Quantum Physics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Monday 31 August
Thermodynamics is operational in the sense that the very concept of thermal equilibrium depends crucially on the choice of observables, while the amount of extractable work is defined relative to the allowed operations. Here we show that isolated quantum many-body states admit a thermodynamic hierarchical structure of observables and allowed operations, where the same state can be thermal at one level of the hierarchy and athermal at another. Enlarging the set of allowed operations therefore renders such hidden athermality a potential resource for work extraction; the resulting work gain is bounded in terms of the difference between the entropy densities of the two levels. This entropy difference takes the form of the mutual-information density. We establish the bounds for three extensions of operational access: increased spatial resolution, increased duration of control, and nonlocal connectivity, which provide access to position-state Holevo information, correlations between neighboring regions and spatially nonlocal correlations, respectively. Thus the difference between entropies at different levels of the thermodynamic hierarchy governs the bound on the work gain associated with moving to a less restricted level.
Spectral Theory
On almost commuting matrices with respect to the normalized Hilbert–Schmidt norm
In this paper, we consider the Rosenthal – Halmos problem of almost commuting matrices with respect to the normalized Hilbert – Schmidt norm $\|\cdot\|_{2,d}=d^{-1/2}\|\cdot\|_2$. We show that if $X$ and $Y$ are self-adjoint matrices with $\|X\|\le 1$, $\|Y\|\le 1$, then there exist commuting self-adjoint matrices $X',Y'$ such that $\|X-X'\|_{2,d}+\|Y-Y'\|_{2,d}\le 5\|[X,Y]\|_{2,d}^{1/3}$, and $[X,X']=0$. Within these constraints, the exponent $1/3$ cannot be improved.
Read the full abstract
Trading and Market Microstructure
Metaorder modelling and identification from public data
Market-order flow in financial markets exhibits long-range correlations. This is a widely known stylised fact of financial markets. A popular hypothesis for this stylised fact comes from the Lillo-Mike-Farmer (LMF) order-splitting theory. However, quantitative tests of this theory have historically relied on proprietary datasets with trader identifiers, limiting reproducibility and cross-market validation. We investigate whether it can be recovered from anonymous public data using synthetic metaorder reconstruction. Using transaction and quote data for the largest 239 stocks by market capitalisation on the JSE as of 13 March 2026 with the data range being 1 January 2023 until 31 December 2025, we conduct a grid search over reconstruction parameters and evaluate each configuration against established metaorder stylised facts and the LMF relation. Configurations selected to minimise errors across the metaorder impact stylised facts reproduce the targeted aggregate properties but yield a poor LMF relation. Configurations selected to minimise the LMF discrepancy recover the relation by construction while retaining several broad impact features, although some stock-level execution and decay fits are weaker. These asymmetric results show that recovering aggregate impact stylised facts alone is insufficient to identify LMF-consistent order splitting, while the LMF-targeted result establishes compatibility within the reconstruction class rather than an independent test. The findings support consistency with, rather than direct validation of, the LMF theory using anonymous market data.
Read the full abstract
Group Theory
Commutator identities, Lie product identities, and Multiplicative Lie algebras
The aim of this paper is to give a formal meaning of universal commutator identities and to see as to how the structure of multiplicative Lie algebra on a group is a potential candidate to describe universal commutator identities. We introduce some higher Schur multipliers. We shall also introduce and study Steinberg and Chevalley Multiplicative Lie algebras, the multiplicative Lie algebras with commutator identities among the root vectors describing Lie product relations. Free multiplicative Lie algebra on Steinberg and Chevalley groups are discussed.
Read the full abstract
Computer Vision
Multi-View Reflective Surface Inspection via Semantic-Saliency Cross-Verification
Reflective smartphone cover glass is challenging to inspect from a single fixed viewpoint because defect visibility varies with viewing geometry and specular reflections. This gives rise to two practical challenges: defects may be weakly observable from certain viewpoints, while the available visual evidence may remain spatially ambiguous. To address these issues, we propose a multi-view inspection framework in which each RGB observation is processed by a shared per-view expert. A vision-language model (VLM) produces class-aware semantic boxes, while a normal-reference reconstruction branch provides class-agnostic saliency. Their spatial agreement is used as supporting evidence to rank semantic proposals without modifying their coordinates or treating saliency as ground truth. The resulting evidence records are combined at product level without cross-view registration. On 282 production-line images, semantic-saliency association improves $AP_{50}$ from 52.6% to 62.6% by re-ranking fixed semantic proposals. Across 94 products, cross-view evidence recall $R_{\rm prod}@0.5$ increases from 75.5% for the best single view to 88.3% using all three views. These results support the complementary roles of semantic-saliency cross-verification and additional optical observations in reflective-surface inspection.
