Earth and Planetary Astrophysics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Tuesday 8 September
Centaurs occupy dynamically unstable orbits between Jupiter and Neptune and are typically removed by giant-planet encounters within a few Myr. Bailey and Malhotra (hereafter BM09) classified the known Centaurs, from 2007 orbits, into a short-lived diffusing class (D), a resonance-hopping class (R), and a long-lived, quasi-stable class (Q); their sole Q object, 2005 TH173, was fit to a 17\,d arc. We revisit both halves of that analysis on modern Jet Propulsion Laboratory Small-Body Database (SBDB) orbits (retrieved 2026-07-26), integrating clone ensembles of each object with the REBOUND WHFast $N$-body integrator. We ask (i) whether tighter orbital precision narrows the dispersion in clone lifetimes, and (ii) whether the BM09 labels survive on updated elements. Among 65 genuine Centaurs with arc $\ge 30$\,d and no censored clones, $\sigma(\log_{10}\mathrm{lifetime})=0.54$\,dex (a factor of 3.5). Conditioned on present-day $a$, $e$, $i$, and $q$, this dispersion is insensitive to relative semimajor-axis uncertainty (slope $=-0.002$\,dex per decade in $\sigma_a/a$; 95% confidence interval $-0.034$ to $+0.030$) — a chaos floor. On the D/Q axis, 49/50 comparable objects retain their BM09 class; 2005 TH173 is the exception (Q$\to$D): 80/100 clones escape before 10\,Myr (median 2.99\,Myr), and a 40\,Myr integration of the same initial conditions returns 88\,D / 12\,R / 0\,Q. None of the 30 multi-opposition 2007 orbits changes class; two short-arc objects leave the Centaur zone.}
Machine Learning
Chimaera: A Mixture-of-Graph-Experts Architecture for Cross-Task and Cross-Dataset Graph Learning
Designing foundation models for graphs is challenging due to the irregular structure of graphs and the different sizes and characteristics of embeddings. Chimaera integrates mixture-of-experts with graph foundation models (GFM). It integrates different GFM architectures, such as graph prompts and linear GNN models. Large language models are used to generate embeddings, and experts can be trained and combined following different strategies, GFMs, embeddings, etc. Furthermore, Chimaera extends existing linear GNNs to support link-level and graph-level tasks in addition to node-level tasks. Empirical analyses are performed on same-task and cross-task experiments with node, link, and graph classification tasks using six benchmark text-attributed graph datasets. The experiments demonstrate the effectiveness of Chimaera and its capabilities for transfer across tasks and datasets. Further insights include the need to use both large and small language models to generate embeddings for the experts, a strong cross-task transferability of simple but effective linear GNNs, and using few samples only to provide strong results.
Read the full abstract
Systems and Control
Model Predictive Control for Dynamic Hydraulic Balancing in Building Radiator Heating Systems: Modeling, Design, and Experimental Validation
Hydronic radiator systems are among the most widely used heating systems in buildings. In the literature, radiator heat outputs are often assumed to be independent of one another and arbitrarily adjustable in time when designing controllers. However, in practice, radiators are supplied by one or multiple common heat sources and are hydraulically coupled through the water circulation system. Maintaining occupant comfort in multi-zone buildings therefore requires not only an appropriate supply temperature but also proper hydraulic balancing to distribute the available water flow according to the heating demand of each zone. To this end, we develop a grey-box thermal model that captures the hydraulic interactions among radiators and the effects of radiator valves, circulation pumps, and heat sources. In a real building, we show that accounting for hydraulic interactions reduces the root mean square error (RMSE) between the measured and modeled zone temperatures by approximately 11% compared with a model that neglects these interactions. Additionally, we integrate the developed model into a model predictive control (MPC) framework for dynamic hydraulic balancing that jointly optimizes valve openings and the supply temperature to maintain thermal comfort in each zone while reducing energy consumption. Through both real-world experiments and numerical case studies, we demonstrate that, compared with existing MPC formulations that neglect hydraulic interactions or do not control radiator valves, the proposed MPC reduces comfort-range violations by at least 27% while requiring a similar or even lower cumulative supply temperature.
