arXiv subjects

Yin Li

Publications and source records attributed to Yin Li.

At least 37 records · Page 2Linked to original sources

arXiv2026

Aspherical Lagrangian submanifolds, Audin's conjecture and cyclic dilations

Given a closed, oriented Lagrangian submanifold $L$ in a Liouville domain $\overline{M}$, one can define a Maurer-Cartan element with respect to a certain $L_\infty$-structure on the string homology $\widehat{H}_\ast^{S^1}(\mathcal{L}L;\mathbb{R})$, completed with respect to the action filtration. When the first Gutt-Hutchings capacity of $\overline{M}$ is finite, and $L$ is a $K(π,1)$ space, we show that $L$ bounds a pseudoholomorphic disc of Maslov index 2. This confirms a general form of Audin's conjecture and generalizes the works of Fukaya and Irie in the case of $\mathbb{C}^n$ to a wide class of Liouville manifolds, which includes low degree smooth affine hypersurfaces in $\mathbb{C}^{n+1}$. In particular, when $\dim_\mathbb{R}(\overline{M})=6$, every closed, orientable, prime Lagrangian 3-manifold $L\subset\overline{M}$ is diffeomorphic either to a spherical space form, or $S^1\timesΣ_g$, where $Σ_g$ is a closed oriented surface.

math.SG↗

arXiv2026

UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillation

Autoregressive video diffusion models have emerged as a promising approach for long video generation, achieving strong performance in streaming settings. However, existing methods are restricted to forward temporal generation, whereas practical video creation often requires flexible generation order, e.g., conditioning on future context to extend backward, or on both past and future context for inbetween generation. We bridge this gap by training an autoregressive model that supports generation in arbitrary temporal directions. A key technical challenge arises from the Causal 3D VAE widely used in video diffusion models, which encodes latents strictly conditioned on past context. While suited for forward generation, this causal structure causes inter-block discontinuities when generation proceeds backward. To address this, we introduce blockwise anchor latents, a set of auxiliary latents that restore the missing past context at block boundaries during backward generation. Built on this design, we propose UniTemp, a bidirectional distillation framework that trains a single autoregressive student model for any-direction video generation. At inference time, UniTemp conditions on arbitrary past and/or future frames, improving controllability for both bidirectional and inbetween generation. Experiments show that UniTemp maintains competitive performance on short and long video generation compared to forward-only methods, while enabling diverse workflows such as bidirectional video extension, inbetween generation, looping video generation, scene transition, and visual story generation. Project website: https://lzhangbj.github.io/projects/unitemp/

cs.CV↗

arXiv2026

Polarity Reversal of the Polar Magnetic Fields in Solar Cycle 25

The polar magnetic field polarity reversal is a key signature of solar cycle evolution, and precise determination of its timing is crucial for dynamo theory validation and solar cycle prediction. We investigate the polar polarity reversal of solar cycle 25 using the vector magnetic field data from the spectropolarimeter on board the Hinode satellite. We constructed polar top-down composite maps from Hinode-view magnetograms. These maps show the year-to-year polar polarity variations, with the northern polar region gradually changing from positive to negative and the southern polar region exhibiting the reverse behavior. The polarity reversals of the northern and southern polar caps (above 70 deg latitude) likely occurred in November 2024 and October 2024, respectively. The northern polarity reversal lagged the northern hemispheric sunspot number maximum by approximately 19 months, while the southern reversal possibly coincided with the southern maximum. Moreover, polarity reversal times calculated at 5 deg latitude intervals above 70 deg reveal a trend of earlier reversal in lower latitudes consistent with that of solar cycle 24. These results offer observational references for modeling polar polarity reversal in solar cycles.

astro-ph.SR↗

arXiv2026

M$^3$Eval: Multi-Modal Memory Evaluation through Cognitively-Grounded Video Tasks

