $25$, $30$ and $31$: (a) a nonorthogonal diagonal linear deformation of the lifted block admits two antipodal points in dimension $25$; (b) rotating the additional coordinates of the lifted vectors and then applying a small orthogonal transformation to the resulting lifted block as a whole admits two antipodal points in dimension $30$; (c) rotating only the additional coordinates of the lifted vectors admits four nonantipodal points in dimension $31$. Together, these constructions yield the new lower bounds $\tau_{25}\geq 197058$, $\tau_{26}\geq 198552$, $\tau_{27}\geq 200046$, $\tau_{28}\geq 204522$, $\tau_{29}\geq 209497$, $\tau_{30}\ge 220442$, and $\tau_{31}\geq 238354$.