Read the full abstract
Materials Science
Pressure Dependence of Ferromagnetism in Uranium Hydride UH$_3$ up to 20 GPa Revealed by NV Centers Magnetometry
The origin of ferromagnetism in the stable $\beta$-UH$_3$ phase is still debated. With a high Curie temperature and a short uranium-uranium interatomic distance, $\beta$-UH$_3$ is placed outside the known systematics of uranium compounds. Pressure provides a powerful means of tuning magnetism toward its suppression, thereby offering valuable insights into its underlying mechanisms, yet the high-pressure behavior of $\beta$-UH$_3$ has remained largely unexplored. Here, we combine in the Diamond Anvil Cell (DAC) the development of pure $\beta$-UH$_3$ synthesis and widefield nitrogen-vacancy (NV) center magnetometry to measure the pressure dependence of the Curie temperature in $\beta$-UH$_3$ up to about 20~GPa. We present two data analysis methods, vector magnetic field reconstruction and statistical analysis of the optically detected magnetic resonance (ODMR) response, which enable us to directly image the magnetic dipole of $\beta$-UH$_3$ during isobaric warming. We observe a linear decrease in the Curie temperature with pressure, yielding a slope of $\text{d}T_C/\text{d}P=-3.86\,(8)$~K/GPa. Ferromagnetism in $\beta$-UH$_3$ is predicted to reach 0~K at approximately 44~GPa, where a quantum critical point may emerge.
Read the full abstract
Number Theory
On a problem of Erdős and Nathanson related to minimal asymptotic bases of order $h$
Let $h\geq 3$ be an integer and $0<\alpha<1/h$. In this paper, we prove that there exists a minimal asymptotic basis of order $h$ with asymptotic density $\alpha$. This solves an open problem posed by Erdős and Nathanson in 1988.
Computer Vision
Enhancing Table Structure Recognition via Bounding Box Guidance
Table Structure Recognition (TSR) aims to extract the bounding boxes of cells and table structure (e.g., HTML) from table images. Although current approaches have made significant progress, the latest image-to-sequence methods overlook the explicit utilization of the bounding box information when predicting HTML sequences, leading to error predictions in complex scenes. In this paper, we introduce a novel framework BGTR (Bounding Box-Guided Table Recognizer). To more effectively utilize bounding box information, we first predict the bounding boxes of cells and then use this information to guide the generation of HTML sequences. While utilizing bounding box information can enhance the accuracy of HTML sequences, for natural scene tables, the data volume is too small to allow for sufficient training of bbox-guided HTML generation. In response, we adopt a progressive training method for natural scene tables and introduce SNSTab, a synthetically generated natural scene table dataset. Our experiments on five benchmark datasets demonstrate SOTA performance.
Read the full abstract
High Energy Astrophysical Phenomena
Supermassive black hole binaries in the multi-messenger context of ground and spaceborne VLBI
Evidence of a gravitational wave background suggests the existence of a population of sub-parsec supermassive black hole binaries (SMBHBs), with characteristic angular separations on the order of 1-10 $\mu$as. Spaceborne extensions of Very Long Baseline Interferometry (VLBI) and the next generation ground arrays introduce the possibility of directly imaging these SMBHBs. In this work, a binary spectral energy distribution model is used to predict the detectability of SMBHBs with ground and spaceborne VLBI. We consider the Black Hole Explorer (BHEX), a proposed spaceborne VLBI mission, as our primary case study. We explore the detectable SMBHB parameter space and identify distinguishable binary signatures in the visibility domain. We find that for a flux-density-limited sample, ground array observations are more effective at detecting a wider region of the binary parameter space, with $M_\mathrm{tot} \gtrsim 10^9$ solar mass systems detectable out to redshift $z=0.075$ and beyond, depending on mass ratio. Conversely, inclusion of a spaceborne element such as BHEX, offering finer angular resolution ($\sim6\,\mu$as) and sampling of the (u,v) plane not limited by Earth rotation synthesis, will provide significant benefits in constraining binary properties. An illustrative Fisher analysis shows improvements in characterisation of the separation and position angle of SMBHBs by a factor of $\sim4$. Near-future ground and/or spaceborne VLBI may achieve the first direct observation of a SMBHB, contributing significantly to multi-messenger studies of such systems with pulsar timing arrays and observations across the electromagnetic spectrum.
