Peer Revenue Past citations are no guarantee of future returns

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Friday 11 September

Systems and Control

Adaptive Agent Design

We consider an agent acting against a general non-Markovian environment. The agent maintains its agent states, but is free to choose a transition kernel across those states and optimize its state-feedback control policies. We study the bi-level agent design problem that optimizes the transition kernel and the policy it induces, given said kernel with offline data of observations and actions obtained via a behavioral policy. For general environments, we show that a soft $Q$-learning algorithm converges almost surely to the fixed point of a soft Bellman equation defined by the stationary averages that the behavioral policy and the chosen kernel induce, and we delineate what separates the resulting policy from an optimal one. In partially observed Markov decision problems, we analyze convergence properties of parametrized transition kernel design via zero-th order and Bayesian optimization techniques.
Read the full abstract

Numerical Analysis

Sharp Error Estimates for a Fully Discrete Finite Element Method for Semilinear SPDEs with Multiplicative Noise and Nonsmooth Initial Data

This article establishes sharp strong error estimates for the fully discrete approximation of semilinear parabolic stochastic partial differential equations (SPDEs) driven by multiplicative noise and subject to nonsmooth initial data. The spatial discretization is based on a standard finite element method, coupled with the linearly implicit Euler scheme in time. By mapping the diffusion operator into negative fractional spaces, our framework accommodates both trace-class and space-time white noise. For nonsmooth initial data, by decoupling the noise regularity parameter $\beta \in (0,2)$ from the initial data regularity parameter $\mu \in (0,2]$, we derive sharp regularity estimates that isolate the exact loss of initial regularity into an integrable temporal singularity. Furthermore, we establish sharp strong convergence rates of $O(h^{\beta-\varepsilon} + k^{\frac{1}{2}\min\{\beta-\varepsilon, 1\}})$ for $\varepsilon>0$ away from $t = 0$. Finally, we consider physically relevant stochastic models, such as the modified Langmuir fractional surface coverage model and the parabolic Anderson model, in our numerical experiments to confirm the theoretical convergence rates.
Read the full abstract

Audio and Speech Processing

Overview and Meta-Analysis of DCASE 2026 Challenge Task 6: Audio Moment Retrieval from Long Audio

This paper presents an overview of the Detection and Classification of Acoustic Scenes and Events (DCASE) 2026 Challenge Task 6, Audio Moment Retrieval (AMR) from Long Audio. Given a several-minute-long audio recording and a free-form text query, AMR aims to retrieve temporal moments in the recording that match the query, where each moment is represented by a pair of start and end timestamps. This task requires effective cross-modal alignment and long-range temporal modeling. We describe the task definition, the evaluation metrics, the development and evaluation datasets, and a baseline system that combines a pre-trained MS-CLAP feature extractor with a Detection Transformer (DETR)-based moment-detection network. On the development data, the baseline trained on a manually annotated dataset and a synthetic dataset achieved Recall1@0.7 of 13.56%, indicating that AMR in long audio remains a challenging problem. The challenge attracted 21 teams, which submitted 59 systems in total. The three best systems achieved Recall1@0.7 of 48.59%, roughly 3.5 times the baseline score. The results show that strengthening the audio-text feature extractor and the moment-detection network led to substantial performance improvements. Furthermore, the top three teams boosted performance by applying confidence score calibration or ensembling across different temporal resolutions of features.
Read the full abstract

All authors Hokuto Munakata, Tatsuya Komatsu, Keisuke Imoto, Taichi Nishimura, Huang Xie, Tuomas Virtanen

Combinatorics

The affine Tverberg theorem revisited

We prove the Bárány–Kalai–Tverberg conjecture extending the affine version of Tverberg's theorem to simplicial balls and polytopes other than the simplex: For any affine map $\phi : P\to \mathbb R^d$ from a convex polytope $P$ of dimension $N=(d+1)(r-1)$ with $r\geq 2$, there exist $r$ pairwise disjoint faces $F_1,\ldots,F_r\subset\partial P$ such that $\phi(F_1)\cap\dots\cap\phi(F_r)\neq\varnothing$. A similar statement holds for a simplicial ball $P$ with $\phi$ assumed affine on all of its faces.
Read the full abstract

Artificial Intelligence

When Does AI Augment Work? A Workflow-Level Framework for Human-Agent Collaboration

We aim to characterise the value of artificial intelligence in the workplace. Current studies largely measure this value in terms of the current automation capabilities and public adoption of AI. However, such metrics ignore the greater impacts of human–agent collaboration in transforming the nature of work. To account for this, we must expand the scope of our analysis beyond atomised tasks of today, and instead focus on how AI can augment entire workflows of the future. To ground this analysis, we establish a precise definition of AI augmentation comprising six conditions, spanning durable net value, meaningful human control, accountability and recovery, and long-term human development through learning, career pathways, and job purpose. We elaborate on these conditions and apply the framework in a case study of AI-mediated social surveys. We conclude by outlining how organisations, researchers, and government leaders can use this framework to make sense of the future of work.
Read the full abstract

