Tootfinder

Neural Scaling Laws for Deep Regression
Tilen Cadez, Kyoung-Min Kim
https://arxiv.org/abs/2509.10000 https://arxiv.org/pdf/2509.10000

Neural Scaling Laws for Deep Regression
Neural scaling laws--power-law relationships between generalization errors and characteristics of deep learning models--are vital tools for developing reliable models while managing limited resources. Although the success of large language models highlights the importance of these laws, their application to deep regression models remains largely unexplored. Here, we empirically investigate neural scaling laws in deep regression using a parameter estimation model for twisted van der Waals magnet…

OpenCodeReasoning-II: A Simple Test Time Scaling Approach via Self-Critique
Wasi Uddin Ahmad, Somshubra Majumdar, Aleksander Ficek, Sean Narenthiran, Mehrzad Samadi, Jocelyn Huang, Siddhartha Jain, Vahid Noroozi, Boris Ginsburg
https://arxiv.org/abs/2507.09075

OpenCodeReasoning-II: A Simple Test Time Scaling Approach via Self-Critique
Recent advancements in reasoning-based Large Language Models (LLMs), particularly their potential through test-time scaling, have created significant opportunities for distillation in code generation and critique. However, progress in both areas fundamentally depends on large-scale, high-quality datasets. In this work, we introduce OpenCodeReasoning-II, a dataset consists of 2.5M question-solution-critique triples (approx. 35K unique programming questions), making it nearly twice the size of th…

Phenomenological Scaling Relations for SQM Stars with a Massive s-Quark in Gravitationally Strong Magnetic Fields under the Spherical Symmetry Approximation
{\L}ukasz Bratek, Joanna Ja{\l}ocha, Marek Kutschera
https://arxiv.org/abs/2507.10756

Phenomenological Scaling Relations for SQM Stars with a Massive s-Quark in Gravitationally Strong Magnetic Fields under the Spherical Symmetry Approximation
Scaling relations with the bag constant parameter are investigated for Strange Quark Matter (SQM) stars in the presence of gravitationally strong magnetic fields minimally coupled with matter, considering both massless and massive strange quark scenarios. Assuming a simple model for such coupling under the approximation of spherical symmetry, a phenomenological scaling formula for the maximum mass of stars is derived as a function of the surface magnetic field and the strange quark mass. This f…

Scaling Up without Fading Out: Goal-Aware Sparse GNN for RL-based Generalized Planning
Sangwoo Jeon, Juchul Shin, Gyeong-Tae Kim, YeonJe Cho, Seongwoo Kim
https://arxiv.org/abs/2508.10747

Scaling Up without Fading Out: Goal-Aware Sparse GNN for RL-based Generalized Planning
Generalized planning using deep reinforcement learning (RL) combined with graph neural networks (GNNs) has shown promising results in various symbolic planning domains described by PDDL. However, existing approaches typically represent planning states as fully connected graphs, leading to a combinatorial explosion in edge information and substantial sparsity as problem scales grow, especially evident in large grid-based environments. This dense representation results in diluted node-level infor…

Interestingn post and replies.
I think that Apple’s incentives are just not aligned with developers in niche segments - purely a matter of scaling and managing the teams.
In most cases open source alternatives are vastly better.
But the question remains, how can you finance the maintenance and evolution in the long term - volunteer work is not sustainable.
It is remarkable that after 25 years we still haven’t found a viable and scalable funding model for this.

LaSM: Layer-wise Scaling Mechanism for Defending Pop-up Attack on GUI Agents
Zihe Yan, Zhuosheng Zhang
https://arxiv.org/abs/2507.10610 https://

LaSM: Layer-wise Scaling Mechanism for Defending Pop-up Attack on GUI Agents
Graphical user interface (GUI) agents built on multimodal large language models (MLLMs) have recently demonstrated strong decision-making abilities in screen-based interaction tasks. However, they remain highly vulnerable to pop-up-based environmental injection attacks, where malicious visual elements divert model attention and lead to unsafe or incorrect actions. Existing defense methods either require costly retraining or perform poorly under inductive interference. In this work, we systemati…

Scaling High-Performance Nanoribbon Transistors with Monolayer Transition Metal Dichalcogenides
Tara Pe\~na, Anton E. O. Persson, Andrey Krayev, \'Ashildur Fri{\dh}riksd\'ottir, Kathryn Neilson, Zhepeng Zhang, Anh Tuan Hoang, Jerry A. Yang, Lauren Hoang, Andrew J. Mannix, Paul C. McIntyre, Eric Pop
https://arxiv.org/abs/2509.09964

Scaling High-Performance Nanoribbon Transistors with Monolayer Transition Metal Dichalcogenides
Nanoscale transistors require aggressive reduction of all channel dimensions: length, width, and thickness. While monolayer two-dimensional semiconductors (2DS) offer ultimate thickness scaling, good performance has largely been achieved only in micrometer-wide channels. Here, we demonstrate both $\it{n}$- and $\it{p}$-type nanoribbon transistors based on monolayer 2DS, fabricated using a multi-patterning process, reaching channel widths down to 25 nm and lengths down to 50 nm. 'Anchored' conta…

Despite hype and optimistic projections, the humanoid robot industry faces hurdles, from battery life and design to limited demand for large-scale deployments (Evan Ackerman/IEEE Spectrum)
https://spectrum.ieee.org/humanoid-robot-scaling

Humanoid Robots: The Scaling Challenge
​It takes more than building a humanoid robot to build a humanoid robot product.

Scaling the memory wall using mixed-precision -- HPG-MxP on an exascale machine
Aditya Kashi, Nicholson Koukpaizan, Hao Lu, Michael Matheson, Sarp Oral, Feiyi Wang
https://arxiv.org/abs/2507.11512

Scaling the memory wall using mixed-precision -- HPG-MxP on an exascale machine
Mixed-precision algorithms have been proposed as a way for scientific computing to benefit from some of the gains seen for artificial intelligence (AI) on recent high performance computing (HPC) platforms. A few applications dominated by dense matrix operations have seen substantial speedups by utilizing low precision formats such as FP16. However, a majority of scientific simulation applications are memory bandwidth limited. Beyond preliminary studies, the practical gain from using mixed-preci…

"This convening is not an isolated event. It’s the first step in scaling a movement that reimagines how values-driven open science organisations can grow together, share resources, and strengthen the ecosystem we collectively serve. By working together, we can ensure that the future of open science is not only innovative, but also equitable, sustainable, and resilient."
https://www.carpentries.org/blog/2025/09/convening-to-reclaim-and-sustain-open-science-communities/
Great to see this forward looking collaborative work being done by @… @… @… @… and OLS

Scaling Learned Image Compression Models up to 1 Billion
Yuqi Li, Haotian Zhang, Li Li, Dong Liu, Feng Wu
https://arxiv.org/abs/2508.09075 https://arxiv.or…

Scaling Learned Image Compression Models up to 1 Billion
Recent advances in large language models (LLMs) highlight a strong connection between intelligence and compression. Learned image compression, a fundamental task in modern data compression, has made significant progress in recent years. However, current models remain limited in scale, restricting their representation capacity, and how scaling model size influences compression performance remains unexplored. In this work, we present a pioneering study on scaling up learned image compression mode…

Accurate Reduced Floating-Point Precision Implicit Monte Carlo
Simon Butson, Mathew Cleveland, Alex Long, Todd Palmer
https://arxiv.org/abs/2506.11962 http…

Accurate Reduced Floating-Point Precision Implicit Monte Carlo
This work describes methodologies to successfully implement the Implicit Monte Carlo (IMC) scheme for thermal radiative transfer in reduced-precision floating-point arithmetic. The methods used can be broadly categorized into scaling approaches and floating-point arithmetic manipulations. Scaling approaches entail re-scaling values to ensure computations stay within a representable range. Floating-point arithmetic manipulations involve changes to order of operations and alternative summation al…

Universal Scaling Laws for Deep Indentation Beyond the Hertzian Regime
Tong Mu, Changhong Linghu, Yanju Liu, Jinsong Leng, Huajian Gao, K. Jimmy Hsia
https://arxiv.org/abs/2506.11461

Universal Scaling Laws for Deep Indentation Beyond the Hertzian Regime
Deep indentation of soft materials is ubiquitous across scales in nature and engineering, yet accurate predictions of contact behaviors under extreme deformations ($δ/R > 1$) remain elusive due to geometric and material nonlinearities. Here, we investigate the indentation of rigid spheres into soft elastic substrates, resolving the highly nonlinear regime where the sphere becomes fully submerged. A universal geometric mapping approach reveals Hertz-type pressure distributions in the deformed c…