Read the full abstract
Sound
TontaubeV1: Streaming Text-to-Speech with Hierarchical Codec Modeling and Bounded Context
Text-to-speech systems often face a trade-off between natural prosody and efficient inference: higher perceptual quality usually comes at increased computational cost and latency. We present TontaubeV1, a model that preserves natural prosody while enabling streaming from a single consumer GPU. Speech is encoded by the hierarchical DualCodec representation at 12.5 Hz, which separates a semantic stream from successive acoustic refinements. Our design assumes that prosodic structure is largely established when the semantic stream is generated, and allocates capacity accordingly: a Qwen3-1.7B-derived transformer predicts that stream and thereby the utterance duration, while three progressively smaller Qwen3-0.6B-derived transformers each add one acoustic refinement. Text is tokenized per character rather than by subword. Paired text and audio markers at shared positions support long-form generation with bounded context, and overlapping DualCodec reconstructions are mapped into the VibeVoice acoustic latent space and decoded causally, enabling streaming despite DualCodec's noncausal decoder. The model accepts up to one minute of reference audio for voice conditioning and is designed primarily for English and German, with additional multilingual support. The four predictors total 2.9B parameters; on a single RTX 5090 the streaming path reaches about 200 ms to first audio. In separate non-streaming measurements, the end-to-end real-time factor (RTF) is 0.08 for one input and the aggregate RTF is 0.02 across eight concurrent inputs. In a blinded study, 40 listeners preferred TontaubeV1's prosody over Fish Audio S2 Pro in 70.25% of 400 comparisons (95% CI: 64.2-75.7%). On an LLM-as-a-judge benchmark, it matches ElevenLabs Flash v2.5 and outperforms the April 2026 Gradium API and Cartesia Sonic 3 on prosody. The weights are released on Hugging Face under the Tontaube Community Model License 1.0.
Read the full abstract
Probability
Scaling Limit of Critical Loop Soup Clusters in Three, Four, and Five Dimensions
We prove that on the metric graph of $\mathbb{Z}^d$ for $d\in \{3,4,5\}$, clusters of the critical loop soup (or equivalently, clusters of the critical Gaussian free field level-set) admit a scaling limit. To the best of our knowledge, this presents the first scaling limit result for a percolation model on a three-dimensional lattice.
Read the full abstract
Probability
Dynamical and stationary non-equilibrium fluctuations for a boundary-driven exclusion process with energy
We consider the exclusion process with energy, a model with mass and energy conservation, and we put it in a boundary-driven setting. We are interested in its hydrodynamic and hydrostatic limits, together with the associated dynamical and stationary fluctuations. We provide a new proof for the derivation of dynamical fluctuations that does not require explicit estimates on the two-point correlation functions. While the hydrodynamic limit is given by two uncoupled heat equations, the dynamical fluctuations are described by a system of coupled generalized Ornstein-Uhlenbeck processes.
Read the full abstract
Instrumentation and Methods for Astrophysics
Full Characterisation of the Polarisation Primary Beam of the GRAO 32-m Telescope
Direction-dependent instrumental polarisation is a major systematic limitation in high-fidelity single-dish radio polarimetry, yet a unified characterisation of beam leakage and the conditioning of polarisation recovery is lacking for the Ghana Radio Astronomy Observatory (GRAO) 32-m telescope. We aim to establish a quantitative, direction-dependent polarimetric beam model at 5.0 and 6.7 GHz. High-resolution GRASP electromagnetic simulations of the nominal telescope configuration are used to derive the full complex Jones response and corresponding Mueller matrices, from which beam shape, beam squint, instrumental Stokes leakage, and intrinsic cross-polarisation ratio (IXR) are evaluated across the primary beam. The Stokes $I$ beams have half-power beam widths of $396.4$ and $295.3$ arcsec at 5.0 and 6.7 GHz, respectively, with main-beam efficiencies of $56.4$ and $55.6$ per cent. At 5.0 GHz, the circular-polarisation beam squint is $9.96$ arcsec ($2.5$ per cent of the HPBW), whereas no statistically significant squint is detected at 6.7 GHz. Leakage from Stokes $I$ into linear polarisation remains below $0.1$ per cent within the half-power beam but increases substantially towards the sidelobes. IXR reaches approximately $80$ dB on axis and decreases with angular offset. These results establish the intrinsic electromagnetic polarimetric response of the GRAO 32-m telescope and provide a quantitative baseline for direction-dependent calibration and subsequent observational validation.
Read the full abstract
Soft Condensed Matter
Effective particle size governs structure and dynamics in rough hard-particle fluids
We numerically investigate how particle surface roughness affects the static and dynamic properties of a hard-particle fluid across a wide range of densities, rho. These simulations of a simple model of granular systems reveal that, although the amplitude and coverage of surface corrugation significantly influence the rho-dependence of system properties, the qualitative behavior of this dependence remains unchanged. These findings can be described quantitatively by introducing an effective particle size, which enables a direct mapping of the rough particle systems to equivalent hard-disk systems. Analytical calculations provide an explicit form of this mapping, and allow us to predict the static and dynamic properties of rough particles using standard liquid-state theories for hard disks.