As multi-modal models advance towards long-form video understanding, memory emerges as a critical capability. Despite substantial efforts in developing video datasets and benchmarks, existing works primarily focus on perception and reasoning, without systematically evaluating memory: what models retain, how faithfully information is preserved, and how robust memory remains under interference. To address this gap, we introduce M$^3$Eval, the first comprehensive evaluation framework and benchmark for probing different memory dimensions in multi-modal models. Grounded in cognitive psychology, our design features carefully constructed tasks that isolate key aspects of memory. Leveraging M$^3$Eval, we conduct extensive experiments across representative multi-modal models, revealing consistent weaknesses and distinctive behaviors. We find that models struggle to maintain disentangled representations when processing parallel video streams, exhibit interference patterns differing substantially from those observed in human memory, ground memory sources more reliably in the spatial domain than the temporal domain, and demonstrate limited symbolic memory. Collectively, our benchmark provides a valuable resource for future research, while our findings highlight memory as a fundamental yet underexplored capability and offer insights for designing more effective memory mechanisms in multi-modal models. Our code and dataset are available at https://pku-value-lab.github.io/m3eval-homepage.

cs.CV↗

arXiv2026

Plumbings of lens spaces and crepant resolutions of compound $A_n$ singularities

For many compound $A_n$ ($cA_n$) singularities $R_f=\mathbb{C}[u,v,x,y]/(uv-f(x,y))$ with crepant resolutions $Y_f$, their mirrors are affine $A_n$ plumbings $W^\circ_f$ of $3$-dimensional lens spaces along circles. We prove two versions of homological mirror symmetry for these Stein $3$-folds. (i) The uncompleted version: there is an equivalence $D^\mathit{perf}\mathcal{W}(W_f^\circ)\simeq D^b\mathit{Coh}(Y^\circ_f)$ between the derived wrapped Fukaya category and the bounded derived category of coherent sheaves on some divisor complement $Y^\circ_f=Y_f\setminus D$. (ii) The completed version: there is an equivalence $D^\mathit{perf}\widehat{\mathcal{W}}(W^\circ_f)\simeq D^b\mathit{Coh}(\widehat{Y}_f)$, where $\widehat{\mathcal{W}}(W^\circ_f)$ is the completion of $\mathcal{W}(W^\circ_f)$ with respect to the word-length filtration of Hamiltonian chords, and $\widehat{Y}_f$ is the complete local version of $Y_f$. As an application of (i), we show that certain infinitely generated subgroup of the pure braid group $\mathit{PBr}_{n+2}$ split injects into the compactly supported symplectic mapping class group of $W^\circ_f$ as long as $R_f$ is isolated, generalizing the work of Keating-Smith in the case of a conifold smoothing. Applying categorical localization to (ii), we obtain an equivalence between the (uncompleted) derived wrapped Fukaya category of the corresponding (non-affine) $A_n$ plumbing $W_f$ of lens spaces along circles and the relative singularity category of $Y_f$. This generalizes the result of Smith-Wemyss in the case of double bubble plumbings and partially answers their realization question.

math.SG↗

arXiv2026

Sustained 70B-Class AWQ Inference on a Single NVIDIA L20: Throughput, Stability, Energy, and Quality Characterization