Scaling Relations, Morphological Stability, and Asymptotic Freedom of Plasma-Surface Deposition Dynamics
Joel Saucedo, Uday Lamba, Hasitha Mahabaduge
https://arxiv.org/abs/2507.10645

Scaling Relations, Morphological Stability, and Asymptotic Freedom of Plasma-Surface Deposition Dynamics
Connecting plasma processing parameters to the resultant film microstructure remains a fundamental challenge in materials synthesis, one that has largely confined process design to empirical approaches. To bridge this gap, we develop a predictive analysis of coupling by applying a renormalization group (RG) analysis to an effective Hamiltonian for the stochastic dynamics of the plasma-surface interface, derived systematically from microscopic principles. The central result from this formalism i…

Spin and thermal current scaling at a $Y$-junction of XX spin chains
Domenico Giuliano, Francesco Buccheri
https://arxiv.org/abs/2508.10267 https://arxiv.o…

Spin and thermal current scaling at a $Y$-junction of XX spin chains
We study the boundary phase diagram and the low-temperature heat and magnetization transport at a $Y$-junction of XX spin chains. Depending on the magnetization axis anisotropy between the magnetic exchange interactions at the junction, the system exhibits two different strong-coupling regimes at low energies/temperatures, similar to the overscreened (topological) four- and to the two-channel Kondo fixed points. Using renormalization group arguments combined with boundary conformal field theo…

Scaling Arabic Medical Chatbots Using Synthetic Data: Enhancing Generative AI with Synthetic Patient Records
Abdulrahman Allam, Seif Ahmed, Ali Hamdi, Khaled Shaban
https://arxiv.org/abs/2509.10108

Scaling Arabic Medical Chatbots Using Synthetic Data: Enhancing Generative AI with Synthetic Patient Records
The development of medical chatbots in Arabic is significantly constrained by the scarcity of large-scale, high-quality annotated datasets. While prior efforts compiled a dataset of 20,000 Arabic patient-doctor interactions from social media to fine-tune large language models (LLMs), model scalability and generalization remained limited. In this study, we propose a scalable synthetic data augmentation strategy to expand the training corpus to 100,000 records. Using advanced generative AI system…

🔧 #MatryoshkaRepresentationLearning technique allows scaling output dimensions from default 3072 💰 Priced at $0.15 per 1M input tokens with free tier available ⚡

Universal self-similarity of hierarchical communities formed through a general self-organizing principle
Shruti Tandon (equal), Nidhi Dilip Sonwane (equal), Tobias Braun, Norbert Marwan, Juergen Kurths, R. I. Sujith
https://arxiv.org/abs/2507.11159

Universal self-similarity of hierarchical communities formed through a general self-organizing principle
Emergence of self-similarity in hierarchical community structures is ubiquitous in complex systems. Yet, there is a dearth of universal quantification and general principles describing the formation of such structures. Here, we discover universality in scaling laws describing self-similar hierarchical community structure in multiple real-world networks including biological, infrastructural, and social networks. We replicate these scaling relations using a phenomenological model, where nodes wit…

Evidence of scaling advantage on an NP-Complete problem with enhanced quantum solvers
Quanfeng Lu, Shijie Wei, Keren Li, Pan Gao, Bao Yan, Muxi Zheng, Haoran Zhang, Jinfeng Zeng, Gui-Lu Long
https://arxiv.org/abs/2508.08869

Evidence of scaling advantage on an NP-Complete problem with enhanced quantum solvers
Achieving quantum advantage remains a key milestone in the noisy intermediate-scale quantum era. Without rigorous complexity proofs, scaling advantage-where quantum resource requirements grow more slowly than their classical counterparts-serves as the primary indicator. However, direct applications of quantum optimization algorithms to classically intractable problems have yet to demonstrate this advantage. To address this challenge, we develop enhanced quantum solvers for the NP-complete one-i…

US ethnic cleansing and what to do about it
Reposting link to source article instead of screenshot of tweet that had no alt text:
Data on arrests shows that ICE was heavily engaged in racial profiling in LA, because their arrest numbers fell by ~66% after that were ordered to stop making arrests based just in factors like skin color, with place, or language spoken.
#ICE #USPol

SwiftSpec: Ultra-Low Latency LLM Decoding by Scaling Asynchronous Speculative Decoding
Ziyi Zhang, Ziheng Jiang, Chengquan Jiang, Menghan Yu, Size Zheng, Haibin Lin, Henry Hoffmann, Xin Liu
https://arxiv.org/abs/2506.11309

SwiftSpec: Ultra-Low Latency LLM Decoding by Scaling Asynchronous Speculative Decoding
Low-latency decoding for large language models (LLMs) is crucial for applications like chatbots and code assistants, yet generating long outputs remains slow in single-query settings. Prior work on speculative decoding (which combines a small draft model with a larger target model) and tensor parallelism has each accelerated decoding. However, conventional approaches fail to apply both simultaneously due to imbalanced compute requirements (between draft and target models), KV-cache inconsistenc…

Comparison of Localization Algorithms between Reduced-Scale and Real-Sized Vehicles Using Visual and Inertial Sensors
Tobias Kern, Leon Tolksdorf, Christian Birkner
https://arxiv.org/abs/2507.11241

Comparison of Localization Algorithms between Reduced-Scale and Real-Sized Vehicles Using Visual and Inertial Sensors
Physically reduced-scale vehicles are emerging to accelerate the development of advanced automated driving functions. In this paper, we investigate the effects of scaling on self-localization accuracy with visual and visual-inertial algorithms using cameras and an inertial measurement unit (IMU). For this purpose, ROS2-compatible visual and visual-inertial algorithms are selected, and datasets are chosen as a baseline for real-sized vehicles. A test drive is conducted to record data of reduced-…

FuXi-\beta: Towards a Lightweight and Fast Large-Scale Generative Recommendation Model
Yufei Ye, Wei Guo, Hao Wang, Hong Zhu, Yuyang Ye, Yong Liu, Huifeng Guo, Ruiming Tang, Defu Lian, Enhong Chen
https://arxiv.org/abs/2508.10615

FuXi-β: Towards a Lightweight and Fast Large-Scale Generative Recommendation Model
Scaling laws for autoregressive generative recommenders reveal potential for larger, more versatile systems but mean greater latency and training costs. To accelerate training and inference, we investigated the recent generative recommendation models HSTU and FuXi-$α$, identifying two efficiency bottlenecks: the indexing operations in relative temporal attention bias and the computation of the query-key attention map. Additionally, we observed that relative attention bias in self-attention mec…

Grids Often Outperform Implicit Neural Representations
Namhoon Kim, Sara Fridovich-Keil
https://arxiv.org/abs/2506.11139 https://arxi…

Grids Often Outperform Implicit Neural Representations
Implicit Neural Representations (INRs) have recently shown impressive results, but their fundamental capacity, implicit biases, and scaling behavior remain poorly understood. We investigate the performance of diverse INRs across a suite of 2D and 3D real and synthetic signals with varying effective bandwidth, as well as both overfitting and generalization tasks including tomography, super-resolution, and denoising. By stratifying performance according to model size as well as signal type and ba…

Scaling behaviour of rotating convection in a spherical shell with different Prandtl numbers
Wei Fan, Qi Wang, Yufeng Lin
https://arxiv.org/abs/2508.09416 https://

Scaling behaviour of rotating convection in a spherical shell with different Prandtl numbers
Rayleigh-Benard convection in a rotating spherical shell provides a simplified model for convective dynamics of planetary and stellar interiors. In this study, we build more than 200 numerical models of rotating convection in a spherical shell over a wide range of $\Pran$ ($10^{-2}\le\Pran\le10^2$). As increasing the Rayleigh number $Ra$, we characterise four different flow regimes, starting from the linear onset to multiple modes, then transiting to the geostrophic turbulence and eventually ap…

FractalSync: Lightweight Scalable Global Synchronization of Massive Bulk Synchronous Parallel AI Accelerators
Victor Isachi, Alessandro Nadalini, Riccardo Fiorani Gallotta, Angelo Garofalo, Francesco Conti, Davide Rossi
https://arxiv.org/abs/2506.11668

FractalSync: Lightweight Scalable Global Synchronization of Massive Bulk Synchronous Parallel AI Accelerators
The slow-down of technology scaling and the emergence of Artificial Intelligence (AI) workloads have led computer architects to increasingly exploit parallelization coupled with hardware acceleration to keep pushing the performance envelope. However, this solution comes with the challenge of synchronization of processing elements (PEs) in massive heterogeneous many-core platforms. To address this challenge, we propose FractalSync, a hardware accelerated synchronization mechanism for Bulk Synchr…

A Prediction for Maximum Supercooling in SU(N) Confinement Transition
Prateek Agrawal, Gaurang Ramakant Kane, Vazha Loladze, John March-Russell
https://arxiv.org/abs/2508.10091 …

A Prediction for Maximum Supercooling in SU(N) Confinement Transition
The thermal confinement phase transition (PT) in $SU(N)$ Yang-Mills theory is first-order for $N\geq 3$, with bounce action scaling as $N^2$. Remarkably, lattice data for the action include a small coefficient whose presence likely strongly alters the PT dynamics. We give evidence, utilizing insights from softly-broken SUSY YM models, that the small coefficient originates from a deconfined phase instability just below the critical temperature. We predict the maximum achievable supercooling in $…

Critical Ising correlations on a torus
Baran Bayraktaroglu, Konstantin Izyurov
https://arxiv.org/abs/2506.11324 https://arxiv.org/pdf…

Critical Ising correlations on a torus
We prove convergence of multi-point spin correlations in the critical Ising model on a torus. Via Pfaffian identities, this also implies convergence of other correlations, including correlations of spins with fermionic and energy observables. We obtain explicit formulae for the scaling limits in terms of theta functions, verifying the predictions in the physics literature.