Read the full abstract
Computation and Language
Record Grouping Controls Evidence Weight in Language Models
Retrieved records are presentation units; a supplied partition determines which records enter a language model as one evidential contribution. We characterize the invariant group-content state that removes within-group copies while retaining complementary canonical content, show that equal group counts can encode different evidence states, and derive a sharp content-aware partition-error bound. Given a supplied partition, our pre-generation representation deduplicates and aggregates content within groups and bounds each group's contribution. Across 104,402 trials and 6 public checkpoints, a central natural-text intervention finds that content-fixed false splits add 10.27-32.66 percentage points and false merges remove 9.13-31.79 points; a matched six-slot control retains the positive direction in all 16 cells. In a new 48-item controlled campaign panel, changing the supplied partition produces measurable, checkpoint-dependent decision shifts across all four models, and the balanced mirror design exposes substantial order interactions. Together, the theory and experiments establish the supplied partition as a controllable pre-generation representation variable and characterize its checkpoint-dependent behavioral effects.
Read the full abstract
High Energy Physics (Theory)
Mass for Particles from Extra Dimensions
We investigate a geometric mechanism for mass generation in a codimension-two hedgehog braneworld. A massless particle propagating in the bulk acquires an effective four-dimensional mass through its motion along the compact angular extra dimension, yielding a dynamically determined Kaluza-Klein spectrum after canonical quantization. While the bosonic sector exhibits no radial confinement, extending the analysis to N = 1, 2 spinning-particle models shows that the spin-curvature coupling modifies the effective potential and introduces a level-dependent localization pattern for the Kaluza-Klein modes. The results establish a particle-mechanics realization of geometric mass generation in flux-supported codimension-two braneworlds and highlight the interplay between bulk geometry, spin, and the effective four-dimensional mass spectrum.
Read the full abstract
Multiagent Systems
MorphoOrgaAgent: A Foundation-Model-Based Multi-Agent System for Autonomous Organoid Analysis
Organoids are three-dimensional tissue models whose morphology provides important insights into tumor development, disease progression, and drug testing. Extracting these morphological features relies heavily on manual segmentation, which is time-consuming and labor-intensive. Furthermore, performing quantitative statistical analysis typically requires custom coding skills and a mathematical background, presenting a major barrier for experimental biologists. To address these challenges, we introduce MorphoOrgaAgent, a multi-agent framework that achieves zero-shot organoid segmentation, automated data analysis, and report generation based on natural language input. The framework consists mainly of three core components: a TaskUnderstandingAgent that identifies requested measurements and visualization types; a hybrid segmentation module that combines Cellpose-derived geometric prompts with text prompts to guide SAM3 for zero-shot organoid instance segmentation; and a ReportAgent that computes quantitative metrics and compiles them alongside generated visualizations into a structured report. We further introduce MorphoOrgaVQA, a benchmark designed for quantitative evaluation of agent systems in organoid morphology analysis. Experimental results demonstrate that MorphoOrgaAgent handles both explicit and descriptive user requests, produces measurements closely matching ground truth, and generates complete analysis reports without requiring manual programming. The complete source code and MorphoOrgaVQA benchmark are publicly available at https://github.com/peng-lab/MorphoOrgaAgent.
Read the full abstract
Instrumentation and Methods for Astrophysics
A Content-Addressed Workflow for Reproducible DANTE Gravitational-Wave Anomaly Analysis
Unsupervised analysis of gravitational-wave detector data combines expensive scientific stages with data acquisition, provenance checks, statistical calibration, follow-up products, and reporting. Reproducing such an analysis requires more than preserving model weights or a final candidate table: the execution graph, scientific contracts, retry semantics, and verification boundary must also be explicit. We present a local workflow layer for the Domain-Adaptive Network for Transient Evaluation (DANTE) that represents a corrected O4a analysis as a content-addressed, 15-stage directed acyclic graph. The architecture separates immutable stage evidence from mutable operational progress, binds command-line and browser interfaces to the same run identity, and resumes only work compatible with the frozen contract. The tagged v1 release records all 15 stages as verified in a machine-readable receipt and passed a documented human usability gate. The release verifies adopted scientific artifacts rather than recomputing them, and therefore makes no new claim of global significance, astrophysical discovery, or public real-time operation. This paper describes the architecture, its fail-closed controls, and the bounded evidence supporting reproducible local use.