All authors CIVIC-AI Collaboration: Jiaying Wu, Caleb Ziems, Raymond Chan, Nancy F. Chen, Corlyss Chua, Gerard Chung, Jungpil Hahn, Wee Sun Lee, Zhengyuan Liu, Jamie Ng, Desmond C. Ong, Jeryl Ong, Da Ren Soon, Tianqi Song, Zhi-Xuan Tan, Sixing Tao, Emily Yang, Yajing Yang, Stella Xin Yin, Min-Yen Kan, Diyi Yang

Signal Processing

Secrecy Sum-Rate Maximization in Finite Blocklength IRS-aided Systems With Perfect and Imperfect CSI

The rapid growth of the Internet of Things (IoT) requires efficient and secure communication technologies to enable ultra-reliable low-latency (URLLC) communication applications. Intelligent reflecting surfaces (IRS) have emerged as a promising solution to enhance IoT network secrecy performance by improving signal quality for legitimate devices (Bob), while thwarting the eavesdropper (Eve) interception. However, ensuring secrecy across multiple users, given their diverse channels and locations, is challenging, especially under finite blocklength (FBR) constraints, which are common in IoT networks. This paper investigates the secrecy performance of IRS-aided URLLC systems under three channel state information (CSI) scenarios: perfect CSI, imperfect CSI, and unknown eavesdropper's CSI. We first formulate a non-convex optimization problem to maximize the system's sum secrecy rate (SSR) by jointly optimizing the transmitter's beamforming and the IRS's passive reflective elements, while maintaining FBR-related latency and transmission duration constraints. For imperfect CSI, the semi-infinite uncertainty constraints are transformed into finite linear matrix inequalities (LMIs) via a successive convex approximation (SCA)-based approach, and the proposed algorithm is proven to converge to a locally optimal solution with low computational complexity.
Read the full abstract

All authors Monir Abughalwa, Nguyen Van Huynh, Diep N. Nguyen, Dinh Thai Hoang, Eryk Dutkiewicz

Number Theory

$S_h$-sets in abelian groups

For a positive integer $h$, a subset $A = \{a_1, \dots, a_k\}$ of an additive abelian group $G$ is called an $S_h$-set of size $k$ if all sums of $h$ distinct elements in $S$ are distinct. For fixed positive integers $h$ and $k$, let $v_h(k)$ denote the order of the smallest abelian group containing an $S_h$-set of size $k$. A lower bound for $v_2(k)$ is known. In this paper, we establish a lower bound for $v_3(k)$. Using our argument for $h=2$, we recover the known bound for $v_2(k)$.
Read the full abstract

Plasma Physics

Nonuniform Hydrodynamic Plasma-Wave Instability in a Gated Electron Channel

We report a theoretical and numerical study of plasma-wave instability in a current-driven InGaAs/GaAs gated electron channel. A finite drain current drives the channel away from a uniform plasma-wave cavity: the carrier density and plasma velocity decrease toward the drain, while the electron drift velocity increases. This current-induced spatial nonuniformity modifies both the propagation time and amplification of plasma waves and becomes increasingly important as the sonic regime is approached. Using a hydrodynamic description combined with the unified charge-control model, we determine the nonuniform steady state and analyze the resulting plasma-wave frequency and instability growth rate. The oscillation frequency is found to be governed by wave propagation through the entire channel rather than by the local drain conditions, leading to a pronounced current-induced frequency reduction that is not captured by a uniform-channel approximation. The instability results from the competition between boundary-induced plasma wave gain and distributed losses due to momentum relaxation and electron viscosity. The finite-viscosity theory shows good quantitative agreement with the simulation data over the subsonic regime and regularizes the singular behavior of the inviscid description near the sonic point. At higher currents, the dynamics become increasingly sensitive to the near-sonic structure and contact boundary conditions, indicating that the eventual suppression of the instability cannot be attributed to bulk hydrodynamics alone.
Read the full abstract

Portfolio Management

Large Signal Libraries: Equal-Weight Limits and the Divergent Spectra of Signals and PnL