Invariant measures on moduli spaces of twisted holomorphic 1-forms and strata of dilation surfaces
Paul Apisa, Nick Salter
https://arxiv.org/abs/2507.10685

Invariant measures on moduli spaces of twisted holomorphic 1-forms and strata of dilation surfaces
The moduli space of twisted holomorphic 1-forms on Riemann surfaces, equivalently dilation surfaces with scaling, admits a stratification and GL(2,R)-action as in the case of moduli spaces of translation surfaces. We produce an analogue of Masur-Veech measure, i.e. an SL(2,R)-invariant Lebesgue class measure on strata or explicit covers thereof. This relies on a novel computation of cohomology with coefficients for the mapping class group. The computation produces a framed mapping class group i…

Universal Driven Critical Dynamics near the Boundary
Yu-Rong Shu, Shuai Yin
https://arxiv.org/abs/2509.10049 https://arxiv.org/pdf/2509.10049

Universal Driven Critical Dynamics near the Boundary
The celebrated Kibble-Zurek mechanism (KZM) describes the scaling of physical quantities when external parameters sweep through a critical point. Boundaries are ubiquitous in real systems, and critical behaviors near the boundary have attracted extensive research. Different boundary universality classes, including ordinary, special, extraordinary, and surface transitions, have been identified. However, the driven critical dynamics near boundaries remains unexplored. Here, we systematically inve…

Duty-Cycling is Not Enough in Constrained IoT Networking: Revealing the Energy Savings of Dynamic Clock Scaling
Michel Rottleuthner, Thomas C. Schmidt, Matthias W\"ahlisch
https://arxiv.org/abs/2508.09620

Duty-Cycling is Not Enough in Constrained IoT Networking: Revealing the Energy Savings of Dynamic Clock Scaling
Minimizing energy consumption of low-power wireless nodes is a persistent challenge from the constrained Internet of Things (IoT). In this paper, we start from the observation that constrained IoT devices have largely different hardware (im-)balances than full-scale machines. We find that the performance gap between MCU and network throughput on constrained devices enables minimal energy delay product (EDP) for IoT networking at largely reduced clock frequencies. We analyze the potentials by in…

Profiling Multi-Level Operator Costs for Bottleneck Diagnosis in High-Speed Data Planes
Zhiyuan Ren, Yutao Liu, Wenchi Cheng, Kun Yang
https://arxiv.org/abs/2508.09574 https://

Profiling Multi-Level Operator Costs for Bottleneck Diagnosis in High-Speed Data Planes
This paper proposes a saturation throughput delta-based methodology to precisely measure operator costs in high-speed data planes without intrusive instrumentation. The approach captures non-linear scaling, revealing that compute-intensive operators like CRC exhibit super-linear behavior, while most others are sub-linear. We introduce the Operator Performance Quadrant (OPQ) framework to classify operators by base and scaling costs, exposing a cross-architecture Quadrant Shift between Arm and x8…

"À la question de savoir si l'intensification des approches actuelles de l'IA pourrait conduire Š l'intelligence artificielle générale (AGI), ou Š une IA Š usage général qui égalerait ou surpasserait la cognition humaine, 76 % des personnes interrogées ont répondu qu'il était "improbable" ou "très improbable" que cela réussisse."
#IA

Primordial Black Hole Formation and Spin in Matter Domination Revisited
Weitao Ye, Yungui Gong, Tomohiro Harada, Zhaofeng Kang, Kazunori Kohri, Daiki Saito, Chul-Moon Yoo
https://arxiv.org/abs/2508.10070

Primordial Black Hole Formation and Spin in Matter Domination Revisited
In this article, we calculate the mass distribution of primordial black holes (PBHs) formed in the matter-dominated (MD) era by the peak theory. We apply the Zel'dovich approximation to track the nonlinear evolution of overdensities and compute the PBH abundance and mass function by incorporating a PBH formation criterion based on the hoop conjecture. We find that the PBH abundance $β$ follows the scaling law $β\simeq A_γσ_h^{*5}$ for $σ_h^*\ll 1$. Here, $σ_h^*$ is the quantity that chara…

Mechanics-Informed Machine Learning for Geospatial Modeling of Soil Liquefaction: Global and National Surrogate Models for Simulation and Near-Real-Time Response
Morgan D. Sanger, Mertcan Geyin, Brett W. Maurer
https://arxiv.org/abs/2509.10962

Mechanics-Informed Machine Learning for Geospatial Modeling of Soil Liquefaction: Global and National Surrogate Models for Simulation and Near-Real-Time Response
Using machine learning (ML), high performance computing, and a large body of geospatial information, we develop surrogate models to predict soil liquefaction across regional scales. Two sets of models - one global and one specific to New Zealand - are trained by learning to mimic geotechnical models at the sites of in-situ tests. Our geospatial approach has conceptual advantages in that predictions: (i) are anchored to mechanics, which encourages more sensible response and scaling across the do…

Time Scaling Makes Accelerated Gradient Flow and Proximal Method Faster in Multiobjective Optimization
Yingdong Yin
https://arxiv.org/abs/2508.07254 https://

Time Scaling Makes Accelerated Gradient Flow and Proximal Method Faster in Multiobjective Optimization
This paper extends a class of single-objective gradient flows and accelerated proximal methods to the multiobjective optimization domain within Euclidean spaces. The proposed gradient flow is a second-order differential equation composed of a second-order term, a first-order term with asymptotic vanishing behavior, and a gradient term with time scaling. We prove the existence of trajectory solutions to the equation and, through Lyapunov analysis, demonstrate that with appropriate parameter choi…

On surface energies in scaling laws for singular perturbation problems for martensitic phase transitions
Angkana R\"uland, Camillo Tissot, Antonio Tribuzio, Christian Zillinger
https://arxiv.org/abs/2507.06773 https://arxiv.org/pdf/2507.06773 https://arxiv.org/html/2507.06773
arXiv:2507.06773v1 Announce Type: new
Abstract: The objective of this article is to compare different surface energies for multi-well singular perturbation problems associated with martensitic phase transformations involving higher order laminates. We deduce scaling laws in the singular perturbation parameter which are robust in the choice of the surface energy (e.g., diffuse, sharp, an interpolation thereof or discrete). Furthermore, we show that these scaling laws do not require the presence of isotropic surface energies but that generically also highly anisotropic surface energies yield the same scaling results. More precisely, the presence of essentially generic partial directional derivatives in the regularization terms suffices to produce the same scaling behaviour as in the isotropic setting. The only sensitive directional dependences are directly linked to the lamination directions of the well structure -- and even for these only the ``inner-most'' lamination direction is of significance in determining the scaling law. In view of experimental applications, this shows that also for higher-order laminates, the precise structure of the surface energies -- which is often very difficult to determine experimentally -- does not have a crucial impact on the scaling behaviour of the investigated structures but only enters when considering finer properties.
toXiv_bot_toot

Primordial Black Hole Formation in a Scalar Field Dominated Universe: Investigation of the Critical nature of the Collapse
Luis E. Padilla, Ethan Milligan, David J. Mulryne, Juan Carlos Hidalgo
https://arxiv.org/abs/2509.10431

Primordial Black Hole Formation in a Scalar Field Dominated Universe: Investigation of the Critical nature of the Collapse
In this paper, we investigate the critical collapse leading to primordial black hole (PBH) formation in a universe dominated by a self-interacting scalar field with a quartic potential, comparing it to the well-known radiation-dominated case. Using fully relativistic nonlinear numerical simulations in spherical symmetry, based on the Misner--Sharp formalism, we analyze the dynamics near the collapse threshold and track the scaling of the black hole mass. Our results confirm that both the scalar…