Read the full abstract
Combinatorics
Matchings and Clusters on Plabic Fences
Fix two positive braid words $\beta_+,\beta_-$, and let $\text{Conf}(\beta_+,\beta_-)$ be the corresponding (type A) double Bott-Samelson variety. Let $\beta$ be a double braid word containing $\beta_+,\beta_-$ as the top, bottom words. We consider the open cluster torus $T(C_\beta)$ associated to a triangulation $C_\beta$ in $\text{Conf}(\beta_+,\beta_-)$ from arXiv:1904.07992, and we identify these with weighted plabic fences, where the usual local moves on planar bipartite graphs naturally correspond to change of torus coordinates in double Bott-Samelson variety. Moreover, $T(C_\beta)$ can be parametrized explicitly by certain matrix products, and also has monomial coordinates given by the cluster variables, which are matrix minors. We interpret these minors as a generalized notion of perfect matchings on weighted plabic fences, which may not be reduced plabic graphs. Using this interpretation, we show that the cluster variables are given by "generalized minimal matchings", extending the minimal matchings from arXiv:1606.08383. Lastly, we derive a Chamber Ansatz formula for double Bott-Samelson varieties via face alternating products from dimer theory. In general, plabic fences are not reduced plabic graphs, yet we are able to extend and apply the standard tools such as local moves on plabic graphs, trips, and minimal matchings to them.
Read the full abstract
Strongly Correlated Electrons
$g$-wave altermagnetic order parameter in hematite
Altermagnets combine the vanishing net magnetization of antiferromagnets with momentum-dependent spin splitting. Magnon band splitting provides a direct probe of altermagnetic order and may enable chirality-selective magnon transport, yet the momentum-space symmetry of this splitting has not been determined quantitatively. Here we use inelastic neutron scattering to map the momentum dependence of altermagnetic magnon splitting in hematite ($\alpha$-Fe$_2$O$_3$). The splitting vanishes along nodal directions and reaches maxima off the nodes, revealing the $g$-wave symmetry of the altermagnetic order parameter. These results agree with linear spin-wave theory calculations based on the altermagnetic model, which further identify the nondegenerate branches as magnons of opposite chirality and trace the splitting to symmetry-inequivalent long-range exchange interactions. Our results provide the first quantitative determination of the momentum-space symmetry of altermagnetic chiral magnons. These findings, together with hematite's high magnetic ordering temperature and low magnon damping, establish it as a promising platform for low-dissipation, symmetry-selective magnonic applications.
Read the full abstract
Computation and Language
Global Divergence, Local Convergence: Representation Geometry in SSMs and Transformers
Recent state-space models (SSMs) such as Mamba achieve language modeling performance comparable to transformers despite relying on fundamentally different architectures. This raises an important question: how do these structural differences influence the geometry and functional nature of their internal representations? We study this question through a multi-scale analysis of representations in transformers, SSMs, and hybrid architecture. First, we find that SSMs distribute their representational information evenly across all dimensions, whereas transformer representations are heavily dominated by a single principal direction. By evaluating hybrid architectures, we observe that the representation space becomes increasingly skewed toward a single dominant direction after each attention layer. Next, we explore how the different geometric spread of representations impacts representational capacity through compressibility. Surprisingly, we find that despite their contrasting geometric structures, both architectures exhibit tightly matched effective capacities. We further investigate whether this skewed geometry affects how concepts are encoded. Using rank-constrained probes, we demonstrate that both architectures encode concepts in subspaces of surprisingly similar dimensionality. Furthermore, we demonstrate that the transformers' dominant principal direction does not inherently encode more conceptual information. Finally, we zoom in and examine the alignment between manifolds, either by analyzing representations of specific topics or by looking at the nearest neighborhoods of tokens, and find that they are highly aligned. Ultimately, our analysis suggests that while transformers and SSMs induce different usage of latent space, they display a striking functional convergence at the level of local semantic manifolds.