Serving 70B-class open-weight language models is usually associated with 80GB accelerators, tensor-parallel multi-GPU systems, or vendor-managed inference profiles. This technical report evaluates whether a single NVIDIA L20 48GB GPU can sustain a useful 70B-class quantized serving workload. We measure Qwen2.5-72B-Instruct-AWQ served with vLLM 0.8.5.post1 and AWQ Marlin on one L20. Under a fixed workload of approximately 512 input tokens and 256 output tokens, a 24-hour concurrency-10 soak completed 36,740/36,740 requests with no request failures and no vLLM CUDA OOM, traceback, or killed-process signatures. The system sustained 108.84 output tokens/s, with p95 time-to-first-token of 6.61s and p95 end-to-end latency of 23.54s. GPU-board power sampled through nvidia-smi produced an estimated 7.92 kWh over the run, corresponding to 0.330 output tokens/J and 1.008 total tokens/J. Repeated fixed-shape runs at concurrency 1, 4, 8, and 16 completed 12/12 runs successfully; the concurrency-16 condition averaged 127.22 +/- 12.68 output tokens/s over three runs. The same AWQ endpoint also produced absolute quality scores of 0.8130 on MMLU, 0.8309 on CMMLU, and 0.8082 on GSM8K, plus 80/80 MT-Bench answer generations and a 60-item 8K LongBench subset. The evidence supports a narrow claim: a carefully configured single L20 can serve Qwen2.5-72B-Instruct-AWQ as a throughput-oriented 70B-class endpoint under the tested fixed-shape workload. It does not prove lossless AWQ quality retention, low-latency interactive serving, broad production SLA coverage, or equivalence to a BF16/FP16 baseline.

cs.DC↗

arXiv2026

EgoIntrospect: An Egocentric Dataset and Benchmark for User-Centric Internal State Reasoning

Despite extensive efforts on egocentric video datasets and benchmarks, understanding users' internal states, which is crucial for enabling seamless AI assistant experiences, remains largely overlooked. In this work, we introduce EgoIntrospect, the first egocentric dataset captured in user-driven scenarios with self-annotations that explicitly reveal users' interactive intentions with AI assistants. EgoIntrospect was collected using a cross-device setup, providing synchronized video, audio, gaze, motion, and physiological signals. It consists of 180 hours of recordings from 60 subjects, with an average recording duration of 3 hours per subject. Leveraging EgoIntrospect, we formalize a suite of tasks centered on user internal states, including affective experience, interactive intent, and cognitive memory. We further process the annotations to construct benchmarks that evaluate the ability of modern multimodal large language models to reason about users' internal states from egocentric observations. Experiments on our benchmark suggest that existing multimodal large language models struggle to effectively leverage multimodal signals to infer users' subjective internal states. The dataset and annotations will be made publicly available to advance research in egocentric vision and wearable AI assistants. Project page: https://ego-introspect.github.io/

cs.CV↗

arXiv2026

When JSON Is Not Enough: Semantic Reliability of Schema-Constrained LLM Ordering Agents

LLM agents are increasingly used as transaction compilers: a user states an intent in natural language, and the model emits a structured object that an API can execute. JSON Schema and provider-level structured-output modes are useful because they remove a large class of parse failures, but they do not by themselves decide whether the object is a safe, faithful transaction. We introduce OrderBench, a deterministic benchmark for restaurant ordering agents that separates syntactic validity, schema validity, status decisions, exact item semantics, constraint preservation, and unsafe acceptances. Across 2,400 Nebius Token Factory calls to four open models in prompt-only and JSON-schema modes, we find that schema-valid output can still have large semantic error rates. In the strongest model, both modes achieve 100% schema validity, yet semantic success remains near 80%; in weaker models, schema-valid unsafe acceptances occur in double digits. The result is a concrete engineering warning: structured output is a necessary interface layer, not a substitute for domain verification and fail-closed execution.

cs.AI↗

arXiv2026

MUltiplexed Survey Telescope (MUST) Science White Paper I: Overview of Large-Scale Structure Cosmology in the Era of Stage-V Spectroscopic Surveys