Read the full abstract
Computation and Language
Faithfulness Is Not Free: Auditing Offline KV-Cache Quantization in Retrieval-Augmented Generation
Retrieval-augmented generation systems can precompute and store key-value caches of retrieved documents to avoid re-encoding context at every query. Quantizing these caches further reduces storage, but no prior work asks whether compression damages faithfulness, whether responses remain grounded in the retrieved evidence. Faithfulness and accuracy are not equivalent: a model can produce a correct answer that is no longer supported by the context it was given. We evaluate Qwen2.5-7B-Instruct under INT8 and INT4 quantization on RGB and HotpotQA, measuring both accuracy and faithfulness with a hallucination detector, NLI entailment, and an LLM judge. INT8 is near-lossless across both metrics. INT4 reduces accuracy and, more critically, even among answers that remain factually correct, over 90% of faithfulness changes are negative, i.e., accuracy metrics are blind to this regression. The harm grows under noisy retrieval and with more retrieved chunks. Faithfulness must be audited before compressed caches are deployed.
Read the full abstract
General Relativity and Quantum Cosmology
Constraining equation of state of neutron star using neutron star-black hole mergers
In this work, we study tidal effects that can be observed in gravitational wave transient from neutron star-black hole (NSBH) mergers. It has been well known that neutron star's tidal deformation has imprint of equation of state and hence composition of neutron star matter. Compared to binary neutron star mergers, where we observe a combined effect of two neutron stars by measuring the parameter known as lambda tilde. In neutron star black hole mergers, we measure the tidal deformability of the neutron star component clearly. In this work, we study the feasibility of measuring the bare tidal deformability of a neutron star with planned gravitational detectors in the near and far future. We use the parameter estimation computation framework, Bilby is used to perform Bayesian inference for hundreds of NSBH mergers in current and future GW detector networks and obtain the posterior probability distribution functions of binary parameters. We find that we need at least 20 events to clearly distinguish the equation state of neutron stars in future detectors, including the Einstein Telescope and Cosmic Explorer.
Read the full abstract
Soft Condensed Matter
Intercoupling of Segregation and Rheology in Spatially Developing Granular Chute Flows
We investigate the flow and segregation of binary granular mixtures with density differences down a long chute using a continuum framework that couples a particle-force-based segregation model with an inertial-number-based local rheology. The steady-state momentum and convection-diffusion-segregation equations are solved simultaneously, explicitly accounting for the two-way coupling between segregation and flow. Predicted concentration and velocity fields at different streamwise locations are validated against representative DEM simulations. The validated model is then used to examine the influence of density ratio, mixture composition, and chute inclination on segregation over a wide range of conditions. The chute length required to achieve fully developed segregation is quantified and compared with the development length for monodisperse granular flow. At low density ratios and/or low inclinations, segregation develops over much longer distances than the velocity field. In contrast, at higher inclinations and larger density contrasts, the two length scales become comparable, demonstrating that neglecting flow development can significantly underestimate segregation evolution.