Replaced article(s) found for q-bio.NC. https://arxiv.org/list/q-bio.NC/new
[1/1]:
- Neuronal correlations shape the scaling behavior of memory capacity and nonlinear computational c...
Shotaro Takasu, Toshio Aoyagi

Numerical analysis of the large deviation regime of a kinetic equation with a nonlocal Hamilton-Jacobi limit
H\'el\`ene Hivert, Tino Laidin
https://arxiv.org/abs/2509.10323 …

Numerical analysis of the large deviation regime of a kinetic equation with a nonlocal Hamilton-Jacobi limit
We develop and study an asymptotic-preserving (AP) numerical scheme for a linear kinetic equation in a large deviation regime. After applying a Hopf-Cole transform to the distribution function, the system exhibits the behavior of rare events, which in the limit is governed by a non-standard, nonlocal Hamilton-Jacobi equation, as identified in [E. Bouin et al., J. Lond. Math. Soc., II. Ser., 2023]. The proposed scheme efficiently handles the stiffness introduced by scaling, with a computationa…

Biological Processing Units: Leveraging an Insect Connectome to Pioneer Biofidelic Neural Architectures
Siyu Yu, Zihan Qin, Tingshan Liu, Beiya Xu, R. Jacob Vogelstein, Jason Brown, Joshua T. Vogelstein
https://arxiv.org/abs/2507.10951

Biological Processing Units: Leveraging an Insect Connectome to Pioneer Biofidelic Neural Architectures
The complete connectome of the Drosophila larva brain offers a unique opportunity to investigate whether biologically evolved circuits can support artificial intelligence. We convert this wiring diagram into a Biological Processing Unit (BPU), a fixed recurrent network derived directly from synaptic connectivity. Despite its modest size 3,000 neurons and 65,000 weights between them), the unmodified BPU achieves 98% accuracy on MNIST and 58% on CIFAR-10, surpassing size-matched MLPs. Scaling the…

Internal Slack message: OpenAI has hired four high-profile engineers from Tesla, xAI, and Meta, including David Lau, former VP of software engineering at Tesla (Wired)
https://www.wired.com/story/openai-new-hires-scaling/

OpenAI Poaches 4 High-Ranking Engineers From Tesla, xAI, and Meta
The new hires are joining OpenAI’s scaling team, according to an internal Slack message sent by cofounder Greg Brockman.

Compartmentalised Agentic Reasoning for Clinical NLI
Ma\"el Jullien, Lei Xu, Marco Valentino, Andr\'e Freitas
https://arxiv.org/abs/2509.10222 https://

Compartmentalised Agentic Reasoning for Clinical NLI
A common assumption holds that scaling data and parameters yields increasingly structured, generalisable internal representations. We interrogate this assumption in clinical natural language inference (NLI) by adopting a benchmark decomposed into four reasoning families, Causal Attribution, Compositional Grounding, Epistemic Verification, and Risk State Abstraction, and introducing CARENLI, a Compartmentalised Agentic Reasoning for Clinical NLI that separates knowledge access from principled in…

Insights for Early Massive Black Hole Growth from JWST Detection of the [Ne v] {\lambda}3427 Emission Line
Benny Trakhtenbrot, Claudio Ricci, Ezequiel Treister, Michael J. Koss, Richard Mushotzky, Kyuseok Oh, Alessandro Peca, Franz E. Bauer, Kriti Kamal Gupta, Tomer Reiss
https://arxiv.org/abs/2507.10681

Insights for Early Massive Black Hole Growth from JWST Detection of the [Ne v] λ3427 Emission Line
We use the narrow [Ne v] $λ$3427 emission line detected in the recently published JWST spectra of two galaxies, at z = 6.9 and 5.6, to study the key properties of the active galactic nuclei (AGN) and the supermassive black holes in their centers. Using a new empirical scaling linking the [Ne v] line emission with AGN accretion-driven (continuum) emission, derived from a highly complete low-redshift AGN sample, we show that the [Ne v] emission in the two z > 5 galaxies implies total (bolometric…

Crosslisted article(s) found for q-bio.BM. https://arxiv.org/list/q-bio.BM/new
[1/1]:
- Physical Principles of Size and Frequency Scaling of Active Cytoskeletal Spirals
Aman Soni, Shivani A. Yadav, Chaitanya A. Athale

Natively Trainable Sparse Attention for Hierarchical Point Cloud Datasets
Nicolas Lapautre, Maria Marchenko, Carlos Miguel Pati\~no, Xin Zhou
https://arxiv.org/abs/2508.10758 ht…

Natively Trainable Sparse Attention for Hierarchical Point Cloud Datasets
Unlocking the potential of transformers on datasets of large physical systems depends on overcoming the quadratic scaling of the attention mechanism. This work explores combining the Erwin architecture with the Native Sparse Attention (NSA) mechanism to improve the efficiency and receptive field of transformer models for large-scale physical systems, addressing the challenge of quadratic attention complexity. We adapt the NSA mechanism for non-sequential data, implement the Erwin NSA model, and…

The Darkfield Approach to Measuring Vacuum Birefringence and Light-by-Light Couplings -- A Proof-of-Principle Experiment
Michal Sm\'id, Pooyan Khademi, Carsten B\"ahtz, Erik Brambrink, Jindrich Chalupsky, Tom E. Cowan, Samuele Di Dio Cafiso, Sebastian G\"ode, J\"org Grenzer, Vera Hajkova, Peter Hilz, Willi Hippler, Hauke H\"opner, Alzbeta Horynova, Oliver Humphries, Simon Jelinek, Libor Juha, Felix Karbstein, Alejandro Laso-Garcia, Robert L\"otzsch, Aim\…

The Darkfield Approach to Measuring Vacuum Birefringence and Light-by-Light Couplings -- A Proof-of-Principle Experiment
Vacuum fluctuations give rise to effective nonlinear interactions between electromagnetic fields. These generically modify the characteristics of light traversing a strong-field region. X-ray free-electron lasers constitute a particularly promising probe, due to their brilliance, the possibility of precise control and favourable frequency scaling. However, the nonlinear vacuum response is very small even when probing a tightly focused high-intensity laser field with XFEL radiation and direct me…

Crosslisted article(s) found for physics.bio-ph. https://arxiv.org/list/physics.bio-ph/new
[1/1]:
- Physical Principles of Size and Frequency Scaling of Active Cytoskeletal Spirals
Aman Soni, Shivani A. Yadav, Chaitanya A. Athale

Replaced article(s) found for cs.CV. https://arxiv.org/list/cs.CV/new
[3/3]:
- Taccel: Scaling Up Vision-based Tactile Robotics via High-performance GPU Simulation
Li, Du, Yu, Li, Zhao, Liu, Jiang, Zhu, Huang

Crosslisted article(s) found for cond-mat.mes-hall. https://arxiv.org/list/cond-mat.mes-hall/new
[1/1]:
- Scaling High-Performance Nanoribbon Transistors with Monolayer Transition Metal Dichalcogenides
na, et al.

Dynamic scaling of growing interfaces
Pierre Le Doussal
https://arxiv.org/abs/2507.08341 https://arxiv.org/pdf/2507.08341

Dynamic scaling of growing interfaces
We give a brief overview of the seminal paper which introduced the Kardar-Parisi-Zhang equation as a paradigmatic model for random growth in 1986. We describe some of the developments to which it gave rise in mathematics and physics over the years, and some examples of applications.