An ensemble of roughly 3,000 signals over 20 assets was reported to have approximately 90% correlation with the leading component of the asset-space return structure. Does having about 158 signals per available linear dimension explain that alignment? The population answer depends on the research process's design distribution and its relation to returns: crowding alone imposes neither a nonzero mean nor agreement with a principal component. The population theory developed here distinguishes four objects: the equal-weight signal, signal principal components, equal-weight profit and loss (PnL), and PnL principal components. The return operator in the motivating observation is a separate object. Independent libraries converge to their design mean; exchangeable libraries can retain a random conditional mean. Cross-sectional signals over $d$ assets, once demeaned and normalized, lie on the unit sphere $S^{q-1}$ of a $q$-dimensional space, $q=d-1$. Under axial symmetry, their nonzero mean is signal-cloud PC1 exactly when its longitudinal second moment exceeds $1/q$, the isotropic energy share. Residual-and-gap bounds quantify approximate alignment. Combining design weights into signals contracts the tangent of their angle to PC1 to at most $\sqrt{\lambda_2/\lambda_1}$ times its value, where $\lambda_1>\lambda_2$ are the leading signal-kernel eigenvalues; the factor is sharp. A target-aligned frame separates transverse signal geometry, which PnL discards, from dispersion weighting and temporal centering, which also change the spectrum. A reproducible synthetic example illustrates the geometric threshold, and a proposed empirical program separates library growth from limited-history estimation. No market-data empirical results are presented.
Read the full abstract

General Relativity and Quantum Cosmology

A semicoherent resampling method for long-transient gravitational wave searches with applications to subsolar-mass primordial black-hole binaries

Resampling removes the modeled phase evolution of a frequency-evolving signal, transforming it into a monochromatic one. We introduce a novel implementation for long-transient gravitational-wave searches that evaluates the Fourier spectrum of the resampled data using a type-I non-uniform fast Fourier transform (NUFFT). This implementation can be up to two orders of magnitude faster than some previous implementations, depending on coherent length, sampling rate, and desired accuracy. To demonstrate its application, we benchmark the method with a semicoherent search for long-duration inspiral signals from subsolar-mass primordial black hole binaries. For a search over $40$–$60\,\mathrm{Hz}$ and chirp masses of $5\times10^{-4}$–$10^{-1}\,M_\odot$, with a coherent duration of $30\,\mathrm{s}$ and a realistic computing cost, the estimated horizon distance exceeds the Galactic Center for all of the parameter space considered and exceeds Andromeda for chirp masses above approximately $2\times10^{-2}M_\odot$. These results establish NUFFT-based resampling as a computationally efficient and broadly applicable approach for searches for modeled, frequency-evolving gravitational-wave signals.
Read the full abstract

All authors Neil Lu, Cristiano Palomba, Ornella J. Piccinni, Ling Sun

Computation and Language

Zipbench: Low-Cost Framework for Compressing Comprehensive Benchmarks of Large Language Models

Comprehensive benchmark suites are essential for improving large language models (LLMs), but many widely used benchmarks are redundant, making evaluation unnecessarily expensive. Although recent benchmark compression methods (BCMs) can mitigate this cost, many strong BCMs rely on large collections of per-sample evaluation results from numerous LLMs to identify representative samples. Building such collections is also expensive unless they are already public, making these methods difficult to extend to newly released benchmarks. To address this challenge, we present ZipBench, a simple and low-cost BCM with theoretical error and rank-consistency guarantees. ZipBench evaluates only a small set of anchor LLMs, synthesizes pseudo evaluation results to broaden coverage, learns compact sample representations, and selects a small yet representative subset. Building on it, we create ZipBench Zoo, a collection of compact versions of 100+ benchmark proxies spanning text, multimodal, and agent tasks. These benchmark achieve mean absolute errors of 0.002–0.02 and average Spearman correlations of ~0.98 with the full benchmarks. Overall, ZipBench reduces the cost of both LLM evaluation and compact benchmark construction, lowering the barrier to broad LLM research for compute-constrained researchers. The code has been released in https://github.com/MilkThink-Lab/ZipBench.
Read the full abstract

All authors Zhongzhan Huang, Junxin Li, Guoming Ling, Yupei Lin, Shanshan Zhong, Hefeng Wu

Plasma Physics

First Experimental Evidence of Helicon Current Drive

Helicon current drive is an attractive solution for driving current to sustain steady state tokamak operation in reactor conditions. Dedicated DIII-D experiments have been conducted with a MW-level helicon system and successfully demonstrated core power deposition and current drive with helicon waves launched via a traveling wave antenna. The profile of the measured electron temperature response to helicon power injection is in good agreement with time-dependent integrated modeling that incorporates ray tracing and the effects of thermal transport simultaneously. When the helicon power is injected continuously to drive co-$I_p$ current, the reconstructed safety factor profile flattens significantly faster and sawteeth are triggered earlier than in comparison shots where the helicon is replaced by a comparable amount of electron cyclotron heating. Calculation of the helicon-driven current profile yields a peaked profile in the core, consistent with the observed power deposition profile and in good agreement with ray tracing predictions. Taken together, these experimental results represent strong evidence for the first definitive observation of auxiliary current drive due to helicon waves on any device.
Read the full abstract

All authors J.B. Lestz, R.I. Pinsker, B. Van Compernolle, S.X. Tang, A. Dupuy, A.M. Garofalo, L. McAllister, C.P. Moeller, C.C. Petty, M. Porkolab, M.P. Ross, G. Rutherford, A. Torrezan