Read the full abstract
History and Overview of Mathematics
The Mysteries of Plato's Parmenides Dispersed: From musical intervals to contacts/"hapseis" to "logoi"
Plato's Parmenides has long been considered the most difficult and enigmatic of Platonic dialogues, marked by seemingly contradictory arguments. We present an interpretation showing that the One of the Second Hypothesis (142b-155e) is an intelligible Being whose structure constitutes the philosophical analogue of a dyad in periodic anthyphairesis. First, by linguistic analysis and Proclus' Platonic Theology, we show that the One and the Being form an indefinite dyad satisfying the philosophical analogue of infinite anthyphairesis. We analyze Plato's initial introduction and eventual rejection of eristic numbers formed with unequal units from the infinite sequence of remainders (143c-144c). Second, we reveal that the presence of the One in the Being (144c-d, 138a) is achieved via contacts ("hapseis", 148d-149d). Based on the Pythagorean musical rule relating terms and intervals, a contact is identified with the ratio between consecutive anthyphairetic remainders. The circularity of contacts (138a) expresses Theaetetus' Logos Criterion for anthyphairetic periodicity, defining dialectical numbers with equalized units and establishing the finiteness of the dialectical number of parts. This framework resolves key Platonic enigmas: Plato's Indivisible Line is identified with the intelligible Being; the Sophist's assertion that "not-Being is a Being" corresponds to the equalization of the dyad; and the compresence of opposite properties separates intelligibles from sensibles. Finally, the self-similar Oneness of an intelligible Being refutes Vlastos' Non-Identity principle, resolving the Third Man Argument.
Read the full abstract
Emerging Technologies
Trustless Accountable Data Sharing for Supply Chains: A Reference Architecture and MoreMedDiet Proof-of-Concept
Modern supply chains are increasingly dependent on the exchange of sensitive business data between heterogeneous stakeholders who do not trust each other. %under partial trust. Conventional integration approaches typically rely on centralized intermediaries or require extensive data disclosure, which hinders collaboration, compliance, and innovation. We propose a generic reference architecture for privacy preserving, trustless data sharing in supply chains. The architecture combines permissioned distributed ledgers, IoT-based evidence ingestion, and modular privacy mechanisms to enable controlled transparency: stakeholders can extract verifiable insights and support audits while minimizing data exposure. We instantiate the architecture in the context of the MoreMedDiet project, which promotes sustainable farm-to-fork food systems and adoption of the Mediterranean diet. By mapping the MoreMedDiet use case into our architecture, we demonstrate how channel topologies, IoT-to-ledger data flows, and smart contract-based quality assurance (QA) can support secure data exchange, transparency, interoperability, and regulatory alignment. Our proof-of-concept illustrates how Self-Sovereign Identity (SSI) and W3C Verifiable Credentials (VCs) can provide portable and auditable access governance over Hyperledger Fabric channels. The resulting design offers a reusable blueprint for multi-actor supply chains, including agri-food and pharmaceutical domains. Moreover, the performance achieved – presenting a verified credential to the next stakeholder in the supply chain in 130 ms – is promising and demonstrates the potential applicability of the approach in real-world supply chain environments.
Read the full abstract
Analysis of PDEs
A block-Toeplitz approach to the spectrum of the Neumann-Poincaré operator on self-similar chains
We characterise the spectrum of the Neumann-Poincaré operator acting on a mean-zero subspace of the Sobolev-Slobodeckij space $W^{s,p}(\Gamma)$, where $\Gamma$ is a non-Lipschitz domain, consisting of an infinite self-similar chain of disjoint smooth domains accumulating at a limit point. We exploit the discrete geometric scaling of the chain to obtain an exact block-Toeplitz representation of the Neumann-Poincaré operator, as well as the single- and double-layer operators. The corresponding operator-valued symbol depends entirely on the single scaling parameter $\alpha=(d-1)/p-s$ and belongs to the Wiener algebra whenever $0<\alpha<d$. This allows us to exploit block-Toeplitz operator theory to characterise the essential spectrum and Fredholm regions within this regime. In the energy space $H^{-1/2}_0(\Gamma)$, corresponding to $\alpha = d/2$, we prove the spectrum is real and characterise any isolated eigenvalues outside the essential spectrum through an operator-valued Wiener-Hopf factorisation. Furthermore, we apply an operator-valued Szegő limit theorem to derive the asymptotic spectral distribution for large finite truncations and establish an eigenvalue counting formula based on the operator symbol. We illustrate these results through explicit analytical computations for a chain of concentric annuli and numerical approximations for a chain of disks.
