arXiv:2605.29020v2 Announce Type: replace-cross Abstract: Stiff and dense fibrin networks in chronic blood clots impede drug penetration, limiting the efficacy of thrombolytic therapies. Acoustic cavitation of microbubbles is a promising strategy to enhance drug delivery in soft tissues. However, the interaction of these bubbles with stiff fibrin networks has yet to be investigated. Here, we show that ultrasound-driven bubbles undergoing periodic oscillations can penetrate and alter dense fibrin networks. The penetrated bubbles create three-dimensional paths that enable nanobeads (matrix transport markers) to infiltrate up to 200 \textmu m deep into the mesh. Radial bubble oscillation is found to be the dominant forcing mechanism on fibrin fibers. Combining mechanical measurements with these observations reveals that the radial stress from a single bubble oscillation is far below the fracture strength of fibrin fibers. Instead, repeated sub-fracture loading from thousands of oscillations progressively accumulates damage and softens the network until it yields - the plausible mechanism for bubble penetration through the fiber mesh. We further explored this fibrin softening at a range of peak applied forces. At low force, the fibrin network initially softens, but is resistant to further damage after hundreds of cycles. At higher forces, networks continue to soften without reaching a stable state, indicating the progressive accumulation of damage. These results show that cavitation can enhance matrix transport in dense fiber mesh by softening and structurally altering fibrin networks. The underlying physics is governed by the viscoplastic mechanics of bubble-fibrin interactions. These findings establish a mechanistic framework to design comprehensive treatment strategies for fibrotic aged clots.
Science Journals
arXiv:2606.16219v1 Announce Type: new Abstract: Digital twin modeling, including control and data assimilation under model uncertainty, often faces an open-ended fidelity problem: adding variables, data streams, and time scales can indefinitely increase model complexity, ultimately producing systems that are difficult to maintain, validate, interpret, and use for stress or safety testing. As an alternative, one can seek parsimonious stochastic surrogate models built only on the variables needed to describe the relevant quantities of interest. We introduce a framework for discovering such variables from observational data by identifying which candidate inputs influence the full conditional law of a target quantity, rather than only its conditional mean. This distinction is essential in stochastic, coarse-grained, or partially observed systems, where dependencies may appear through changes in variability, tail behavior, multimodality, or uncertainty rather than through deterministic functional relationships. The framework couples conditional generative modeling, which learns the conditional distribution of the target given candidate inputs, with Gaussian-process-based analysis of variance (through kernel mode decomposition), which enables iterative pruning of non-influential inputs and interpretable structure discovery. In control settings, the resulting surrogate can be interpreted as a learned Markov decision process: the method identifies not only a transition model, but also the state, action, and memory variables needed to make the learned dynamics effectively Markovian. Across examples involving stochastic dynamical systems, missing variables, PDE control, reinforcement learning, and economic data, the discovered structures yield interpretable stochastic surrogates whose downstream performance is comparable to models trained on the full variable set.
arXiv:2606.14737v1 Announce Type: cross Abstract: Molecular dynamics (MD) simulations generate trajectories in a high-dimensional configuration space whose analysis critically depends on molecular descriptors, typically handcrafted observables or learned kinetic embeddings. Designing descriptors that are both expressive and broadly applicable, however, remains challenging. We study persistent homology (PH) as a general-purpose representation for MD and introduce the masked Flood complex, a protein-tailored modification of a recently introduced simplicial complex construction that emphasizes inter-residue structure at low computational cost. Vectorized persistence diagrams then provide information-rich, geometry-aware summaries of protein conformations, which we evaluate on protein class prediction, frame-level observable regression, and Markov state model (MSM) estimation from learned low-dimensional coordinates in a single shared representation space. Results on the mdCATH dataset show that PH-based descriptors are competitive across tasks, with masked Flood PH yielding the most consistent overall performance. Further, when using topologically-informed MSMs as a drop-in replacement within the recent MarS-FM framework for generative modeling of protein conformations, we obtain consistently better ensemble statistics than MSMs based on physical observables. Finally, we explore the transferability of the generative model to qualitatively different, fast folding, proteins.