Computer Vision

Partition-Invariant Tuning for 3D Scene Understanding

Scene-level point cloud understanding remains challenging due to diverse geometries and spatial layouts. While pre-trained 3D point cloud foundation models (PFMs) offer strong transferability, full fine-tuning (FFT) incurs substantial computational and storage costs. Parameter-efficient fine-tuning (PEFT) provides a promising alternative, but existing PEFT methods largely focus on object-level point clouds and overlook serialization-induced partition variations in large-scale scenes. To address this issue, we propose PointPiT, a partition-invariant tuning framework for scene-level point clouds. Specifically, a Scene-aware Structural Adapter (SSA) integrates local geometric patterns with global scene context to mitigate partition-induced representation shifts. Moreover, Gradient Subspace Optimization (GSO) selects informative and partition-stable update directions, suppressing partition-dependent variations during optimization. Extensive experiments across multiple scene-level benchmarks demonstrate that PointPiT achieves competitive or even superior performance to full fine-tuning with less than 1% of backbone's parameters, while achieving consistent state-of-the-art performance among representative PEFT methods.
Read the full abstract

All authors Hongqiang Lin, Tianle Wang, Shuiwang Li, Dongxu Zhang, Yiding Sun, Zihao Guo, Dongfu Yin

Artificial Intelligence

TripPattern: A Pattern-based Text Watermarking Method for Large Language Models

Text watermarking techniques have gained significant attention for identifying machine-generated text and mitigating risks from large language models (LLMs). Existing methods typically divide an LLM's vocabulary into green and red tokens, but encouraging generation toward green tokens can reduce text quality and naturalness. To address this, we propose TripPattern, a watermarking framework that formulates text watermarking as a pattern-based matching task using three vocabulary partitions. TripPattern divides the vocabulary into one neutral group and two pattern groups. During generation, the model alternates token selection between the two pattern groups to embed detectable patterns, while neutral tokens are selected independently to improve flexibility and preserve naturalness. For detection, TripPattern uses pattern-based statistical tests that provide interpretable p-values by measuring how often adjacent tokens alternate between the pattern groups. Theoretical analysis and empirical evaluations on four multilingual datasets show that TripPattern maintains LLM generation quality while achieving robust watermark detectability.
Read the full abstract

All authors Sangjun Moon, Dasom Choi, Jingun Kwon, Hidetaka Kamigaito, Taro Watanabe, Manabu Okumura

Computation and Language

AMDKernelVault: Large-Scale Datasets and Agentic Training for AMD GPU Kernel Optimization

We introduce AMDKernelVault, an open HIP and Triton kernel corpus and training framework for recent AMD CDNA GPUs. Existing LLM-based kernel agents are largely CUDA/NVIDIA-centric and often depend on repeated frontier-LLM calls for generation, reflection, and optimization. To address this gap, we develop HIPKernelGen and TritonKernelGen, agent-driven pipelines that transform PyTorch references into HIP or Triton kernels, compile and validate candidates under ROCm, and latency-profile them on AMD hardware. The corpus contains 62,153 execution-verified HIP kernel samples, 2,377 production-grounded ROCm Libraries QA entries, and 39,893 Triton kernels. We further train Qwen3-8B with supervised fine-tuning and execution-aware reinforcement learning as a demonstration of the corpus's utility. Under fixed evaluation budgets, it achieves the highest correctness among the compared models on PyTorch-to-HIP (34.0% Pass@1), TritonBench-G (33.2% Corr@3), and ROCmBench (41.94% Corr@3), but does not uniformly lead compilation or speed metrics. The corpus and documentation are available at https://huggingface.co/datasets/amd/AIG-Datasets, and the associated training and kernel-generation code is available at https://github.com/AMD-AGI/hip_kernel_llm_lab.
Read the full abstract

All authors Ji Liu, Saptarshi Majumder, Yiqing Huang, Wenwen Ouyang, Umang Pandey, Zeping Li, Chushi Chen, Zihao An, Puyuan Yang, Zekai Li, Sina Rafati, Ziqiong Liu, Pratik Prabhanjan Brahma, Dong Li, Zicheng Liu, Sharon Zhou, Emad Barsoum

Machine Learning

A Differentially Private Federated Proximal Optimization Framework for Customer Churn Prediction in Heterogeneous Federated Telecom Networks