Read the full abstract
Differential Geometry
Convergence of symmetries: nilpotency, dimension, and perfectness
Let $(X_i,p_i)$ be a sequence of pointed $n$-dimensional Riemannian manifolds with a uniform lower Ricci curvature bound, and $G_i \leq \operatorname{Iso} (X_i)$ a sequence of closed groups of isometries. We show that if the triples $(X_i, G_i, p_i)$ converge in the equivariant Gromov–Hausdorff sense to a triple $(X,G,p)$, then $\mathfrak{g}$, the Lie algebra of $G$, admits an ideal $\mathfrak{h} \trianglelefteq \mathfrak{g}$ with $\operatorname{dim} (\mathfrak{h}) \leq \limsup_i \operatorname{dim} (G_i)$ and $\mathfrak{g}/ \mathfrak{h}$ nilpotent. Moreover, if the sequence $(X_i, p_i ) $ is non-collapsing, we show that $\mathfrak{h} $ can be taken of dimension $\limsup_i \operatorname{dim} (G_i)$.
Read the full abstract
High Energy Astrophysical Phenomena
The First Array-Wide Diffuse Flux Search for UHE Neutrinos with the Askaryan Radio Array
The Askaryan Radio Array (ARA) is an ultrahigh energy (UHE) neutrino detector at the South Pole that searches for impulsive broadband radio signals from neutrino-induced particle showers in glacial ice. ARA consists of five autonomous stations with receiving antennas deployed up to 200 m deep and has accumulated the largest livetime of any in-ice radio array. We present the first array-wide diffuse UHE neutrino search using data collected from January 2013 to December 2023, incorporating improved detector characterization and simulation, including data-driven noise and electronics models, revised antenna responses, and updated neutrino interaction and lepton propagation modeling, within a unified framework for event processing, background rejection, and cut optimization across independently operating stations. This search is expected either to identify the first UHE neutrino candidates observed by an in-ice radio detector or to set the most stringent UHE neutrino diffuse flux limits above a few EeV, while informing analysis strategies for under-construction and future radio arrays, such as RNO-G and IceCube-Gen2 Radio.
Read the full abstract
Instrumentation and Methods for Astrophysics
METIS high-contrast imaging simulations: From instrument modelling to science readiness
The Mid-infrared Extremely Large Telescope (ELT) Imager and Spectrograph (METIS) instrument, expected to see first light in early 2030, aims to detect and characterise exoplanets and circumstellar disks through high- contrast imaging (HCI) and spectroscopy. The High-contrast End-to-End Performance Simulator (HEEPS), initially developed to support the design of the METIS HCI modes, has evolved into a crucial tool for the METIS science team to prepare and optimize observations. HEEPS is an open-source Python-based software with a modular architecture, integrating the wavefront Fresnel propagation package PROPER, and HCI image processing with the Vortex Image Processing (VIP) package. Though designed for METIS, its modularity has been applied to other HCI instruments as well. This work presents recent updates to HEEPS, including modelling of the final METIS pupil and Lyot stops, revised quasi-static non-common path aberrations (NCPA) and Talbot effect simulations informed by as-built optical surface errors, and updated METIS Single Conjugated Adaptive Optics (SCAO) simulations. We also discuss advancements in NCPA control strategies focusing on framerate, latency and sensing performance optimization, particularly for mitigating water vapor seeing effects using the asymmetric Lyot wavefront sensor (ALF) algorithm. With these refinements, we present a comprehensive grid of HCI performance simulations for METIS, covering a range of magnitudes in the L, M, and N-bands, and several HCI observing modes. These simulations produce updated 5-sigma sensitivity contrast curves and mock HCI observations, providing key insights on HCI performance for instrument optimization and science observation planning. Our results underscore the key role of end-to-end simulations in bridging instrumental design and scientific readiness in the ELT era.
Read the full abstract
Computation and Language
Stick to What You Know: A Study of Knowledge-Aligned Supervised Fine-Tuning
Supervised fine-tuning (SFT) trains a base language model to imitate target responses, and these targets may require knowledge the base model has not robustly internalized. We study this as a source of hallucinations and frame a group of mitigation methods as knowledge-aligned SFT: constraining SFT training targets to the base model's parametric knowledge. Under a unified setup, we compare existing generation-based and estimation-based knowledge-alignment methods and introduce two new variants: Evidence Rewrite, which verifies base-model generations using external evidence, and Recall Rewrite, which retains claims only when they can be consistently recalled by the base model. Experiments with Qwen 3 4B and OLMo 3 7B show that knowledge-aligned SFT can reduce factual hallucinations on WildHalu and Biography while largely preserving general capabilities. Recall Rewrite yields the strongest factuality gains and improves refusal behavior on UnknownBench. It thereby confirms that SFT targets beyond the base model's knowledge drive hallucination behavior.