arXiv:2407.11163v3 Announce Type: replace Abstract: In this paper, we propose a family of label recovery problems on weighted Euclidean random graphs. The vertices of a graph are embedded in $\mathbb{R}^d$ according to a Poisson point process, and are assigned to a discrete community label. Our goal is to infer the vertex labels, given edge weights whose distributions depend on the vertex labels as well as their geometric positions. Our general model provides a geometric extension of popular graph and matrix problems, including submatrix localization and $\mathbb{Z}_2$-synchronization, and includes the Geometric Stochastic Block Model (proposed by Sankararaman and Baccelli) as a special case. We study the fundamental limits of exact recovery of the vertex labels. Under a mild distinctness of distributions assumption, we determine the information-theoretic threshold for exact label recovery, in terms of a Chernoff-Hellinger divergence criterion. Impossibility of recovery below the threshold is proven by a unified analysis using a Cram\'er lower bound. Achievability above the threshold is proven via an efficient two-phase algorithm, where the first phase computes an almost-exact labeling through a local propagation scheme, while the second phase refines the labels. The information-theoretic threshold is dictated by the performance of the so-called genie estimator, which decodes the label of a single vertex given all the other labels. This shows that our proposed models exhibit the local-to-global amplification phenomenon.
arXiv:2510.16882v4 Announce Type: replace Abstract: Supervised fine-tuning (SFT) is a commonly used technique to adapt large language models (LLMs) to downstream tasks. In practice, SFT on a full dataset is computationally expensive and sometimes suffers from overfitting or bias amplification. This facilitates the rise of data curation in SFT, which prioritizes the most valuable data to optimze. This work studies the online batch selection family that dynamically scores and filters samples during the training process. However, existing popular methods often (i) rely merely on the utility of data to select a subset while neglecting other crucial factors like diversity, (ii) rely on external resources such as reference models or validation sets, and (iii) incur extra training time over full-dataset training. To address these limitations, this work develops UDS (Utility-Diversity Sampling), a framework for efficient online batch selection in SFT. UDS leverages the nuclear norm of the logits matrix to capture both data utility and intra-sample diversity, while estimating inter-sample diversity through efficient low-dimensional embedding comparisons with a lightweight memory buffer of historical samples. Such a design eliminates the need for external resources and unnecessary backpropagation, securing computational efficiency. Experiments on multiple benchmarks demonstrate that UDS consistently outperforms state-of-the-art online batch selection methods under varying data budgets, and significantly reduces training time compared to full-dataset fine-tuning. Code is available at https://github.com/gfyddha/UDS.
arXiv:2606.04621v2 Announce Type: replace Abstract: We present MeshFlow, a new method for generating artist-like 3D meshes. Current mesh generators often adopt Auto-Regressive (AR) next-token prediction, a natural choice given the discrete nature of mesh topology. However, AR methods scale poorly because the inference cost is quadratic in mesh size. They also require discretizing the vertex coordinates, which introduces quantization errors. To address these challenges, we introduce a Variational Autoencoder (VAE) that, supervised with a contrastive loss, represents both continuous vertex positions and discrete connectivity in a continuous latent space. This latent space is significantly more compact than prior token-based mesh representations. We then build a 3D generator based on a Rectified Flow transformer, generating all mesh vertices and edges in parallel. Our model generates meshes 18x faster than the fastest AR generator while also achieving excellent accuracy across standard mesh-generation metrics. Homepage: https://mesh-flow.github.io/, Code: https://github.com/facebookresearch/meshflow
arXiv:2606.14708v1 Announce Type: cross Abstract: Data-driven machine learning approaches have become increasingly attractive for nonlinear system identification, but standard models often fail to preserve the underlying physical structure and remain difficult to interpret, especially when no analytical model is available. In this context, port-Hamiltonian (pH) models provide a natural physics-informed representation. However, when these models are parameterized with standard multilayer perceptrons (MLPs), the learned constitutive components often remain poorly interpretable. In this paper, we propose a structure-preserving identification framework for nonlinear port-Hamiltonian systems based on Kolmogorov-Arnold Networks (KANs). The proposed PH-KAN model parameterizes the interconnection matrix, dissipation matrix, Hamiltonian, and input mapping using dedicated KAN blocks, while enforcing the port-Hamiltonian constraints by construction. This yields constitutive representations in which the nonlinear functions defining the identified pH components can be explicitly inspected, leading to a more interpretable model than with standard MLP-based parameterizations.