Customer churn is one of the major issues in the telecommunication industry. To predict customer churn, conventional centralized machine learning approaches have been widely used. This centralized approach requires customer data to be stored in a central repository, which raises privacy concerns and may violate data protection regulations. Federated learning addresses this problem by allowing multiple telecom operators to collaboratively train a global model without transferring their raw customer data. However, real-world customer data are often heterogeneous (non-IID), which may negatively affect the performance of standard federated learning. Trained models can also suffer from privacy attacks. To address those issues, we propose a Differentially Private (DP) based Federated Proximal optimization (FedProx) framework. All experiments were performed on two publicly available telecom churn datasets. We trained Federated Averaging (FedAvg), DP-FedAvg, FedProx, and the proposed DP-FedProx framework. For baseline comparison, we also used several centralized and local models. To evaluate the models, we employed seven widely used evaluation metrics. The experimental results show that the FedProx based models consistently outperform the FedAvg based models. Compared with the best centralized model, the proposed DP-FedProx framework achieves competitive prediction performance with only a small reduction in accuracy while providing privacy guarantees. To explain our model, we conducted SHAP analysis which shows that DP-FedProx method priorities revenue group features. These results indicate that the proposed DP-FedProx framework provides a practical balance between prediction performance and data privacy protection.
Read the full abstract

Computation and Language

Not All Speech Is Intent: Adaptive Self-Correcting Inference Layer for Post-ASR False Wake-Up

False wake-up activations remain a persistent challenge in conversational AI. Speech phonetically similar to a device's wake word can produce a syntactically valid and semantically coherent ASR transcript that the assistant incorrectly executes. Most existing systems make a single intent decision in isolation, without a mechanism to learn from recurring errors over time or adapt to individual users through personalized learning. We introduce the Feedback-Driven Adaptive Self-Correcting Inference Layer (ASCIL), a complementary post-ASR correction framework that re-evaluates wake-up intent before response generation by fusing acoustic embeddings, linguistic cues, device context, and patterns from past misclassifications. ASCIL interprets implicit signals, including hesitation, disengagement, and silence, and explicit signals, including cancellation and repetition, as automatically inferred, noisy behavioral indicators of potential misclassification. These signals drive online pattern updates without manual annotation, whereas the intentional/unintentional reference labels used for offline evaluation are human-annotated. It generalizes from prior errors, applies corrective adjustments at inference time, and continuously updates in parallel with natural-language execution. Evaluated on a proprietary dataset of 3,667 interactions with human-annotated intentional/unintentional reference labels spanning 14 acoustic and contextual conditions, ASCIL achieves 54.27% relative error reduction on a session-disjoint subset constructed from baseline failures, and up to 24.39% relative error reduction at threshold 0.90 on the issue-tagged evaluation slice. These gains are achieved while improving intentional acceptance rates, with a median added latency below 60 ms in the reported benchmark.
Read the full abstract

Statistical Methodology

Gibbs Sampling for Bayesian Generalized Poisson Matrix Factorization

Generalized Poisson matrix factorization (GPMF) is a matrix factorization method for count data with overdispersion based on the generalized Poisson distribution. While GPMF provides point estimates of the model parameters through maximum likelihood estimation, it does not quantify estimation uncertainty. In this paper, we propose a Bayesian extension of GPMF, referred to as Bayesian GPMF, and develop a Gibbs sampler for posterior inference. The proposed method is based on a compound Poisson representation of the generalized Poisson distribution, which introduces latent variables and yields closed-form full conditional posterior distributions for all model parameters. To efficiently sample the overdispersion parameter, we derive an infinite mixture representation of the exponentially tilted beta (EBeta) distribution and develop a finite approximation based on this representation. Simulation studies demonstrate that the proposed Bayesian GPMF achieves smaller mean squared errors than the conventional GPMF while providing credible intervals with empirical coverage probabilities close to the nominal level. Furthermore, the proposed finite approximation attains estimation accuracy comparable to Sampling/Importance Resampling (SIR) while requiring less computation. An application to football event data further illustrates the usefulness of Bayesian GPMF for extracting interpretable latent structures and quantifying estimation uncertainty. These results demonstrate that the proposed framework provides an effective Bayesian approach to generalized Poisson matrix factorization.
Read the full abstract

Solar and Stellar Astrophysics

Magnetic Fields and Asymmetric Accretion in the Class 0 Protostellar System L1527 IRS

Magnetic fields play a crucial role in regulating the collapse of dense molecular clouds, shaping the early stages of protostellar evolution. We present polarization observations in submillimeter wavelengths with the SCUBA-2/POL-2 instrument on James Clerk Maxwell Telescope toward the Class 0 protostellar system L1527 IRS (IRAS 04368+2557). The magnetic field morphology varies across the core of L1527 IRS. Magnetic fields in the eastern region are perpendicular to the outflow axis, while those in the western region show a pinched morphology, overall aligned with the outflow cavity. These distinct bipolar outflow regions also have a clear color difference in near infrared observed by James Webb Space Telescope. In addition, the spectral index derived from 450 {\mu}m and 850 {\mu}m observations reveals that the northwestern region of the protostellar system is colder and denser compared to the southeastern region. Furthermore, in the large-scale structures and magnetic fields revealed by Herschel and Planck, we find a ~0.1 pc scale filamentary structure parallel to the large-scale magnetic fields in the eastern region and a relatively isotropic mass distribution in the western region. Based on these results, we propose an accretion scenario of the protostellar system in which an asymmetric mass distribution causes the distinct features observed in near-infrared and submillimeter wavelengths.
Read the full abstract