Read the full abstract
Machine Learning
Controlling Refusal Behavior of LLMs via Stiefel-Constrained Rotation Steering
Activation steering has emerged as a lightweight approach for controlling model refusal at inference time. A growing line of research explores trainable rotations of activations to develop geometrically principled intervention mechanisms. However, existing techniques rely on auxiliary constructs, such as refusal vectors, to define these rotations. In our work, we develop a self-contained methodology for learning parameter-efficient rotational transformations based on Riemannian optimization. We empirically validate the proposed scheme, demonstrating its superiority in intervention efficiency. An extensive ablation study highlights the importance of key design choices in our method. Our results identify the proposed rotation-based steering scheme as a promising direction for more reliable control over the behavior of LLMs.
Read the full abstract
Cosmology and Nongalactic Astrophysics
Resonant Dynamics of Gravitational Wave and Chiral Alfvén wave in the Early Universe
In this work, we investigate the resonant interaction between stochastic gravitational waves (GWs) and chiral Alfvén waves in a magnetized chiral plasma in the early Universe. Starting from covariant chiral magnetohydrodynamics coupled to linearized gravity, we derive a closed system of equations for the coupled chiral-Alfvén velocity and magnetic-field perturbations. Both analytics and numerics show a parametric resonance at the sum frequency of the two CVE-split branches, with an instability band that widens as the GW strain grows. This CVE-to-Alfvén ratio is not a free parameter. In the early Universe, it is fixed by the Standard Model's relativistic degrees of freedom. Including the plasma's backreaction on the GW makes the energy exchange nonlinear, breaking the usual single-frequency Manley-Rowe picture. For a self-consistent choice of parameters, where backreaction is only a small correction, the unsuppressed resonance still drives the system to a finite-time blow-up within a few Hubble times. We then work out the resonant frequencies and growth rates involved, mapping out what current and future gravitational-wave detectors could observe.
Read the full abstract
Signal Processing
Semantic-Aware Sub-Band Allocation for Terahertz Communications
This paper studies semantic-aware sub-band al- location for terahertz (THz) communication systems, where frequency-selective molecular absorption creates highly non- uniform sub-band qualities. Unlike conventional formulations, semantic fidelity depends nonlinearly on the signal-to-noise ratio (SNR) and is also sentence-specific, leading to a non-separable assignment problem that is generally not solvable using simple ordering-based policies. To address this, we use a sentence-BERT (SBERT)-based surrogate model that predicts semantic fidelity from the sentence embedding and sub-band SNR. We propose an importance-aware scheduler that assigns sentences to sub- bands based on their semantic contribution using an oracle utility function that captures importance-weighted semantic similarity across sentence-sub-band pairs. A neural scheduler is then trained through imitation learning to approximate the oracle policy at more than 200x lower runtime than full DeepSC- based oracle evaluation. Integrated with a deep-learning-enabled semantic communication (DeepSC) system, the proposed method consistently outperforms all benchmark schemes and approaches oracle-level performance under realistic THz channel conditions.