arXiv:2606.05742v2 Announce Type: replace Abstract: Speculative decoding accelerates generation by verifying multiple drafted tokens in a single target-model forward pass, reducing sequential decoding iterations. Model-free variants avoid auxiliary draft models by reusing text and model states already available during generation, but their speedup depends on the reliability of the constructed drafts. We identify two limitations of existing reuse-based methods: lexically anchored retrieval has limited recall under surface-form variation, and deterministic span copying can be brittle when the retrieved context does not uniquely determine the continuation. We propose \emph{AdaPLD}, a training-free method that adaptively improves both retrieval and draft construction. AdaPLD preserves high-precision lexical reuse while using semantic similarity to recover additional reuse opportunities when lexical matching fails. It further constructs branched reuse hypotheses to account for continuation uncertainty, rather than relying on a single copied span. Across diverse benchmarks, AdaPLD reduces target-model forward passes and achieves up to $3.10\times$ decoding speedup.
arXiv:2606.14910v1 Announce Type: cross Abstract: In this two-part paper, we investigate synthetic aperture radar (SAR) moving target imaging using planar antenna arrays. For a target moving over a three-dimensional terrain, its accurate localization requires the joint estimation of the motion-induced cross-range shift and the target height. In Part I of this two-part paper, starting from the planar array imaging geometry and the corresponding signal model, we show that these two quantities can be unified into a two-dimensional parameter vector and represented, after two-dimensional discrete Fourier transform (2D-DFT) processing across the planar array, through a natural vector remainder formulation. We first develop a general 2D-DFT matrix modulus framework and show that, in the two-dimensional setting, the associated 2D-DFT matrix modulus affects the propagation of vector remainder errors. Under a fixed array geometry and antenna number constraint, we derive an optimal construction of this matrix modulus and adopt it in the subsequent analysis. Under this construction, a single planar array provides only a folded estimate when the true parameter vector lies outside its unambiguous range. To resolve this ambiguity, we develop a multi-subarray framework in which multiple planar subarrays generate multiple vector remainders with different matrix moduli, and the desired parameter vector is recovered through the multidimensional Chinese remainder theorem (MD-CRT). To account for practical errors introduced by 2D-DFT quantization and additive noise, we further introduce an approximate 2D-DFT peak model for non-integer frequency vectors, incorporate robust MD-CRT, and establish sufficient conditions together with explicit reconstruction error bounds for both noiseless and noisy settings. Numerical results verify that the proposed multi-subarray framework enlarges the unambiguous range compared with a single planar array.
arXiv:2606.04678v2 Announce Type: replace Abstract: End-to-end ASR systems typically use fixed-depth acoustic encoders at inference, making it difficult to trade additional test-time computation for improved recognition without training a larger model. A natural approach is to reuse a shared Transformer block recurrently, but we find that naive looping does not fully exploit additional recurrent compute. We introduce LARM, a depth-conditioned looped Transformer that turns recurrent encoder depth into a controllable test-time compute axis. LARM combines sparse CTC checkpoints, supervision-clock embeddings, FiLM depth conditioning, and delayed soft-posterior feedback. These components structure the loop into recognition checkpoints separated by latent refinement phases and allow shared weights to specialize across recurrent steps. On LibriSpeech, LARM improves WER as the number of inference loops increases and achieves performance competitive with deeper unshared-parameter baselines. Our results show that test-time compute scaling can extend beyond autoregressive language-model reasoning to continuous non-autoregressive speech recognition.