Exploring Quantum Annealing for Coarse-Grained Protein Folding
Timon Scheiber, Matthias Heller, Andreas Giebel
https://arxiv.org/abs/2508.10660 https://arx…

Exploring Quantum Annealing for Coarse-Grained Protein Folding
We explore the potential application of quantum annealing to address the protein structure problem. To this end, we compare several proposed ab initio protein folding models for quantum computers and analyze their scaling and performance for classical and quantum heuristics. Furthermore, we introduce a novel encoding of coordinate based models on the tetrahedral lattice, based on interleaved grids. Our findings reveal significant variations in model performance, with one model yielding unphysic…

Admissibility of Stein Shrinkage for Batch Normalization in the Presence of Adversarial Attacks
Sofia Ivolgina, P. Thomas Fletcher, Baba C. Vemuri
https://arxiv.org/abs/2507.08261

Admissibility of Stein Shrinkage for Batch Normalization in the Presence of Adversarial Attacks
Batch normalization (BN) is a ubiquitous operation in deep neural networks used primarily to achieve stability and regularization during network training. BN involves feature map centering and scaling using sample means and variances, respectively. Since these statistics are being estimated across the feature maps within a batch, this problem is ideally suited for the application of Stein's shrinkage estimation, which leads to a better, in the mean-squared-error sense, estimate of the mean and …

SWE-Mirror: Scaling Issue-Resolving Datasets by Mirroring Issues Across Repositories
Junhao Wang, Daoguang Zan, Shulin Xin, Siyao Liu, Yurong Wu, Kai Shen
https://arxiv.org/abs/2509.08724

SWE-Mirror: Scaling Issue-Resolving Datasets by Mirroring Issues Across Repositories
Creating large-scale verifiable training datasets for issue-resolving tasks is a critical yet notoriously difficult challenge. Existing methods on automating the Gym environment setup process for real-world issues suffer from low success rates and high overhead. Meanwhile, synthesizing new tasks within existing Gym environments leaves the vast pool of authentic, human-reported problems untapped. To maximize the utilization of existing Gym environments and also the rich data of issue-resolving h…

Handows: A Palm-Based Interactive Multi-Window Management System in Virtual Reality
Jindu Wang, Ke Zhou, Haoyu Ren, Per Ola Kristensson, Xiang Li
https://arxiv.org/abs/2508.09469

Handows: A Palm-Based Interactive Multi-Window Management System in Virtual Reality
Window management in virtual reality (VR) remains a challenging task due to the spatial complexity and physical demands of current interaction methods. We introduce Handows, a palm-based interface that enables direct manipulation of spatial windows through familiar smartphone-inspired gestures on the user's non-dominant hand. Combining ergonomic layout design with body-centric input and passive haptics, Handows supports four core operations: window selection, closure, positioning, and scaling. …

InGaN Nanopixel Arrays on Single Crystal GaN Substrate
Nirmal Anand, Sadat Tahmeed Azad, Christy Giji Jenson, Dipon Kumar Ghosh, Md Zunaid Baten, Pei-Cheng Ku, Grzegorz Muziol, Sharif Sadaf
https://arxiv.org/abs/2506.11408

InGaN Nanopixel Arrays on Single Crystal GaN Substrate
Indium gallium nitride (InGaN) quantum well (QW) micro- and nano-scale light-emitting diodes (LEDs) are vital to the next-generation optoelectronic applications such as ultrahigh-speed optical interconnects and near-eye augmented/virtual reality displays. However, scaling to nanoscale dimensions introduces several challenges, including an increased nonradiative recombination from surface states and defect-related emission degradation. In this work, we present a high-performance, strain-relaxed …

Counting sums of two powers
Anand Patel
https://arxiv.org/abs/2507.08337 https://arxiv.org/pdf/2507.08337

Counting sums of two powers
We generalize two well-known enumerative facts. The first, due to Clebsch, says that a general binary sextic form is expressible as the sum of a cube and a square in 40 different ways. The second, due to Zariski and later Vakil, states that a general pencil of binary 12-ics contains exactly 3762 cubes plus squares, ignoring scaling.

SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning
Haozhan Li, Yuxin Zuo, Jiale Yu, Yuhao Zhang, Zhaohui Yang, Kaiyan Zhang, Xuekai Zhu, Yuchen Zhang, Tianxing Chen, Ganqu Cui, Dehui Wang, Dingxiang Luo, Yuchen Fan, Youbang Sun, Jia Zeng, Jiangmiao Pang, Shanghang Zhang, Yu Wang, Yao Mu, Bowen Zhou, Ning Ding
https://arxiv.org/a…

SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning
Vision-Language-Action (VLA) models have recently emerged as a powerful paradigm for robotic manipulation. Despite substantial progress enabled by large-scale pretraining and supervised fine-tuning (SFT), these models face two fundamental challenges: (i) the scarcity and high cost of large-scale human-operated robotic trajectories required for SFT scaling, and (ii) limited generalization to tasks involving distribution shift. Recent breakthroughs in Large Reasoning Models (LRMs) demonstrate tha…

The Knowledge-Reasoning Dissociation: Fundamental Limitations of LLMs in Clinical Natural Language Inference
Ma\"el Jullien, Marco Valentino, Andr\'e Freitas
https://arxiv.org/abs/2508.10777

The Knowledge-Reasoning Dissociation: Fundamental Limitations of LLMs in Clinical Natural Language Inference
Large language models are often assumed to acquire increasingly structured, generalizable internal representations simply by scaling data and parameters. We interrogate this assumption by introducing a Clinical Trial Natural Language Inference benchmark comprising four reasoning families, Causal Attribution, Compositional Grounding, Epistemic Verification, and Risk State Abstraction. Each item is paired with a targeted Ground Knowledge and Meta-Level Reasoning Verification (GKMRV) probe, allowi…

HMD says it will "scale back" US operations, citing "a challenging geopolitical and economic environment", and appears to have stopped US sales of Nokia devices (Dominic Preston/The Verge)
https://www.theverge.com/news/705046/hmd-gl…

HMD is ‘scaling back’ in the US, killing Nokia all over again
HMD Global, which sells its own phones as well as Nokia ones, has stopped selling phones on its US site, and confirms it will ‘scale back’ its US operation.

The Hidden Width of Deep ResNets: Tight Error Bounds and Phase Diagrams
L\'ena\"ic Chizat
https://arxiv.org/abs/2509.10167 https://arxiv.org/pdf/2…

The Hidden Width of Deep ResNets: Tight Error Bounds and Phase Diagrams
We study the gradient-based training of large-depth residual networks (ResNets) from standard random initializations. We show that with a diverging depth $L$, a fixed embedding dimension $D$, and an arbitrary hidden width $M$, the training dynamics converges to a Neural Mean ODE training dynamics. Remarkably, the limit is independent of the scaling of $M$, covering practical cases of, say, Transformers, where $M$ (the number of hidden units or attention heads per layer) is typically of the orde…

SSRL: Self-Search Reinforcement Learning
Yuchen Fan, Kaiyan Zhang, Heng Zhou, Yuxin Zuo, Yanxu Chen, Yu Fu, Xinwei Long, Xuekai Zhu, Che Jiang, Yuchen Zhang, Li Kang, Gang Chen, Cheng Huang, Zhizhou He, Bingning Wang, Lei Bai, Ning Ding, Bowen Zhou
https://arxiv.org/abs/2508.10874

SSRL: Self-Search Reinforcement Learning
We investigate the potential of large language models (LLMs) to serve as efficient simulators for agentic search tasks in reinforcement learning (RL), thereby reducing dependence on costly interactions with external search engines. To this end, we first quantify the intrinsic search capability of LLMs via structured prompting and repeated sampling, which we term Self-Search. Our results reveal that LLMs exhibit strong scaling behavior with respect to the inference budget, achieving high pass@k …

RewardDance: Reward Scaling in Visual Generation
Jie Wu, Yu Gao, Zilyu Ye, Ming Li, Liang Li, Hanzhong Guo, Jie Liu, Zeyue Xue, Xiaoxia Hou, Wei Liu, Yan Zeng, Weilin Huang
https://arxiv.org/abs/2509.08826

RewardDance: Reward Scaling in Visual Generation
Reward Models (RMs) are critical for improving generation models via Reinforcement Learning (RL), yet the RM scaling paradigm in visual generation remains largely unexplored. It primarily due to fundamental limitations in existing approaches: CLIP-based RMs suffer from architectural and input modality constraints, while prevalent Bradley-Terry losses are fundamentally misaligned with the next-token prediction mechanism of Vision-Language Models (VLMs), hindering effective scaling. More critical…

A scalable quantum-neural hybrid variational algorithm for ground state estimation
Minwoo Kim, Kyoung Keun Park, Uihwan Jeong, Sanghyeon Lee, Taehyun Kim
https://arxiv.org/abs/2507.11002

A scalable quantum-neural hybrid variational algorithm for ground state estimation
We propose the unitary variational quantum-neural hybrid eigensolver (U-VQNHE), which improves upon the original VQNHE by enforcing unitary neural transformations. The non-unitary nature of VQNHE causes normalization issues and divergence of the loss function during training, leading to exponential scaling of measurement overhead with qubit number. U-VQNHE resolves these issues, significantly reduces required measurements, and retains improved accuracy and stability over standard variational qu…

Casimir scaling in glueballs in SU($N$) and Sp($2N$) gauge theories: hints from constituent approaches
F. Buisseret, C. Chevalier, V. Mathieu, C. Semay
https://arxiv.org/abs/2509.09454