Read the full abstract
Robotics
Autonomously Acquiring Robot Manipulation Skills with Language-Driven Quality-Diversity
Quality-diversity (QD) algorithms have been gaining traction in robot learning, where diverse motion primitive libraries allow robots to adapt zero-shot to constraints at deployment time. However, such methods typically require expert designers to write the success condition, fitness and diversity metrics, and this strongly limits the robot's autonomy. On the other hand, existing LLM-based reward-shaping techniques allow robots to learn autonomously but only output single high-performing solutions, limiting the robot's adaptability. In this paper, we propose an approach designed to output diverse motion primitive archives by autonomously leveraging quality-diversity algorithms, only requiring a free-form description of the task in common language. To address the difficulty of designing relevant fitness and diversity metrics, we propose an autonomous exploration mechanism able to reliably output sets of functionals covering the fitness and behavior descriptor (BD) space. First, we pose policy exploration as a functional design problem, where the functional spaces are lower-dimensional than the full BD and fitness spaces, and propose an LLM-based exploration scheme to sample from these low-dimensional spaces without any task-specific prompts, fine-tuning or expert intervention. We adapt a multi-BD variant of the MAP-Elites success (MES) algorithm, designed to leverage the heterogeneous BD samples. Finally, through experiments based on the genesis simulator, we show that our method effectively generates archives of diverse motion primitives, outperforming classical QD algorithms with inferred and hand-written parametrizations on a set of $4$ robotic manipulation tasks.
Read the full abstract
Optimization and Control
Efficiency of cooperation incentives in evolutionary population games under payoff-observation errors
Traditional studies on evolutionary dynamics of cooperation have concentrated on an idealized game setup free of payoff-observation errors. However, in real-world scenarios, individuals frequently encounter errors when observing the payoffs of their opponents during game interactions, resulting from unintentional mistakes, such as data misrecording, overlooking critical details, and miscalculations. This gap between idealized error-free game models and the error-prone real-world game interactions leads to the lack of insight into the impact of payoff-observation errors on the evolutionary dynamics of population games, in particular the efficiency of cooperation incentives. In this paper, we construct a research framework for population games with payoff-observation errors, which enables us to investigate the effects of errors on the evolutionary dynamics of cooperation in the evolutionary Prisoner's Dilemma game with combined incentives. To quantify the implementation costs of incentives in the presence of errors, we devise an index function and employ optimal control theory to derive the optimal incentive protocols. Our theoretical and numerical results reveal that payoff-observation errors can lower the costs compared to error-free cases, and we also derive the theoretical conditions for these results. Finally, we formulate an optimization problem to explore the cost difference between the optimal incentive protocols with and without errors, and further design an algorithm to obtain the numerical solution that minimizes this difference.
Read the full abstract
Soft Condensed Matter
Particle-Mediated Tuning of Defect Stability in Lamellar Block Copolymer Systems
We study how colloidal inclusions modify the formation energy of dislocation pairs in lamellar block copolymer systems. Using a hybrid particle/Ginzburg–Landau model, we calculate defect formation energies by comparing defect-free and defect-containing states with and without embedded colloids. Finite-size scaling is used to obtain formation energies in the thermodynamic limit. The effect of colloid insertion depends strongly on particle sizes and surface patterning. Homogeneous particles increasingly stabilize dislocation pairs with increasing particle sizes. Particles larger than one lamellar domain preferentially occupy the dislocation cores, where they replace strained polymer rather than deforming defect-free lamellae. The magnitude of this stabilization depends on surface affinity. Balanced Janus particles instead increase the formation energy, because their competing surface preferences cannot be satisfied simultaneously near the curved core. Varying the patch ratio interpolates between these limits. Surface patterning has little effect for particles smaller than one lamellar domain but changes the formation energy by several tens of $k_BT$ for larger particles. These results provide quantitative guidelines for controlling topological defect stability in lamellar block copolymer systems.
Read the full abstract
Computation and Language
Evaluating and Improving LLM Self-Modeling
We study self-modeling: an LLM's ability to answer questions about its own behavior. We focus on verifiable behavioral questions, such as whether a prompt edit would change the model's final answer. To measure this capability, we introduce a benchmark that tests diverse types of self-modeling questions. Current models show non-trivial but limited self-modeling skill, and make systematic mistakes on simple counterfactual questions about their own behavior. To improve self-modeling skill, we develop a scalable synthetic-data pipeline that produces self-modeling training data, and show that reinforcement-learning can improve aggregate self-modeling skill across three open-source model families with some transfer to held-out tasks. These gains, however, do not seem to constitute introspection consistently: improved self-modeling may not arise from privileged access to the model's internal decision process.