arXiv:2606.14750v1 Announce Type: cross Abstract: Recent advances in pixel-based text modeling show that representing text as images enables models to exploit visual cues for language understanding. Grounding text in its visual form allows structurally similar characters with different Unicode encodings to produce similar embeddings, benefiting cross-lingual and zero-shot scenarios. Conventional text-based approaches treat each character independently, limiting generalization to unseen characters and requiring embedding expansion during cross-lingual adaptation. We propose Pixel-TTS, the first framework for visually grounded speech synthesis. It renders text as images and projects them through a 2D convolutional layer to generate embeddings. This design eliminates embedding matrix expansion during fine-tuning while improving robustness to unseen characters and orthographic variations. Extensive experiments show Pixel-TTS achieves competitive performance with strong baselines, faster convergence and robust zero-shot generalization.
arXiv:2606.16231v1 Announce Type: new Abstract: High-performance CUDA kernels are essential for scalable AI systems, while Large Language Models (LLMs) still struggle to generate correct kernels due to strict and implicit execution constraints. Existing LLM-based approaches either rely on costly agentic or reinforcement-learning (RL) pipelines, or adopt supervised fine-tuning (SFT) objectives that fail to explicitly model CUDA sensitivity, namely code tokens or regions tightly coupled with execution constraints. In this work, we investigate CUDA sensitivity from the perspective of token confidence patterns, showing that CUDA sensitivity appears at both token and region levels, where most CUDA-sensitive tokens are predicted with high confidence, while a smaller low-confidence subset forms regions corresponding to execution-critical structures. These findings suggest that effective CUDA kernel generation should both leverage high-confidence CUDA-sensitive tokens and preserve low-confidence CUDA-sensitive regions. Building on these insights, we propose \textbf{\underline{CU}DA-\underline{Se}nsitive Instruction \underline{T}uning (CuSeT)}, a low-cost post-training method within a simple SFT framework. CuSeT follows the principle of ``from tokens to regions'' by combining \emph{adaptive token-level masking} with \emph{region-aware sample reweighting}. Experiments show that CuSeT consistently improves functional correctness across multiple model families and scales, outperforming standard SFT and advanced SFT variants, while achieving competitive performance against frontier CUDA kernel generation models with substantially lower inference cost.
arXiv:2606.15517v1 Announce Type: new Abstract: Large language models often struggle with sensitive prompts. They may refuse outright, provide generic safety boilerplate, or fail to address the user's legitimate informational needs that can be answered safely. We introduce SHARD, a self-reframing distillation method to improve safe-helpfulness. It first rewrites sensitive prompts to surface benign intent using philosophical guidelines, then reframes its original responses into safe, more helpful ones, and finally fine-tunes the model on its self-reframed responses. Across DNA and the English subset of LINGUASAFE, SHARD improves helpfulness for most model families while preserving safety. It also remains competitive with distillation from a larger teacher model, suggesting that models can internalize safe and helpful behavior elicited from their own. Warning: This paper contains content that may be offensive or harmful.
arXiv:2606.15519v1 Announce Type: new Abstract: Adjoint formulations of magnetohydrodynamic (MHD) equilibrium are presented for sensitivity analysis and optimization-based equilibrium calculations. We derive the adjoint of the three-dimensional equilibrium equations under fixed-boundary conditions and analyse its axisymmetric reduction. The reduced operator coincides with the perturbed Grad-Shafranov operator, thereby making explicit the connection between the threedimensional adjoint construction and the axisymmetric adjoint problem. The resulting framework provides a basis for relating adjoint structures in both axisymmetric and fully three-dimensional ideal MHD equilibrium.
arXiv:2606.15213v1 Announce Type: cross Abstract: Time series forecasting largely benefits from combining the strengths of different models, especially using a scheme where a model corrects another model by capturing supplementary patterns from forecasting errors. Concurrently, quantum models are providing a means to augment the classical capacity, including in time series forecasting, by acting alongside classical models in hybrid architectures. In this work, we propose the first forecasting system based on error correction that jointly uses quantum and classical models. Here, quantum models first extract patterns by exploring quantum phenomena, and classical models capture the remaining patterns from the quantum errors. Compared to classical single models and classical-classical hybrid models based on error correction, the complementary capacity that emerges from this quantum-classical system provided the best results in most of the addressed problems. Therefore, this work paves the way to introduce quantum models in established hybridization schemes for time series forecasting.