All authors Hanju Nam, Woojin Kwon, Youngwoo Choi, Ian W. Stephens, Leslie W. Looney

Statistical Mechanics

Tensor-network Monte Carlo approach based on time-evolving block decimation

We propose a tensor-network Monte Carlo (TNMC) approach for unitary evolution following the compression sequence of the time-evolving block decimation (TEBD) algorithm. In the TNMC approach, the obtained results contain evaluable statistical errors rather than truncation errors, unlike ordinary singular-value-decomposition-based methods such as the TEBD algorithm. Consequently, one can estimate unbiased expectation values within statistical errors even with a finite bond dimension. Since the sampling scheme is introduced in the simulations of unitary evolution, the proposed Monte Carlo scheme may suffer from a sign problem. We observe that the sign problem can be mitigated by increasing the bond dimension. We apply the proposed TNMC approach to the Hamiltonian and the Floquet dynamics. Numerical experiments show that the TNMC approach can estimate accurate expectation values of observables even when the TEBD method with the same bond dimension cannot. The proposed approach can be a new direction for improving the classical simulatability of unitary evolution.
Read the full abstract

Astrophysics of Galaxies

The constrainability of galaxy positions in the $M_*$ – SFR plane from SED fitting

Using a sample of 6,029 $z<0.06$ galaxies from the GAMA survey, we characterise the spread of galaxy property inferences on the M$_*$ – SFR plane via Spectral Energy Distribution (SED) fitting. We analyse 24 different model configurations using the stellar population library code ProGeny, and the SED-fitting code ProSpect, and quantify how modelling assumptions affect inferred stellar masses, star formation rates, and the resulting classification of galaxies. We find that the reproducibility of inferred galaxy positions depends strongly on location within the $M_*$ – SFR plane, with uncertainties increasing dramatically below $\log_{10}(\rm{sSFR}/\rm{yr}^{-1}) \sim -11$. This boundary closely tracks the transition between robustly and poorly constrained SFR estimates, regardless of whether SFRs are derived from SED fitting or H$_{\alpha}$ measurements. We find that star-bursting galaxies cannot be robustly isolated from SED fitting alone, while the separation of star-forming and passive galaxies is most consistently conducted by a cut in sSFR, rather than a dex cut relative to either a curved or linear SFS (Star Forming Sequence). We further show that transitioning (or Green Valley) selections using SED fitting are highly unstable, with only $\sim$25% of the sample being repeatedly selected, on average.
Read the full abstract

All authors Sabine Bellstedt, Aaron S. G. Robotham, Luke J. M. Davies, Luca Cortese, Simon P. Driver

Artificial Intelligence

Beyond Vector Similarity: Hierarchical Context-Aware Graph RAG vs Standard RAG in Enterprise Code Migration

As enterprises modernize legacy monolithic systems to microservices, Large Language Models (LLMs) are heavily utilized for automated code translation. However, traditional vector-based Retrieval-Augmented Generation (Standard RAG) struggles to capture topological relationships. It fetches isolated chunks that sever inheritance chains, leading to high compilation failure rates. This paper introduces a Hierarchical Context-Resident Graph (HCRG) methodology to resolve these limitations. Our pipeline uses tree-sitter for Abstract Syntax Tree (AST) extraction, maps architectural edges into a Google Cloud Spanner Property Graph, and serializes this structure into a Gemini Context Cache for topological, parent-first code translation. We shift evaluation from naive text-overlap to a custom 7-metric Software Engineering framework. Traditional metrics like CodeBLEU (which scored 91% for both methods) effectively masked Standard RAG's structural failures behind syntactically plausible but broken code. Empirically, Graph RAG decisively mitigates dependency loss: API hallucination rates dropped from 56.4% to 16.2%, Dependency Resolution Quality improved from 34.8% to 65.9%, and Parent-Child Consistency rose from 26.7% to 45.5%. However, Graph RAG introduces specific trade-offs. The dense global context causes defensive over-engineering by the LLM, reducing Cyclomatic Complexity Consistency from 71.6% to 46.7%, and slightly degrades Docstring Preservation (67.0% to 61.0%). Ultimately, while trading code complexity for reduced hallucinations, Graph RAG provides a substantially more viable, architecturally sound path for automated enterprise codebase modernization.
Read the full abstract

All authors Nilesh Jaiswal, Aniket Agrawal, Arjit Shukla, Divya Malhotra, Saurabh Garg, Suchit Puri, Suddhasatwa Bhaumik