Read the full abstract
Combinatorics
A $(q,t)$-Overview of $q$-Analogs
By a simple homogenization process, one may turn a $q$-analog into a symmetric $(q,t)$-analog. Although simple, this bijective correspondence makes many concepts and identities become more natural, and proofs follow readily from classical properties of symmetric functions in two variables. A more intricate non-homogeneous extension is also developed. We illustrate this approach, revisiting recent developments in the study of $\gamma$-positivity, Lucas analogues, and the monoid of Cyclotomic generating functions. This also leads naturally to new constructions and conjectures. We further explore other avenues of investigations, included graded and equivariant $\gamma$-positivity and $\gamma$-anti-positivity (also known as alternatingly $\gamma$-positive).
Read the full abstract
Machine Learning
Sparse Competition during Training For the Emergence of Specialized Modules
Modularity in deep neural networks has been proposed as a means of improving both interpretability and training by promoting disentangled representations and reducing redundancy. In this work, we study the emergence of modular structure through competition dynamics between groups of neurons during training. We introduce a method that (i) maintains near-baseline accuracy, (ii) induces usage-based modularity by sparsely routing inputs to neuron groups, and (iii) encourages specialization of these modules, such that their activations are correlated with input classes. We evaluate the proposed approach on ImageNet-100 and CIFAR-100 and show that with it, specialized modules emerge without module-level supervision. These modules capture a meaningful high-level structure in the data, with individual modules responding to semantic categories (e.g., dogs or vehicles). We also study the emergence of a hierarchical partition of sub-tasks depending on the number of modules. Our results suggest that competitive dynamics can serve as a simple mechanism for inducing functional modularity in standard architectures.
Read the full abstract
Systems and Control
Adaptive Observer of Nonlinear One-Sided Lipschitz Systems Using Estimated State Regressors With Finite Excitation
For systems with unknown parameters, finite excitation and concurrent learning can potentially yield parameter convergence without persistent excitation but the regressor may still depend on inaccessible states, leading to regressor mismatch. In this paper, this problem is addressed for a class of nonlinear systems with one-sided Lipschitz properties and quadratically inner-bounded nonlinearities with bounded disturbances and linearly parametrized uncertainties. To this aim, an output-integral regression is utilized by using measured outputs and estimated states, and history-stack residual is explicitly bounded in terms of state-estimation error and disturbance. Furthermore, a perturbation bound between the estimated-state and true-state information matrices is derived. Additionally, an OSL-QIB LMI condition is applied for the observer design and a projected adaptive law is designed without needing exact output matching. Stability analysis's results indicate the proposed observer and parameter estimation outperform observers without history-stack learning term.
Read the full abstract
Machine Learning
CastClaw: A Human-in-the-Loop Autonomous Agent for Interactive Time Series Forecasting
Practical time-series forecasting requires domain knowledge, constraints, plausibility checks, and human intervention. We present CastClaw, an interactive forecasting system centered on an autonomous forecasting loop and supported by a forecasting-oriented harness. Users create and refine tasks in natural language through command-line and Web interfaces. An internal large language model (LLM) agent reads the task and evidence, selects a permitted tool action or request for human feedback, and uses checked results to decide whether to continue, revisit an analysis, switch a candidate, or stop. The harness supplies tool integration, controlled execution, state management, historical validation, and execution limits. npm distribution connects installation and configuration to interactive execution and forecast and report export. Across five public electricity-price datasets, CastClaw achieves the lowest point-estimate MSE and MAE among 16 baselines. A Nord Pool case study illustrates its inspectable decision process, while provincial electricity-load data from North China (January–June 2026) demonstrate offline task execution.