arXiv:2606.16656v1 Announce Type: new Abstract: We study stochastic linear bandits with delayed feedback under several delay models and establish near-optimal regret guarantees. Our results identify when delayed linear bandits exhibit the same qualitative behavior as multi-armed bandits (MAB), and when the linear structure creates fundamentally new challenges. Specifically, (1) for \emph{loss-independent delays}, where the delay does not depend on the realized loss (but potentially depends on the arm), we show that delays incur only an additive regret penalty. Under stochastic delays, this penalty scales with the expected delay, while under adversarial delays, it scales with the maximum number of outstanding observations. Notably, both delay penalties are dimension-free, improving upon the state-of-the-art results; (2) for \emph{loss-dependent delays}, we show that linear bandits are substantially harder than MAB: unlike in MAB, we prove matching (up to log factors) upper and lower bounds in linear bandits, whose delay penalty depends on the square root of the dimension. (3) for the \emph{delay-as-payoff model}, a special case of loss-dependent delay, we show that the optimal MAB guarantee, which depends only on the delay of the optimal arm, is also unattainable in linear bandits. Together, these results provide a sharp characterization of how delayed feedback interacts with linear generalization.
arXiv:2606.16678v1 Announce Type: new Abstract: We introduce a mean-field-like computational model for calculating ionization potentials (IPs) based on the pair Coupled Cluster Doubles (pCCD) wave function. Specifically, our model combines the extended Koopmans' theorem (EKT) with the advantages of a variationally orbital-optimized (oo)-pCCD ansatz. The computational cost of the EKT(pCCD) method is negligible (O(N^3)) as the response 1- and 2-particle reduced density matrices used to construct the generalized Fock matrix are readily available after an oo-pCCD calculation. We benchmarked our new computational model for IPs of atoms, small molecules, and a set of organic acceptor molecules against experimental and theoretical reference data. The EKT(pCCD) model significantly improves upon the modified Koopmans' approach [J. Chem. Phys. 162, 184110 (2025)], and the obtained IPs are comparable to those of computationally more expensive IP-EOM-pCCD-based models, approaching CCSD(T) reference values (with a mean error of 0.05 eV). Most importantly, the EKT(pCCD) approach is almost independent of the basis set size, and reliable IPs are already obtained with small basis sets.
arXiv:2511.18018v2 Announce Type: replace Abstract: Subglacial blisters form due to the rapid drainage of supraglacial lakes into grounded ice sheets, and are characterised by elastic ice uplift and transient ice-velocity anomalies. Although blister occurrence is confirmed by observations, the dynamics of blisters and their impacts on ice flow remain poorly represented in current subglacial hydrology models, as typical cavity-channel system models cannot capture short-timescale blister formation, propagation, and relaxation. Here we present a unified, self-consistent modelling framework that directly couples blister evolution with the subglacial drainage system, extending existing subglacial hydrology models to account for transient responses to rapid lake drainage events. Numerical simulations, motivated by field observations of wintertime lake drainages, reveal distinct seasonal behavior: during summer, lake drainage generates short-lived blisters that rapidly leak water into a pre-existing drainage system of efficient, channelised water pathways, whereas winter drainage results in persistent blisters that propagate slowly and serve as the primary meltwater pathway at the ice-bed interface. The dynamics of blister propagation and leakage in our model are governed by effective viscosity and a characteristic leakage length scale, which reflects the connection between the blister and the surrounding hydrological network. This unified model offers a valuable tool for investigating blister dynamics and their interplay with subglacial hydrology, facilitating the interpretation of observed surface uplift and ice-velocity variations following supraglacial lake drainage events.