Casimir scaling in glueballs in SU($N$) and Sp($2N$) gauge theories: hints from constituent approaches
We show that the lattice glueball masses $M_G$ versus $N$ in SU($N$) and Sp($2N$) Yang-Mills theories scale as $\frac{M_G}{\sqrtσ}\sim \sqrt{\frac{C_2(adj)}{C_2(f)}}$, with $σ$ the fundamental string tension and $C_2(adj)$ and $C_2(f)$ the quadratic Casimir of the gauge algebra in the adjoint and fundamental representations. This scaling behaviour is followed by the great majority of available lattice glueball states, and may set constraints on $SU(3)$ models by imposing a specific behaviour …

Characterizing the Efficiency of Distributed Training: A Power, Performance, and Thermal Perspective
Seokjin Go, Joongun Park, Spandan More, Hanjiang Wu, Irene Wang, Aaron Jezghani, Tushar Krishna, Divya Mahajan
https://arxiv.org/abs/2509.10371

Characterizing the Efficiency of Distributed Training: A Power, Performance, and Thermal Perspective
The rapid scaling of Large Language Models (LLMs) has pushed training workloads far beyond the limits of single-node analysis, demanding a deeper understanding of how these models behave across large-scale, multi-GPU systems. In this paper, we present a comprehensive characterization of LLM training across diverse real-world workloads and hardware platforms, including NVIDIA H100/H200 and AMD MI250 GPUs. We analyze dense and sparse models under various parallelism strategies -- tensor, pipeline…

From Kardar-Parisi-Zhang scaling to soliton proliferation in Josephson junction arrays
Mikheil Tsitsishvili, Reinhold Egger, Karsten Flensberg, Sebastian Diehl
https://arxiv.org/abs/2509.08479

From Kardar-Parisi-Zhang scaling to soliton proliferation in Josephson junction arrays
We propose Josephson junction arrays as realistic platforms for observing nonequilibrium scaling laws characterizing the Kardar-Parisi-Zhang (KPZ) universality class, and space-time soliton proliferation. Focusing on a two-chain ladder geometry, we perform numerical simulations for the roughness function. Together with analytical arguments, our results predict KPZ scaling at intermediate time scales, extending over sufficiently long time scales to be observable, followed by a crossover to the a…

Optimal trace norms for Helmholtz problems
Benedikt Gr\"a{\ss}le
https://arxiv.org/abs/2506.11944 https://arxiv.org/pdf/2506.119…

Optimal trace norms for Helmholtz problems
The natural $H^1(Ω)$ energy norm for Helmholtz problems is weighted with the wavenumber modulus $σ$ and induces natural weighted norms on the trace spaces $H^{\pm1/2}(Γ)$ by minimial extension to $Ω\subset\mathbb R^n$. This paper presents a rigorous analysis for these trace norms with an explicit characterisation by weighted Sobolev-Slobodeckij norms and scaling estimates, highlighting their dependence on the geometry of the extension set $Ω\subset\mathbb R^n$ and the weight $σ$. The anal…

Semi-empirical constraints on the HI mass function of star-forming galaxies and $\Omega_{\rm HI}$ at $z\sim 0.37$ from interferometric surveys
Francesco Sinigaglia, Alessandro Bianchetti, Giulia Rodighiero, Lucio Mayer, Miroslava Dessauges-Zavadsky, Ed Elson, Mattia Vaccari, Matt J. Jarvis
https://arxiv.org/abs/2506.11280…

Semi-empirical constraints on the HI mass function of star-forming galaxies and $Ω_{\rm HI}$ at $z\sim 0.37$ from interferometric surveys
The HI mass function is a crucial tool to understand the evolution of the HI content in galaxies over cosmic times and to constrain both the baryon cycle in galaxy evolution and the reionization history of the Universe. We aim to derive semi-empirical constraints at $z\sim 0.37$ by combining literature results on the stellar mass function from optical surveys with recent findings on the $M_{\rm HI}-M_\star$ scaling relation derived via spectral stacking analysis applied to 21-cm line interferom…

The growth of magnetic energy during the nonlinear phase of the subsonic and supersonic small-scale dynamo
Neco Kriel, James R. Beattie, Mark R. Krumholz, Jennifer Schober, Patrick J. Armstrong
https://arxiv.org/abs/2509.09949

The growth of magnetic energy during the nonlinear phase of the subsonic and supersonic small-scale dynamo
Small-scale dynamos (SSDs) amplify magnetic fields in turbulent plasmas. Theory predicts nonlinear magnetic energy growth $E_\mathrm{mag} \propto t^{p_\mathrm{nl}}$, but this scaling has not been tested across flow regimes. Using a large ensemble of SSD simulations spanning sub- to supersonic turbulence, we find linear growth ($p_\mathrm{nl} = 1$) in subsonic flows and quadratic growth ($p_\mathrm{nl} = 2$) in supersonic flows. In all cases, we find a dynamo efficiency of $\sim 1/100$ and a dur…

Certifying and learning quantum Ising Hamiltonians
Andreas Bluhm, Matthias C. Caro, Francisco Escudero Guti\'errez, Aadil Oufkir, Cambyse Rouz\'e
https://arxiv.org/abs/2509.10239

Certifying and learning quantum Ising Hamiltonians
In this work, we study the problems of certifying and learning quantum Ising Hamiltonians. Our main contributions are as follows: Certification of Ising Hamiltonians. We show that certifying an Ising Hamiltonian in normalized Frobenius norm via access to its time-evolution operator requires only $\widetilde O(1/\varepsilon)$ time evolution. This matches the Heisenberg-scaling lower bound of $Ω(1/\varepsilon)$ up to logarithmic factors. To our knowledge, this is the first nearly-optimal algor…

Cyclic Data Streaming on GPUs for Short Range Stencils Applied to Molecular Dynamics
Martin Rose, Simon Homes, Lukas Ramsperger, Jose Gracia, Christoph Niethammer, Jadran Vrabec
https://arxiv.org/abs/2507.11289

Cyclic Data Streaming on GPUs for Short Range Stencils Applied to Molecular Dynamics
In the quest for highest performance in scientific computing, we present a novel framework that relies on high-bandwidth communication between GPUs in a compute cluster. The framework offers linear scaling of performance for explicit algorithms that is only limited by the size of the dataset and the number of GPUs. Slices of the dataset propagate in a ring of processes (GPUs) from one GPU, where they are processed, to the next, which results in a parallel-in-time parallelization. The user of th…

Source: Nvidia is scaling back DGX Cloud to primarily internal R&D use; DGX Cloud was initially envisioned to compete with major cloud providers like AWS (Anissa Gardizy/The Information)
https://www.theinformation.com/articles/nvidia-steps-back-cloud-ef…

Nvidia Steps Back From Cloud Effort to Compete with AWS
Nvidia is stepping back from its nascent cloud computing business, which had put it in quasi-competition with Amazon Web Services. It has lessened its efforts to attract businesses to the cloud service, dubbed DGX Cloud, said a person with direct knowledge of the situation. Instead, Nvidia plans ...

Video Parallel Scaling: Aggregating Diverse Frame Subsets for VideoLLMs
Hyungjin Chung, Hyelin Nam, Jiyeon Kim, Hyojun Go, Byeongjun Park, Junho Kim, Joonseok Lee, Seongsu Ha, Byung-Hoon Kim
https://arxiv.org/abs/2509.08016

Video Parallel Scaling: Aggregating Diverse Frame Subsets for VideoLLMs
Video Large Language Models (VideoLLMs) face a critical bottleneck: increasing the number of input frames to capture fine-grained temporal detail leads to prohibitive computational costs and performance degradation from long context lengths. We introduce Video Parallel Scaling (VPS), an inference-time method that expands a model's perceptual bandwidth without increasing its context window. VPS operates by running multiple parallel inference streams, each processing a unique, disjoint subset of …

Hilbert subspace imprint: a new mechanism for non-thermalization
Hui Yu, Jiangping Hu, Shi-Xin Zhang
https://arxiv.org/abs/2506.11922 https://

Hilbert subspace imprint: a new mechanism for non-thermalization
The search for non-ergodic mechanisms in quantum many-body systems has become a frontier area of research in non-equilibrium physics. In this Letter, we introduce Hilbert subspace imprint (HSI)-a novel mechanism that enables evasion of thermalization and bridges the gap between quantum many-body scars (QMBS) and Hilbert space fragmentation (HSF). HSI manifests when initial states overlap exclusively with a polynomial scaling (with system size) set of eigenstates. We demonstrate this phenomenon …

Electron Heating in Hypersonic Flows: A New Thermodynamically Consistent Model
Felipe Martin Rodriguez Fuentes, Bernard Parent
https://arxiv.org/abs/2506.11457