Read the full abstract
Optics
Quantized transport of two-dimensional multifrequency solitons
We consider Thouless pumping of two-dimensional quadratic solitons composed from coherently interacting fundamental frequency (FF) and second harmonic (SH) components propagating in a periodic $\chi^{(2)}$ material. The pumping is induced by two mutually sliding two-dimensional lattices defined by shallow transverse and longitudinal periodic refractive index modulation. Focusing on solitons in the semi-infinite gap, we find three distinct pumping scenarios: the absence of transport for small-amplitude solitons, non-quantized transport in a transient regime at intermediate amplitudes, and stable quantized transport for solitons with relatively large amplitudes. Different directions of the sliding velocity were investigated resulting in different trajectories of soliton center. The transition to the regime of quantized transport is found to depend strongly on phase mismatch between FF and SH waves that also influences stability properties of two-dimensional $\chi^{(2)}$ lattice solitons. We also show that pumping with longitudinal periods in the x and y directions, whose ratio approximates an irrational number, induces quantized transport whose direction rapidly converges, with increase of the accuracy of the approximation, to a limiting pumping direction determined by this irrational number and corresponding to truly incommensurate longitudinal periods.
Read the full abstract
Machine Learning
Hyperparameter Scaling Laws Across MoE Sparsity
Mixture-of-Experts (MoE) models expand model capacity without a proportional increase in training compute, but increasing sparsity makes reliable hyperparameter transfer challenging. In this work, we show that conventional hyperparameter scaling laws are insufficient for ultra-sparse MoEs: the optimal learning rate and batch size vary with activation ratio, and these shifts cannot be explained by either total or activated parameter count alone. To characterize this dependence, we conduct 1,800 pre-training runs spanning six activated-parameter scales and models with up to 6B total non-embedding parameters, processing approximately 20 trillion tokens at a cost of 200,000 equivalent H800 GPU-hours. Our results reconcile conflicting findings in prior work by revealing two scaling regimes. At fixed sparsity, the optimal batch size follows a power-law relationship with training tokens $D$, whereas the optimal learning rate scales with training compute $C$ and remains robust to the allocation between model size and data. Across sparsity levels, the activation ratio $A$ enters both relationships as an additional multiplicative power-law factor. These observations lead to unified hyperparameter scaling laws that transfer across MoE sparsity levels. Large-scale evaluation shows that the scaling form outperforms alternative functional forms. On a held-out ultra-sparse MoE with 12B total parameters and only 1/64 of its experts activated, the predicted hyperparameters remain close to the observed optima, supporting joint extrapolation across model scale and sparsity. Further experiments demonstrate transfer across expert granularities and isolate the effect of activation ratio from that of total expert count.
Read the full abstract
Computation and Language
When Victorian Becomes a Prompt: Literary Periodization as a Generative Constraint in 100 AI-Generated Novels
Generative AI inverts the typical periodization of literary history: the periodizing tag Victorian can now come first and influence what is written. Generative periodization, defined and tested here, describes the use of literary-period designations in generating texts. I test this approach on 100 book-length novels produced under Victorian and Zero-Style conditions using GPT, Qwen, and Llama workflows. The Period Alignment Score (PAS), trained on nineteenth-century literature and benchmarked against human Zero-Style prose, assesses alignment using topic-reduced grammatical features. Victorian prompts produce consistent historical-direction shifts in GPT and Qwen, but not robustly in Llama. Victorian-only recalibration and harder comparison corpora preserve the GPT and Qwen effects. Cross-model transfer also shows a shared direction of grammatical change. The measurable target is the broader nineteenth century rather than the Victorian period per se.
Read the full abstract
Distributed Computing
Measuring Sustainability in Multi-Scale High-Performance Computing
The transition from traditional High Performance Computing (HPC) to the Computing Continuum emphasizes efficient resource management and sustainable practices across Multi-Scale hybrid architectures. This paper introduces a multidimensional metric framework to characterize these systems and guide deployment strategies for modern workloads. The framework combines Architectural Performance metrics (such as Throughput, Latency, Scalability), System Utilization, and key Sustainability and Accuracy indicators (such as Energy Efficiency and Power Consumption). Using a modular hybrid testbed, experiments reveal complex relationships among metrics, especially the trade-offs between accuracy and energy, and the efficiency of hybrid nodes. The guidelines help identify optimal operating points and lay the groundwork for improving orchestrators and schedulers (e.g., Kubernetes) to assign demanding applications, including AI and Quantum Computing, to suitable system modules, ensuring high performance and sustainability.