Human-Enhanced Loop Modeling (HELM): Agent-Based Finite Element Modeling of Concrete Bridge Barriers
arXiv:2606.12025v3 Announce Type: replace Abstract: Finite element (FE) modeling of safety-critical infrastructure such as bridge barriers requires high-fidelity nonlinear dynamic analysis, yet the current FE modeling process remains labor-intensive and lacks automation. This paper presents the Human-Enhanced Loop Modeling (HELM) framework, a collaborative human-agent protocol that decomposes long-sequence finite element modeling into discrete, visually verifiable checkpoints across geometry generation, boundary condition definition, and material assignment. The framework is demonstrated through a 20-case matrix of reinforced concrete bridge barriers under MASH TL-4 and TL-5 lateral loading conditions, interfacing specialized agents with two widely used commercial FE softwares, i.e., ANSYS and LS-PrePost. Experimental results show that HELM improves the baseline autonomous modeling success rate from 20% to 75%, with agent-level pass rates for geometry and boundary condition tasks approximately doubling. Error analysis reveals that spatial reasoning and algebraic logic limitations constitute the primary failure modes, underscoring the value of structured human-in-the-loop intervention for modeling automation. The complete agent design code and prompts are open-sourced and can be accessed at: https://github.com/SimAgentDev/Ansys-LSPP-AgentKit.
arXiv:2606.15520v1 Announce Type: new Abstract: We describe a Lean 4 formalization of the algorithms and domain types from NYU Computer Science Technical Report \#232, \emph{An ICON Package for Experimenting with Euclidean Domains} (Ericson, 1986). The original system implemented Lipson's catalog of procedures over integers, rationals, modular rings, polynomial rings, and truncated power series via a custom runtime dispatch mechanism in Icon. The present work separates three concerns: mathematical definitions grounded in Mathlib's \texttt{EuclideanDomain} hierarchy, computable mirrors suitable for evaluation and regression testing, and report-formatting infrastructure that reproduces the 1986 benchmark output line-for-line. All fourteen application algorithms from Section 3 of the report are defined and typecheck without \texttt{sorry}; those grounded in Mathlib -- chiefly integer gcd and extended Euclid -- additionally carry machine-checked proofs. We classify each procedure by its epistemic status relative to Mathlib, enumerate the coherence obligations between the proof and computable layers, and state precisely what is theorem-backed versus regression-trusted. The formalization makes explicit the verification boundary that the 1986 package crossed only informally.
arXiv:2606.08174v2 Announce Type: replace Abstract: Array superdirectivity is traditionally derived through singular optimization of densely spaced antenna arrays. In this paper, we show that the phenomenon admits a geometric interpretation as a concentration effect induced by spectral collision. As the spacing of an $M$-element linear array tends to zero, the exponential family generated by a linear array undergoes a spectral collision, and the associated finite-dimensional subspaces converge in reproducing kernel to a polynomial jet space. The maximum achievable array gain equals the diagonal evaluation of the reproducing kernel, and is therefore governed by the reciprocal Christoffel function. For the classical flat $L^2([-1,1])$ geometry, the Christoffel--Darboux kernel exhibits boundary concentration, yielding the quadratic $M^2$ superdirective law as a direct consequence of kernel asymptotics. This viewpoint separates intrinsic gain limits from numerical conditioning and identifies superdirectivity as a manifestation of a more general concentration mechanism. The framework further shows that the classical $M^2$ scaling is not universal: alternative spectral geometries produce different concentration laws through their associated Christoffel asymptotics. The results establish a direct connection between superdirectivity, reproducing kernels, orthogonal polynomials, and concentration phenomena arising from singular spectral limits.
arXiv:2410.21803v3 Announce Type: replace Abstract: Emergent Communication (EmCom) investigates how agents develop symbolic communication through interaction without predefined language. Recent frameworks, such as the Metropolis--Hastings Naming Game (MHNG), formulate EmCom as the learning of shared external representations negotiated through interaction under joint attention, without explicit success or reward feedback. However, MHNG relies on sampling-based updates that suffer from high rejection rates in high-dimensional perceptual spaces, making the learning process sample-inefficient for complex visual datasets. In this work, we propose the SimSiam Naming Game (SSNG), a feedback-free EmCom framework that replaces sampling-based updates with a symmetric, self-supervised representation alignment objective between autonomous agents. Building on a variational inference--based probabilistic interpretation of self-supervised learning, SSNG formulates symbol emergence as an alignment process between agents' latent representations mediated by message exchange. To enable end-to-end gradient-based optimization, discrete symbolic messages are learned via a Gumbel--Softmax relaxation, preserving the discrete nature of communication while maintaining differentiability. Experiments on CIFAR-10 and ImageNet-100 show that the emergent messages learned by SSNG achieve substantially higher linear-probe classification accuracy than those produced by referential games, reconstruction games, and MHNG. These results indicate that self-supervised representation alignment provides an effective mechanism for feedback-free EmCom in multi-agent systems.