Electron Heating in Hypersonic Flows: A New Thermodynamically Consistent Model
Accurate prediction of electron temperature ($T_{\rm e}$) in hypersonic non-equilibrium flows is critical, yet hampered by inadequate models for electron heating from vibrationally excited nitrogen. Prior models often relied on ad-hoc scaling or flawed applications of detailed balance that failed to ensure the convergence of electron and vibrational temperatures ($T_{\rm v}$) at thermal equilibrium. This paper introduces a novel, thermodynamically consistent electron heating model derived rigor…

Ergodicity detection algorithms: Scaling of ergodicity in random symbolic dynamics
M. S\"uzen
https://arxiv.org/abs/2508.08319 https://arxiv.org/pdf/2…

Ergodicity detection algorithms: Scaling of ergodicity in random symbolic dynamics
The mathematical definitions of distinct concepts that are needed in building an ergodicity detec- tion algorithm are introduced in a framework. This algorithmic framework is expressed in a discrete setting in an accessible manner for broader quantitative practitioners without loss of generality. At the same time, the common misconceptions of the requirement of visiting all available states in the time-averaged quantities for physical systems and non-existence of an ergodic process are resolved…

Coordinated Reinforcement Learning Prefetching Architecture for Multicore Systems
Mohammed Humaid Siddiqui, Fernando Guzman, Yufei Wu, Ruishu Ann
https://arxiv.org/abs/2509.10719

Coordinated Reinforcement Learning Prefetching Architecture for Multicore Systems
Hardware prefetching is critical to fill the performance gap between CPU speeds and slower memory accesses. With multicore architectures becoming commonplace, traditional prefetchers are severely challenged. Independent core operation creates significant redundancy (up to 20% of prefetch requests are duplicates), causing unnecessary memory bus traffic and wasted bandwidth. Furthermore, cutting-edge prefetchers such as Pythia suffer from about a 10% performance loss when scaling from a single-co…

Compass-v3: Scaling Domain-Specific LLMs for Multilingual E-Commerce in Southeast Asia
Sophia Maria
https://arxiv.org/abs/2509.09121 https://arxiv.org/pdf/…

Compass-v3: Scaling Domain-Specific LLMs for Multilingual E-Commerce in Southeast Asia
Large language models (LLMs) excel in general-domain applications, yet their performance often degrades in specialized tasks requiring domain-specific knowledge. E-commerce is particularly challenging, as its data are noisy, heterogeneous, multilingual, and highly dynamic. We present Compass-v3, a vertical-domain Mixture-of-Experts (MoE) model with 245B total parameters and 71B active per token, designed for Southeast Asian e-commerce. Compass-v3 adopts fewer but larger experts, combined with h…

Sticker-TTS: Learn to Utilize Historical Experience with a Sticker-driven Test-Time Scaling Framework
Jie Chen, Jinhao Jiang, Yingqian Min, Zican Dong, Shijie Wang, Wayne Xin Zhao, Ji-Rong Wen
https://arxiv.org/abs/2509.05007

Sticker-TTS: Learn to Utilize Historical Experience with a Sticker-driven Test-Time Scaling Framework
Large reasoning models (LRMs) have exhibited strong performance on complex reasoning tasks, with further gains achievable through increased computational budgets at inference. However, current test-time scaling methods predominantly rely on redundant sampling, ignoring the historical experience utilization, thereby limiting computational efficiency. To overcome this limitation, we propose Sticker-TTS, a novel test-time scaling framework that coordinates three collaborative LRMs to iteratively e…

Pony AI says it rolled out 200 Gen-7 Robotaxis since mass production started two months ago, putting it on track for its 1,000-vehicle goal by the end of 2025 (Bloomberg)
https://www.bloomberg.com/news/articles/2025-08-13/po…

Microscopic calculation of two-particle-two-hole meson-exchange currents in $^{40}$Ar and asymmetric scaling properties for neutrino and electron scattering
V. L. Martinez-Consentino, J. Segovia, J. E. Amaro
https://arxiv.org/abs/2509.08916

Microscopic calculation of two-particle-two-hole meson-exchange currents in $^{40}$Ar and asymmetric scaling properties for neutrino and electron scattering
We present a microscopic calculation of two particle-two hole meson exchange current response functions in asymmetric nuclei, with particular emphasis on the $^{40}$Ar nucleus. Employing a relativistic mean-field and relativistic Fermi gas framework, we compute the nuclear response for $^{40}$Ar and compare it with that of the symmetric $^{40}$Ca nucleus, analyzing the role of proton-neutron imbalance. The model incorporates distinct proton and neutron Fermi momenta to accurately capture the nu…

Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search
Xin Lai, Junyi Li, Wei Li, Tao Liu, Tianjian Li, Hengshuang Zhao
https://arxiv.org/abs/2509.07969 …

Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search
Recent advances in large multimodal models have leveraged image-based tools with reinforcement learning to tackle visual problems. However, existing open-source approaches often exhibit monotonous reasoning patterns and allow only a limited number of interaction turns, making them inadequate for difficult tasks that require trial-and-error exploration. In this work, we address this limitation by scaling up tool-based interactions and introduce Mini-o3, a system that executes deep, multi-turn re…

Multi-agent Reinforcement Learning-based In-place Scaling Engine for Edge-cloud Systems
Jovan Prodanov, Bla\v{z} Bertalani\v{c}, Carolina Fortuna, Shih-Kai Chou, Matja\v{z} Branko Juri\v{c}, Ramon Sanchez-Iborra, Jernej Hribar
https://arxiv.org/abs/2507.07671

Multi-agent Reinforcement Learning-based In-place Scaling Engine for Edge-cloud Systems
Modern edge-cloud systems face challenges in efficiently scaling resources to handle dynamic and unpredictable workloads. Traditional scaling approaches typically rely on static thresholds and predefined rules, which are often inadequate for optimizing resource utilization and maintaining performance in distributed and dynamic environments. This inefficiency hinders the adaptability and performance required in edge-cloud infrastructures, which can only be achieved through the newly proposed in-…

Multiparameter quantum metrology at Heisenberg scaling for an arbitrary two-channel linear interferometer with squeezed light
Atmadev Rai, Danilo Triggiani, Paolo Facchi, Vincenzo Tamma
https://arxiv.org/abs/2509.07574

Multiparameter quantum metrology at Heisenberg scaling for an arbitrary two-channel linear interferometer with squeezed light
We present a framework for simultaneously estimating all four real parameters of a general two-channel unitary U(2) with Heisenberg-scaling precision. We derive analytical expressions for the quantum Fisher information matrix and show that all parameters attain the 1/N scaling in the precision by using experimentally feasible Gaussian probes such as two-mode squeezed states or two single-mode squeezed states. Our results extend multiparameter metrology to its most general two-mode setting and e…

Constraint correlation functions of the Ising model in the scaling limit
Ivan Balog, Adam Ran\c{c}on
https://arxiv.org/abs/2509.08557 https://arxiv.org/pdf…

Constraint correlation functions of the Ising model in the scaling limit
We study the correlation function of the one-dimensional Ising model at fixed magnetization. Focusing on the scaling limit close to the zero-temperature fixed point, we show that this correlation function, in momentum space, exhibits surprising oscillations as a function of the magnetization. We show that these oscillations have a period inversely proportional to the momentum and give an interpretation in terms of domain walls. This is in sharp contrast with the behavior of the correlation func…

Uncovering Scaling Laws for Large Language Models via Inverse Problems
Arun Verma, Zhaoxuan Wu, Zijian Zhou, Xiaoqiang Lin, Zhiliang Chen, Rachael Hwee Ling Sim, Rui Qiao, Jingtan Wang, Nhung Bui, Xinyuan Niu, Wenyang Hu, Gregory Kang Ruey Lau, Zi-Yu Khoo, Zitong Zhao, Xinyi Xu, Apivich Hemachandra, See-Kiong Ng, Bryan Kian Hsiang Low
https://

Uncovering Scaling Laws for Large Language Models via Inverse Problems
Large Language Models (LLMs) are large-scale pretrained models that have achieved remarkable success across diverse domains. These successes have been driven by unprecedented complexity and scale in both data and computations. However, due to the high costs of training such models, brute-force trial-and-error approaches to improve LLMs are not feasible. Inspired by the success of inverse problems in uncovering fundamental scientific laws, this position paper advocates that inverse problems can …

Speed Always Wins: A Survey on Efficient Architectures for Large Language Models
Weigao Sun, Jiaxi Hu, Yucheng Zhou, Jusen Du, Disen Lan, Kexin Wang, Tong Zhu, Xiaoye Qu, Yu Zhang, Xiaoyu Mo, Daizong Liu, Yuxuan Liang, Wenliang Chen, Guoqi Li, Yu Cheng
https://arxiv.org/abs/2508.09834