Read the full abstract
Game Theory
Comparison-Based Fair Division of Indivisible Chores
We investigate the query complexity of fairly allocating $m$ indivisible chores among $n$ agents with additive cost functions. We depart from the standard cardinal model and assume only comparison access: an algorithm may ask an agent which of two bundles is less costly, but never observes numerical costs. Our first results concern proportionality up to one item (PROP1). We design comparison-based algorithms that compute PROP1 allocations using $O(n^3\log m)$ comparison queries. When the chores are arranged in a fixed order and allocations are required to be contiguous, we compute a contiguous PROP1 allocation using $O(n^3 \log^2 m)$ comparison queries. Our main result concerns the maximin share (MMS) guarantee. We show that for any fixed number of agents $n$ and constant $\varepsilon>0$, a $\left(13/11 +\varepsilon\right)$-MMS allocation can be computed with a comparison complexity logarithmic in $m$. Remarkably, comparison access suffices to match the state-of-the-art $13/11$ cardinal-access guarantee of Huang and Segal-Halevi up to an arbitrarily small loss. Furthermore, our result implies that the MMS distortion of comparison access (i.e., the worst-case multiplicative loss in MMS fairness incurred by observing only comparisons rather than numerical costs) is at most $13/11$. Finally, we show that, for three agents, an allocation satisfying envy-freeness up to one item (EF1) can be computed using $O(\log m)$ comparison queries.
Read the full abstract
Computer Vision
CausalChapter: Improving Long-Video Chaptering with Interventional Dependency Modeling
Long-form instructional videos require automatic chaptering to support browsing, navigation, and knowledge access. Recent long-context language models can perform chaptering from textualized video inputs, but they remain costly and brittle for content-dense lecture videos with long transcripts, smooth topic transitions, and detailed chapter outputs. A scalable segment-then-caption paradigm reduces this cost, but introduces two new challenges: boundary error propagation and fragmented cross-chapter context. We propose CausalChapter, an intervention-inspired framework for long-video chaptering that estimates prediction-level influence through lightweight masking and removal interventions. For boundary localization, our Local Dependency Shift module detects drops in predictive dependency between adjacent temporal windows; for chapter description generation, our Cross-Segment Support Selection module reranks historical contexts according to their support for the current prediction. Experiments on long-video chaptering benchmarks show that CausalChapter improves boundary localization, chapter description quality, and cross-chapter coherence.
Read the full abstract
Machine Learning
HOPE: Heterophily-Aware Open-Set Node Classification with Pseudo-Extrapolation
Standard open-set node classification methods rely on the homophily assumption, where connected nodes share labels. However, real-world graphs are often heterophilic, exposing the limitations of current methods and posing new challenges to open-set node classification. On the one hand, cross-class connectivity causes representations from different known or unknown classes to become intertwined after aggregation, undermining their discriminative capacity. On the other hand, structural mixture invalidates threshold-based open-set methods and cross-class feature interpolation, leading to unreliable unknown-class rejection. To address these challenges, we propose HOPE, a Heterophily-aware Open-set node classification method with Pseudo-Extrapolation. To adapt open-set graph neural networks (GNNs) to heterophilic scenarios, HOPE uses a structure-augmented feature initialization layer to capture multi-hop structural patterns. Meanwhile, we design a trustworthy neighborhood aggregation mechanism for standard GNNs to dynamically filter noisy cross-class neighbors. To enhance unknown-class rejection, we introduce a heterophily-guided pseudo-extrapolation strategy. It dynamically maintains known-class centers and extrapolates along cross-class neighborhood displacement directions, synthesizing pseudo-unknown proxies near structurally ambiguous regions. Finally, we optimize the network with joint classification and logit margin regularization, routing synthetic proxies into a dedicated rejection slot without imposing geometric margin constraints in the representation space. Extensive experiments on multiple datasets show that HOPE consistently outperforms state-of-the-art models, validating its effectiveness, robustness, and efficiency.
Read the full abstract
Representation Theory
From Pre-triangulation to Triangulation: Obstructions and Exact Lifting
We give a new equivalent formulation of Verdier's octahedral axiom. An initial square in a pre-triangulated category determines an obstruction in a quotient of ordinary Hom groups. Its vanishing is equivalent to the existence of a good completion, and $\mathrm{TR4}$ is equivalent to this vanishing for the squares associated with composable morphisms. As a consequence, if two pre-triangulations have the same underlying additive category and one is triangulated, then the other satisfies $\mathrm{TR4}$ precisely when their relative inverse Heller comparison admits an exact lift along every short exact sequence in the Freyd category. As applications, we prove that the exotic pre-triangulated category of Díaz Cabrera–Muro is in fact a triangulated category over every algebraically closed field. We construct scalar families for type-$A_{3a-1}$ preprojective algebras over arbitrary fields and for a range of Dynkin preprojective algebras in characteristic two; in each family the zero parameter is the unique triangulated member. The local equations also yield separable descent, including canonical triangulations for non-modular equivariantizations and for the Markman–Mehrotra K3 deformation categories.