arXiv:2411.02908v2 Announce Type: replace Abstract: Scaling large language models (LLMs) demands extensive data and computing resources, which are traditionally constrained to data centers by the high-bandwidth requirements of distributed training. Low-bandwidth methods like federated learning (FL) could enable collaborative training of larger models across weakly-connected GPUs if they can effectively be used for pre-training. To achieve this, we introduce Photon, the first complete system for federated end-to-end LLM training, leveraging cross-silo FL for global-scale training with minimal communication overheads. Using Photon, we train the first federated family of decoder-only LLMs from scratch. We show that: (1) Photon can train model sizes up to 7B in a federated fashion while reaching an even better perplexity than centralized pre-training; (2) Photon model training time decreases with available compute, achieving a similar compute-time trade-off to centralized; and (3) Photon outperforms the wall-time of baseline distributed training methods by 35% via communicating 64x-512xless. Our proposal is robust to data heterogeneity and converges twice as fast as previous methods like DiLoCo. This surprising data efficiency stems from a unique approach combining small client batch sizes with extremely high learning rates, enabled by federated averaging's robustness to hyperparameters. Photon thus represents the first economical system for global internet-wide LLM pre-training.
arXiv:2606.12632v2 Announce Type: replace Abstract: Stochastic Galerkin discretizations of nonlinear hyperbolic conservation laws may lose hyperbolicity because the standard pseudospectral product is generally nonassociative, leading to non-commuting blocks in the flux Jacobian matrix. We develop a novel framework for constructing hyperbolicity-preserving stochastic Galerkin systems based on associative truncated products on polynomial spaces. In one stochastic dimension, we characterize associative truncated products through a single polynomial datum and identify examples with useful symmetry, positivity, and spectral properties, including collocation products and an associative symmetric product based on Gaussian quadrature nodes. We prove a consistency result showing that, under suitable projection-error assumptions, these products converge to the classical product as the polynomial degree grows. For systems with rational fluxes, we derive sufficient conditions under which the resulting stochastic Galerkin flux remains hyperbolic on the corresponding admissible set. Applications to the one-dimensional isothermal and compressible Euler equations show accurate statistical approximation and robust hyperbolicity preservation of the computed stochastic Galerkin states.
arXiv:2606.16684v1 Announce Type: new Abstract: Vibration-based bearing fault diagnosis requires resolving three interrelated measurement challenges, including the trade-off between global statistical feature efficiency and local transient signal fidelity, insufficient traceability of measurement features to underlying fault physics, and ineffective multi-source measurement information fusion across diagnostic scales. This paper presents a progressive physics-guided multi-scale vibration signal processing framework that addresses all three challenges within a unified diagnostic pipeline. An 81-dimensional measurement descriptor, derived from bearing kinematic theory and characteristic defect frequencies, establishes a physically traceable feature space enabling real-time fault screening at approximately 20 ms per sample. A fault-adaptive signal segmentation mechanism then directs analytical attention toward fault-relevant waveform regions guided by physics-based priors, without manual feature engineering. Structured fault mechanism knowledge is further encoded implicitly in model parameters during training, enabling autonomous multi-scale measurement fusion without external knowledge dependencies at inference. Validated on four public benchmark datasets under diverse operating conditions, the framework achieves 98.49% diagnostic accuracy with a 12.6-fold reduction in computational cost relative to signal-level baselines. Interpretability analysis confirms that diagnostic feature activations align with established bearing fault mechanics, supporting measurement traceability in safety-critical industrial systems.