Optimization and Control

Linear Exponential Quadratic Gaussian Covariance Steering

We formulate and analyze the linear exponential quadratic Gaussian (LEQG) covariance steering problem in continuous time over a given deadline (finite time horizon). The solution for this problem can be seen as a risk-sensitive Schrödinger bridge between Gaussian endpoints in the linear quadratic setting. Unlike the risk-neutral case, the LEQG covariance steering controller–still a linear state feedback–can no longer be written in closed form. We show that the optimal controller is parameterized by a symmetric matrix solving an algebraic equation that encodes the implicit dependence on the risk-sensitivity parameter. We explain how the structure of this optimal controller significantly generalizes the existing results for the risk-neutral case. Building on these results, for the matched noise and input channel case, we prove the existence-uniqueness of solution for the LEQG covariance steering problem in the neighborhood of the known risk-neutral optimal solution. We give an illustrative numerical example.
Read the full abstract

Numerical Analysis

A Structural Relationship Between Crouzeix-Raviart Immersed Finite Elements for Elliptic and Stokes Interface Problems

In this paper, we study the structural relationship between immersed finite element (IFE) spaces for scalar elliptic and Stokes interface problems on unfitted meshes. Although IFE methods have been widely developed for individual partial differential equation models, the algebraic connection between scalar and vector IFE spaces has not been systematically explored. We establish a precise unisolvence relationship between the immersed Crouzeix-Raviart (CR) element for elliptic interface problems and the immersed CR-$P_0$ element for Stokes interface problems in both gradient and stress formulations. By analyzing the block structure of the local IFE matrices, we show that the determinant of the Stokes IFE matrix factorizes in terms of the determinant of the corresponding elliptic IFE matrix. This factorization reveals an intrinsic algebraic link between the two classes of immersed elements and explains the unisolvence of the Stokes IFE spaces through the scalar elliptic case. The same result extends naturally to three dimensions and significantly simplifies the corresponding analysis. Numerical experiments confirm optimal convergence rates for both velocity and pressure in three-dimensional Stokes interface problems.
Read the full abstract

Solar and Stellar Astrophysics

Superoutbursts and Superhumps of Cataclysmic Variables observed with TESS

Superoutbursts and superhumps are characteristic signatures of SU UMa-type cataclysmic variables, yet the full diversity of superhump evolutionary behaviour remains poorly constrained owing to the limitations of ground-based photometric monitoring. Here we report a systematic analysis of 30 SU UMa-type dwarf novae observed with the Transiting Exoplanet Survey Satellite, covering 37 superoutburst events. We detect coherent superhump signals in 29 systems-including five first-time detections-and determine or refine orbital periods for 13 objects. The high-cadence light curves clearly resolve the canonical three-stage (A-B-C) superhump period evolution. Using Stage A periods, we derive dynamical mass ratios for 18 systems, consistent with established cataclysmic variable evolutionary tracks. The measured Stage B period derivatives match the range of values from ground-based campaigns, with prominent positive drifts concentrated in short-period, low-mass-ratio systems. Repeated superoutbursts yield consistent evolutionary patterns, demonstrating that superhump morphology is an intrinsic, repeatable property. Two systems break the standard template: RZ LMi shows an inverted stage sequence inconsistent with classical precession theory, while ASASSN-14kj exhibits no stage evolution but a coherent long-period modulation. Persistent superhumps during normal outbursts in three systems provide direct evidence that eccentric disk structures can survive beyond their parent superoutbursts, revealing a decoupling between thermal accretion and tidal eccentricity cycles. These results demonstrate the power of space-borne photometry for probing accretion disk dynamics and testing precession models across the cataclysmic variable parameter space.
Read the full abstract

All authors Qi-Bin Sun, Sheng-Bang Qian, Li-Ying Zhu, Qin-Mei Li, Wen-Ping Liao, Fu-Xing Li, Wen-Xu Lin, Min-Yu Li, Ping Li, Fang-Zhou Yan, Zi-Mo Li, Tao Wang, Mao-Han Zhang

Materials Science

Enhancing Conformality in Atomic Layer Deposition through Low Growth Per Cycle

Atomic layer deposition (ALD) is a key enabling technology for advanced microelectronics as it enables the growth of functional thin films on complex three-dimensional substrates with unmatched atomic-scale precision. Conformal film growth in high-aspect-ratio (HAR) structures is limited by slow diffusion of reactant molecules into deep, narrow features. This work elucidates an approach to grow conformal films faster by investigating the relationship between the ALD growth per cycle (GPC) and the film penetration depth in HAR structures through modeling and experiments. Diffusion-reaction simulations reveal, under Knudsen diffusion conditions, an inverse square root relationship between the film penetration depth and GPC. The prediction is validated experimentally with a model zinc oxide ALD process on rectangular lateral HAR structures, using an inhibitor molecule to decrease the GPC. While ALD is traditionally optimized for "high GPC," this work shows that "low GPC" may increase efficiency when conformality is key.
Read the full abstract

