arXiv preprints from January 1, 2026 through July 21, 2026 — 15:19:01 EST
The Muon optimizer, a matrix-structured algorithm that leverages spectral orthogonalization of gradients, is a milestone in the pretraining of large language models. However, the underlying mechanisms of Muon -- particularly the role of gradient orthogonalization -- remain poorly understood, with very few works providing end-to-end...
Magnetic fields and turbulence may play a key role in the evolution of protoclusters, influencing the formation of dense cores and stars. Here, we examine the morphology of the magnetic fields in the G327.29 protocluster using both the velocity gradient technique (VGT) extracted from molecular line emissions and linear polarization in...
We investigate visual reasoning limitations of both multimodal large language models (MLLMs) and image generation models (IGMs) by creating a novel benchmark to systematically compare failure modes across image-to-text and text-to-image tasks, enabling cross-modal evaluation of visual understanding. Despite rapid growth in machine...
We study the thermalization properties of a fully nonlinear lattice model originally derived from the two-dimensional cubic defocusing nonlinear Schrödinger equation (NLS) using analytical and numerical methods. Our analysis reveals both ergodic and nonergodic regimes; importantly, we find broad parameter ranges where the dynamics is...
Building upon previous research, this paper further explores the topological foundations for maintaining universal rules within ultra-large-scale societies. It finds that in small-scale societies, absolute egalitarianism and the rule of law can be compatible through peer monitoring within a fully connected network. However, in...
The deployment of large language models (LLMs) has raised security concerns due to their susceptibility to producing harmful or policy-violating outputs when exposed to adversarial prompts. While alignment and guardrails mitigate common misuse, they remain vulnerable to automated jailbreaking methods such as GCG, PEZ, and GBDA, which...
Efficiently serving Large Language Models (LLMs) with persistent Prefix Key-Value (KV) Cache is critical for applications like conversational search and multi-turn dialogue. Serving a request requires loading the pre-computed prefix KV cache and generating the first token, defined as the Re-Prefill Phase. Offloading this shared prefix...
Large language models (LLMs) are increasingly deployed over knowledge bases for efficient knowledge retrieval and question answering. However, LLMs can inadvertently answer beyond a user's permission scope, leaking sensitive content, thus making it difficult to deploy knowledge-base QA under fine-grained access control requirements....
We present S$^2$Voice, the winning system of the Singing Voice Conversion Challenge (SVCC) 2025 for both the in-domain and zero-shot singing style conversion tracks. Built on the strong two-stage Vevo baseline, S$^2$Voice advances style control and robustness through several contributions. First, we integrate style embeddings into the...
This paper presents PRIMAL, a processing-in-memory (PIM) based large language model (LLM) inference accelerator with low-rank adaptation (LoRA). PRIMAL integrates heterogeneous PIM processing elements (PEs), interconnected by 2D-mesh inter-PE computational network (IPCN). A novel SRAM reprogramming and power gating (SRPG) scheme...
Predicting highly-cited papers is a long-standing challenge due to the complex interactions of research content, scholarly communities, and temporal dynamics. Recent advances in large language models (LLMs) raise the question of whether early-stage textual information can provide useful signals of long-term scientific impact. Focusing...
In this paper, we introduce the symmetric multiple Eisenstein series, a variant of the multiple Eisenstein series. As a fundamental result, we show that they satisfy the linear shuffle relation. As a case study, we investigate the vector space spanned by symmetric double Eisenstein series of weight $k$. When $k$ is even, it coincides...
We use the methods introduced by Lue Pan to study the locally analytic vectors of the completed cohomology of Shimura curves associated to an indefinite quaternion algebra $D$ which is ramified at a prime number $p$. Let $D_p^{\times}$ be the group of units of $D$ at $p$. Using $p$-adic uniformization of the quaternionic Shimura...
With the increasing number of high-precision astronomical observations, physical quantities that were previously inaccessible to accurate calculations, such as cosmic birefringence, have once again become a focal point of interest. Such phenomena induce a nonvanishing cross-correlation between the $E$- and $B$-mode polarizations of...
We demonstrate deterministic control of branched flow of light using anisotropic nematic liquid crystals. By sculpting the director field via photoalignment, we create spatially programmable optical potentials that govern light scattering and propagation. This platform enables configurable, anisotropic branched flow of light and...
Iterative code generation with Large Language Models (LLMs) can be viewed as an optimization process guided by textual feedback. However, existing LLM self-correction methods predominantly operate in a stateless, trial-and-error manner akin to first-order search, failing to leverage past problem-solving experiences. To bridge this...
Large vision-language models (LVLMs) are typically trained using autoregressive language modeling objectives, which align visual representations with linguistic space. While effective for multimodal reasoning, this alignment can weaken vision-centric capabilities, causing LVLMs to underperform their base vision encoders on tasks such...
Black holes contain more entropy than any other component of the observable universe. Gravitational-wave observations from LIGO and Virgo have shown evidence of a previously unknown black hole mass range, which provides new information to update the entropy budget. Increases in entropy due to binary black hole mergers, as implied in...
Twisted bilayer systems host a wealth of emergent phenomena, such as flat-band superconductivity, ferromagnetism, and ferroelectricity, arising from moiré superlattices and unconventional interlayer coupling. Despite their central role, direct and quantitative access to the out-of-plane atomic structure in these systems has remained...
Topological states of matter in disordered systems without translation symmetry have attracted great interest in recent years. These states with topological characters are not only robust against certain disorders, but also can be counterintuitively induced by disorders from a topologically trivial phase in the clean limit. In this...
Maritime communication is becoming a vital component of 6G networks, driven by the rapid expansion of the maritime economy. However, existing technologies face critical challenges in signal coverage, availability, and robustness, especially under harsh sea conditions. This paper proposes a novel framework for the maritime...
Quantum state distribution over vast distances is essential for global-scale quantum networks and fundamental test of quantum physics at space scale. While satellite platforms have demonstrated thousand-kilometer entanglement distribution, quantum key distribution and quantum teleportation with ground, future constellations and...
Using new Institut de Radioastronomie Millimétrique (IRAM) 30m telescope $\rm N_2H^+$, $\rm C^{18}O$ $J$=1-0 and Atacama Pathfinder Experiment (APEX) telescope $\rm ^{13}CO$ and $\rm C^{18}O$ $J$=2-1 maps together with archival far-infrared continuum data, and $\rm ^{12}CO$, and $\rm ^{13}CO$ $J$=1-0 data, we present a comprehensive...
Hybrid microgrids integrating Grid-Following (GFL) and Grid-Forming (GFM) inverters present complex control challenges arising from the decoupling between long-term economic dispatch and real-time dynamic regulation, as well as the distinct physical limitations of heterogeneous inverters under cyber uncertainties. This paper proposes...
Causal discovery is fundamental to scientific understanding and reliable decision-making. Existing approaches face critical limitations: purely data-driven methods suffer from statistical indistinguishability and modeling assumptions, while recent LLM-based methods either ignore statistical evidence or incorporate unverified priors...