Read the full abstract
Computer Vision
MR-JEPA: A General Purpose Video Foundation Model for Cardiac MRI
Cardiac magnetic resonance imaging (CMR) produces rich sequential data such as temporal cine videos and spatial LGE/mapping stacks, yet most deep learning approaches process individual 2D slices, discarding this context. We present MR-JEPA, a self-supervised video foundation model for CMR that extends LeJEPA to 3D spatiotemporal inputs through tubelet tokenization, spatiotemporal masking augmentation, and initialization from a 2D CMR foundation model. Unlike prior CMR video models limited to cine data, MR-JEPA is pretrained on multi-sequence data (cine, LGE, mapping) from 10,505 patients across two centers without annotations. We evaluate the frozen encoder on six downstream tasks using a unified multi-view gated attention architecture: LV ejection fraction, RV ejection fraction, three myocardial strains (GLS, GCS, GRS), and four-class disease detection. MR-JEPA outperforms other compared methods on all five regression tasks, including both a domain-specific CMR model pretrained on more data with text supervision and a natural-video foundation model, achieving an LV EF MAE of 4.79% (r =0.764) and a GLS MAE of 1.87 (r=0.805), with 21-27% MAE reductions over baselines on strain tasks. For disease detection, MR-JEPA achieved a macro AUG of 0.868, remaining competitive with the domain-specific baseline despite using a fully self-supervised pretraining objective. These results demonstrate the potential of a unified video encoder for robust, multi-view utilization of diverse CMR sequences in clinical cardiac quantification and diagnosis.
Read the full abstract
Sound
CoJEPA: Combining Contrastive Learning and JEPA for Global-Local Music Representations
Joint-Embedding Predictive Architecture (JEPA) has shown strong performance in learning rich representations through self-supervised prediction in latent space. However, it typically relies on teacher–student architecture with an EMA to stabilise training, and can tend to yield uninformative representations. Contrastive learning is stable to train and produces strong global representations, but remains limited on local tasks by the global nature of its objective. In this work, we combine both into CoJEPA: a single shared backbone jointly trained with a JEPA objective on masked sequence tokens and a contrastive objective on the class token. The contrastive gradient provides stability, removing the need for an EMA teacher entirely, while JEPA enriches the sequence tokens via local predictions that contrastive learning alone cannot provide. Crucially, no extra parameters are added to the backbone: the same model is guided towards richer representations purely through the design of its training signal. CoJEPA takes the best of both worlds, outperforming or matching both individual methods across global and local MIR tasks, with a particularly strong advantage on tonal and harmonic understanding, and without any task-specific architectural changes. CoJEPA shows that combining objectives with complementary inductive biases can substitute for scale, encouraging future work to invest in smarter training objectives over ever-larger models.
Read the full abstract
Geometric Topology
Pseudo-Anosov flow and dynamics on guts
We relate the dynamics on a closed 3-manifold to the topological invariant, homology guts, proposed by Agol and Zhang. Given a closed manifold $M$ that admits a pseudo-Anosov flow $\varphi$ without perfect fits, we construct a canonical semiflow on the homology guts associated to homology classes carried by $\varphi$. We show that, up to orbit equivalent outside half annuli, the semiflow on the guts is invariant on each Thurston open cone. As an application, the fundamental group and the sutured structure of homology guts encode closed orbits of $\varphi$ with particular type. Besides, for a positive class in $H^{1}(M)$, the canonical semiflow on the guts has a well-defined growth rate.
Read the full abstract
Earth and Planetary Astrophysics
Öpik-type collision frequency for Kozai-driven projectiles: Target bodies on eccentric and inclined orbits
For high-inclination projectiles undergoing Lidov–Kozai secular evolution, coupled variations in eccentricity and inclination complicate estimates of long-term collision frequencies with a target body. Previous collision-frequency studies for Kozai-driven projectiles have considered target-body orbits that are either circular or confined to the reference plane. The present study considers the general case in which the target body's orbit has both non-zero eccentricity and non-zero inclination. For each complete Lidov–Kozai cycle, the initial relative nodal longitude and the target body's initial argument of periapsis serve as two independent orientation variables. The single-cycle collision frequency is evaluated for each pair of initial values. Under the incommensurability condition considered here, the long-term mean is then obtained by explicitly averaging these values over both orientation variables. For the special cases in which the target body's orbit is circular or lies in the reference plane, the method is shown analytically to reduce to the corresponding formulations of previous studies. Numerical comparisons confirm these reductions. For the general eccentric and inclined case, the resulting collision frequency predicts a projectile survival curve that agrees well with the fraction of projectiles remaining in independent direct dynamical integrations. Finally, it is shown that, for a fixed dynamical setup and under the incommensurability condition, the long-term mean collision frequency associated with a given closed Hamiltonian level curve is independent of both the projectile's initial secular state along that level curve and its initial longitude of ascending node.