All authors Christine Gonsalves, Jorge A. Velasco, Ahmed Othman, Ville Miikkulainen, Riikka L. Puurunen

Artificial Intelligence

EvoRS: On-Policy Self-Evolution of Reward Systems for Open-Ended Reinforcement Learning

Open-ended reinforcement learning often relies on rubric-based rewards for tasks without directly verifiable answers. Yet the policy and reward system form a dynamic feedback loop: as the policy optimizes the current reward, an initially useful reward system may become unreliable due to reward hacking or reduced response discriminability. The reward system should therefore evolve rather than remain fixed during training. Existing dynamic-rubric methods adapt evaluation criteria, but reward failures can also arise from scoring mechanisms or signal composition. We introduce EvoRS, a self-evolving RL framework that evolves the reward system from on-policy experience, representing it as an executable Reward-DAG. Specifically, an agentic designer updates this system from on-policy rollouts and reward traces to maintain train-time reliability. Across writing and roleplay, EvoRS achieves the best quality under all three judges, outperforming the policy by \(2.107\) and \(4.767\) points, respectively, while reducing reward hacking and coverage failures and preserving reward informativeness. Ablations confirm that a comprehensive fixed reward system cannot remain reliable in open-ended tasks and must evolve throughout training.
Read the full abstract

All authors Weiyuan Li, Aili Chen, Xintao Wang, Yikai Zhang, Qingqing Dong, Jinghan Xu, Hongru Hou, Wenxuan Zhao, Chengkun Lang, Jun Gao, Yuanli Guo, Hongcheng Guo, Yanghua Xiao, Deqing Yang

Applied Physics

Current-dipole detection of microscale metal defects

An effective electric current dipole provides a compact description of localized current disturbances and has been employed across a broad range of physical systems and length scales. Despite its extensive use in macroscopic systems, its applicability to microscale material diagnostics remains unexplored. In this study, we demonstrate that the current-dipole representation provides an effective physical framework for the detection and characterization of microscopic defects in conductive materials. When an external current is applied to a metal, defects locally perturb the current distribution, thus generating an effective in-plane current dipole that yields a characteristic magnetic-field pattern. By measuring the magnetic-field distribution above the metal and fitting it with a dipole model, we reconstruct the position and effective strength of the dipole, thereby enabling the inference of the defect location and an effective defect-volume metric. Using wide-field magnetic imaging as a measurement platform, we observe a defect-associated magnetic signal from a 38-um-long defect on the front surface of copper samples and identify a defect on the back surface using a 0.5-mm-thick copper plate. We further examine the applicability of this approach to magnetic materials, where permeability contrasts modify the field distribution. Our results establish current-dipole imaging as a general physical framework for magnetic detection of defects and highlight its potential for nondestructive inspection across a wide range of length scales.
Read the full abstract

All authors Eikichi Kimura, Eisuke Oba, Yu Saito, Yuto Yamakawa, El Mustapha Mansouri, Takatada Saito, Seiji Goto, Hiroshi Mashima, Keigo Arai

Software Engineering

Hieronym: Leveraging Hierarchical Multi-Source Information for Function Renaming in Stripped Binary

Function renaming in stripped binaries can substantially assist reverse engineers by improving code readability, yet it is a challenging task. The difficulty stems from the need to accurately capture function semantics from low-level binary code across diverse instruction sets, architectures, and compiler optimizations, and to express these semantics in concise, human-readable names. Existing approaches either inadequately capture comprehensive function semantics or exhibit limited generalization to previously unseen binaries. In this paper, we present Hieronym, a generative large language model (LLM)-based framework for stripped binary function renaming. Hieronym adopts a hierarchical summarization-driven domain adaptation strategy and integrates multi-source information, including global binary context, local calling context, and intrinsic function semantics, to enhance the LLM's understanding of binary code. To enable systematic evaluation, we further propose a dual-layer evaluation framework that incorporates both token-level and whole-name-level metrics. We evaluate Hieronym on binary functions compiled with four compiler optimization levels (O0-O3) for four architectures (x64, x86, ARM, and MIPS). Experimental results demonstrate that Hieronym significantly outperforms state-of-the-art methods, achieving token-level improvements of 50.12% in precision, 41.75% in recall, and 45.10% in F1-score, as well as a 79.94% improvement in name-level accuracy, while also exhibiting strong generalization capability. Moreover, experiments on real-world malware samples further validate the practical effectiveness of Hieronym in security-critical scenarios.
Read the full abstract

All authors Xiaoling Zhang, Jian Sun, Dawei Wang, Chongyu Wang, Li Chen, Zhaoteng Yan, Peipei Liu, Lixiao Zhang, Dan Li