Read the full abstract
Machine Learning
Neither Adversarial Training Nor Purification: Emergent Adversarial Robustness from Oscillatory Predictive Learning
Adversarial robustness in computer vision is still largely achieved through adversarial training or test-time adversarial purification, both of which introduce significant computational overhead by generating adversarial examples during training or performing iterative denoising at test time. We study whether empirical robustness can instead emerge from architectural and representation-learning inductive biases. We introduce Oscillatory Predictive Learning (OPL), a two-stage framework that combines Artificial Kuramoto Oscillatory Neurons (AKOrN) with predictive self-supervised pretraining using X-PhiNet. Because our default checkpoint uses randomized initial oscillator states, we compare it with other randomized adversarial defense methods that provide precise, reproducible, and strong attack protocols. Experiments on CIFAR-10 and CIFAR-100, with additional corruption evaluation on CIFAR-10-C, demonstrate that our method achieves competitive results under the AutoAttack-rand evaluation protocol. On CIFAR-10 and CIFAR-100, OPL attains 76.63$\pm$0.76$\%$ and 50.44$\%$ robust accuracy, respectively, under $\ell_\infty$, $\epsilon=8/255$, AutoAttack-rand with EoT $K=20$.
Read the full abstract
Hardware Architecture
HDA-MoE: Hybrid Parallelism and Dynamic, Adaptive Scheduling for Mixture-of-Experts with 3D Near-Memory Processing
Mixture-of-Experts (MoE) architectures have become a key technique for scaling Large Language Models (LLMs), enabling high model capacity with reduced computational cost. However, this efficiency comes at the expense of increased memory capacity and bandwidth demands. Recent 3D Near-Memory Processing (NMP) architectures, which vertically integrate memory and compute through hybrid bonding, provide high internal bandwidth and energy efficiency, making them attractive for accelerating MoE inference. Nevertheless, the distributed memory and compute organization of NMP systems introduces new challenges for mapping MoE workloads. Existing parallelization strategies, such as Tensor Parallelism (TP) and Expert Parallelism (EP), suffer from either high communication costs or unbalanced computation utilization, leading to inferior efficiency. In addition, the dynamic routing behavior of MoE models further complicates efficient deployment. To address these challenges, we present HDA-MoE, a framework that optimizes MoE execution on NMP architectures through hybrid parallel deployment and runtime scheduling. HDA-MoE integrates an offline hybrid parallel mapping algorithm with an online dynamic and adaptive scheduling mechanism to reduce communication overhead while improving computation utilization. Experimental results show that HDA-MoE achieves a speedup of 1.1x–3.4x over TP, 1.1x–1.5x over EP, 1.1x–3.7x over the Hybrid TP-EP compute-balanced baseline, and 1.1x–1.3x over HD-MoE. Source code is available at https://github.com/PKU-SEC-Lab/HDA-MoE-TCAD26.
Read the full abstract
Software Engineering
Beyond Fixed Fault Models: Comparing LLM-Based and Rule-Based Fault Injection in OpenStack
Software Fault Injection (SFI) supports testing of cloud systems by introducing software defects and observing their manifestation. Rule-based injectors such as ProFIPy provide controlled and reproducible source-level mutations but require fault patterns to be encoded manually. Large Language Models (LLMs) offer a data-driven alternative by generating context-dependent software faults. We compare two code LLMs, Qwen2.5-Coder and DeepSeek-Coder, with ProFIPy in OpenStack's Nova and Cinder services. On shared injection targets, activation and observable-failure rates are comparable, but operational profiles differ: LLM-generated faults produce more Catastrophic outcomes on Nova, whereas ProFIPy produces more Silent and Multi-component effects. The sampled LLM outputs also differ in how they manifest failure, while showing greater agreement in their propagation scope. These findings show that LLM-based fault injection extends the behavioral coverage of fixed fault models without establishing general superiority, and that practical adoption still requires controlled generation, runtime validation, system-level oracles, and reproducible experimental provenance.