Speed Always Wins: A Survey on Efficient Architectures for Large Language Models
Large Language Models (LLMs) have delivered impressive results in language understanding, generation, reasoning, and pushes the ability boundary of multimodal models. Transformer models, as the foundation of modern LLMs, offer a strong baseline with excellent scaling properties. However, the traditional transformer architecture requires substantial computations and poses significant obstacles for large-scale training and practical deployment. In this survey, we offer a systematic examination of…

Unreal is all you need: Multimodal ISAC Data Simulation with Only One Engine
Kongwu Huang, Shiyi Mu, Jun Jiang, Yuan Gao, Shugong Xu
https://arxiv.org/abs/2507.08716

Unreal is all you need: Multimodal ISAC Data Simulation with Only One Engine
Scaling laws have achieved success in LLM and foundation models. To explore their potential in ISAC research, we propose Great-X. This single-engine multimodal data twin platform reconstructs the ray-tracing computation of Sionna within Unreal Engine and is deeply integrated with autonomous driving tools. This enables efficient and synchronized simulation of multimodal data, including CSI, RGB, Radar, and LiDAR. Based on this platform, we construct an open-source, large-scale, low-altitude UAV …

Meek Models Shall Inherit the Earth
Hans Gundlach, Jayson Lynch, Neil Thompson
https://arxiv.org/abs/2507.07931 https://arxiv.org/pdf…

Meek Models Shall Inherit the Earth
The past decade has seen incredible scaling of AI systems by a few companies, leading to inequality in AI model performance. This paper argues that, contrary to prevailing intuition, the diminishing returns to compute scaling will lead to a convergence of AI model capabilities. In other words, meek models (those with limited computation budget) shall inherit the earth, approaching the performance level of the best models overall. We develop a model illustrating that under a fixed-distribution n…

Google and IBM believe the first industrial-scale quantum computer is in sight, potentially by 2030, but challenges like scaling from ~200 to 1M qubits remain (Richard Waters/Financial Times)
https://www.ft.com/content/2fe4b1a3-b0d3-403a-bdd3-f033b3e5e56a

Crown, Frame, Reverse: Layer-Wise Scaling Variants for LLM Pre-Training
Andrei Baroian, Kasper Notebomer
https://arxiv.org/abs/2509.06518 https://arxiv.org…

Crown, Frame, Reverse: Layer-Wise Scaling Variants for LLM Pre-Training
Transformer-based language models traditionally use uniform (isotropic) layer sizes, yet they ignore the diverse functional roles that different depths can play and their computational capacity needs. Building on Layer-Wise Scaling (LWS) and pruning literature, we introduce three new LWS variants - Framed, Reverse, and Crown - that redistribute FFN widths and attention heads via two or three-point linear interpolation in the pre-training stage. We present the first systematic ablation of LWS an…

On the commutator scaling in Hamiltonian simulation with multi-product formulas
Kaoru Mizuta
https://arxiv.org/abs/2507.06557 https://

On the commutator scaling in Hamiltonian simulation with multi-product formulas
A multi-product formula (MPF) is a promising approach for Hamiltonian simulation efficiently both in the system size $N$ and the inverse allowable error $1/\varepsilon$ by combining Trotterization and the linear combination of unitaries (LCU). It achieves poly-logarithmic cost in $1/\varepsilon$ like LCU [G. H. Low, V. Kliuchnikov, N. Wiebe, arXiv:1907.11679 (2019)]. The efficiency in $N$ is expected to come from the commutator scaling in Trotterization, and this appears to be confirmed by the …

Scaling RL to Long Videos
Yukang Chen, Wei Huang, Baifeng Shi, Qinghao Hu, Hanrong Ye, Ligeng Zhu, Zhijian Liu, Pavlo Molchanov, Jan Kautz, Xiaojuan Qi, Sifei Liu, Hongxu Yin, Yao Lu, Song Han
https://arxiv.org/abs/2507.07966

Scaling RL to Long Videos
We introduce a full-stack framework that scales up reasoning in vision-language models (VLMs) to long videos, leveraging reinforcement learning. We address the unique challenges of long video reasoning by integrating three critical components: (1) a large-scale dataset, LongVideo-Reason, comprising 52K long video QA pairs with high-quality reasoning annotations across diverse domains such as sports, games, and vlogs; (2) a two-stage training pipeline that extends VLMs with chain-of-thought supe…

CTTS: Collective Test-Time Scaling
Zhende Song, Shengji Tang, Peng Ye, Jiayuan Fan, Tao Chen
https://arxiv.org/abs/2508.03333 https://arxiv.org/pdf/2508.03…

CTTS: Collective Test-Time Scaling
Test-time scaling (TTS) has emerged as a promising research field for enhancing the effectiveness of large language models (LLMs) without extra training. However, most existing approaches, e.g., Best-of-N and Self-Consistency rely on a single agent interacting with a reward model (SA-SR), constrained by limited capabilities of a single test-time scaling (STTS) paradigm. On the other hand, recent works demonstrate that collective-agent methods can break through the upper bound of single-agent sy…

On the Impact of Classical and Quantum Communication Networks Upon Modular Quantum Computing Architecture System Performance
Pau Escofet, Abhijit Das, Sahar Ben Rached, Santiago Rodrigo, Jordi Domingo, Fabio Sebastiano, Masoud Babaie, Batuhan Keskin, Edoardo Charbon, Peter Haring Bol\'ivar, Maurizio Palesi, Elena Blokhina, Bogdan Staszewski, Avishek Nag, Artur Garcia-S\'aez, Sergi Abadal, Eduard Alarc\'on, Carmen G. Almud\'ever

On the Impact of Classical and Quantum Communication Networks Upon Modular Quantum Computing Architecture System Performance
Modular architectures are a promising approach to scaling quantum computers beyond the limits of monolithic designs. However, non-local communications between different quantum processors might significantly impact overall system performance. In this work, we investigate the role of the network infrastructure in modular quantum computing architectures, focusing on coherence loss due to communication constraints. We analyze the impact of classical network latency on quantum teleportation and ide…

GPT-5's release was underwhelming, offering incremental improvements and failing to meet expectations, showing that pure scaling simply isn't the path to AGI (Gary Marcus/Marcus on AI)
https://garymarcus.substack.com/p/gpt-5-overdue-overhyped-and-underwhel…

GPT-5: Overdue, overhyped and underwhelming. And that’s not the worst of it.
A new release botched … and a breaking research new paper that spells trouble

HierMoE: Accelerating MoE Training with Hierarchical Token Deduplication and Expert Swap
Wenxiang Lin, Xinglin Pan, Lin Zhang, Shaohuai Shi, Xuan Wang, Xiaowen Chu
https://arxiv.org/abs/2508.09591

HierMoE: Accelerating MoE Training with Hierarchical Token Deduplication and Expert Swap
The sparsely activated mixture-of-experts (MoE) transformer has become a common architecture for large language models (LLMs) due to its sparsity, which requires fewer computational demands while easily scaling the model size. In MoE models, each MoE layer requires to dynamically choose tokens to activate particular experts for computation while the activated experts may not be located in the same device or GPU as the token. However, this leads to substantial communication and load imbalances a…

Video-RTS: Rethinking Reinforcement Learning and Test-Time Scaling for Efficient and Enhanced Video Reasoning
Ziyang Wang, Jaehong Yoon, Shoubin Yu, Md Mohaiminul Islam, Gedas Bertasius, Mohit Bansal
https://arxiv.org/abs/2507.06485

Video-RTS: Rethinking Reinforcement Learning and Test-Time Scaling for Efficient and Enhanced Video Reasoning
Despite advances in reinforcement learning (RL)-based video reasoning with large language models (LLMs), data collection and finetuning remain significant challenges. These methods often rely on large-scale supervised fine-tuning (SFT) with extensive video data and long Chain-of-Thought (CoT) annotations, making them costly and hard to scale. To address this, we present Video-RTS, a new approach to improve video reasoning capability with drastically improved data efficiency by combining data-ef…

Symmetry-protected many-body Ramsey spectroscopy: precision scaling and robustness
Sijie Chen, Jiahao Huang, Min Zhuang, Chaohong Lee
https://arxiv.org/abs/2509.08291 https://…

Symmetry-protected many-body Ramsey spectroscopy: precision scaling and robustness
Quantum entanglement is a powerful quantum resource for enhancing measurement precision beyond classical limit. % Here we propose an entanglement-enhanced symmetry-protected destructive many-body Ramsey interferometry for precise parameter estimation. % Through matching the symmetry of input states and Hamiltonian, the spectral shift of Ramsey spectroscopy caused by interparticle interaction, noise, decoherence and experimental imperfection can be mitigated for both time-independent and time-de…