The MUltiplexed Survey Telescope (MUST) is a 6.5-meter telescope under development. Dedicated to highly-multiplexed, wide-field spectroscopic surveys, MUST observes over 20,000 targets simultaneously using 6.2-mm pitch positioning robots within a ~5 deg$^2$ field of view. MUST aims to conduct the first Stage-V spectroscopic survey in the 2030s, mapping the 3D Universe with over 100 million galaxies and quasars, spanning from the nearby Universe to a redshift of z ~ 5.5, corresponding to approximately 1 billion years after the Big Bang. To cover this extensive redshift range, we present an initial conceptual target selection algorithm for different types of galaxies, ranging from local bright galaxies and luminous red galaxies to emission-line galaxies, and high-redshift (2 < z < 5.5) Lyman-break galaxies. Using Fisher forecasts, we demonstrate that MUST can address fundamental questions in cosmology, including the nature of dark energy, tests of gravity theories, and investigations into primordial physics. This is the first paper in the series of science white papers for MUST, with subsequent developments focusing on additional scientific cases such as galaxy and quasar evolution, Milky Way physics, and dynamic phenomena in the time-domain Universe.

astro-ph.CO↗

arXiv2026

DUALVISION: RGB-Infrared Multimodal Large Language Models for Robust Visual Reasoning

Multimodal large language models (MLLMs) have achieved impressive performance on visual perception and reasoning tasks with RGB imagery, yet they remain fragile under common degradations, such as fog, blur, or low-light conditions. Infrared (IR) imaging, a well-established complement to RGB, offers inherent robustness in these conditions, but its integration into MLLMs remains underexplored. To bridge this gap, we propose DUALVISION, a lightweight fusion module that efficiently incorporates IR-RGB information into MLLMs via patch-level localized cross-attention. To support training and evaluation and to facilitate future research, we also introduce DV-204K, a dataset of ~25K publicly available aligned IR-RGB image pairs with 204K modality-specific QA annotations, and DV-500, a benchmark of 500 IR-RGB image pairs with 500 QA pairs designed for evaluating cross-modal reasoning. Leveraging these datasets, we benchmark both open- and closed-source MLLMs and demonstrate that DUALVISION delivers strong empirical performance under a wide range of visual degradations. Our code and dataset are available at https://abrarmajeedi.github.io/dualvision.

cs.CV↗

arXiv2026

Into the Gompverse: A robust Gompertzian reionization model for CMB analyses

Cosmic reionization is driven by the formation of sources of ultraviolet photons, and hence it is an intrinsically asymmetric process, where its earlier stages occur at a slower pace relative to its later stages. Yet most modern cosmic microwave background (CMB) analyses rely on a hyperbolic tangent template, i.e. a symmetric sigmoid, that is not well suited for joint fitting of CMB and reionization observations. In this work, we introduce a physically motivated Gompertzian reionization model with three astrophysical (nuisance) parameters, designed to enable joint analyses of CMB and reionization data and to be applicable to a wide range of datasets and cosmological models. This robust Gompertzian model leverages the connection between cosmology and reionization, typically ignored in standard CMB analyses, to demote the optical depth ($τ_{\rm reio}$) to derived parameter, reducing its uncertainty by approximately a factor of three compared to the conventional $\tanh$ prescription. The $τ_{\rm reio}$ improvement enables tighter constraints on the sum of the neutrino masses, revealing potential tension with neutrino oscillation experiments even after accounting for the known relaxation of neutrino mass bounds in $w_0w_a$CDM models -- a tension that is partially obscured by the conventional treatment of reionization. In addition, the inferred constraints on the astrophysical parameters governing reionization naturally synergize with current and upcoming 21 cm experiments, providing physically informed parameter ranges for future 21 cm studies.

astro-ph.CO↗

arXiv2026

Digital Guardians: The Past and The Future of Cyber-Physical Resilience

Resilience in cyber-physical systems (CPS) is the fundamental ability to maintain safety and critical functionality despite adverse "perturbations," which includes security attacks, environmental disruptions, and hardware or software failures. This survey provides a comprehensive review of CPS resilience, framing the field through five interconnected themes that are required in an integrated whole to achieve real-world resilience. The article first posits that resilience is a system-wide property emerging from interactions between hardware, software, and human users. Second, it addresses the challenges of learning-enabled CPS, which often operate in data-scarce environments characterized by imbalanced or noisy data, requiring innovative solutions like synthetic data generation and foundation model adaptation. Third, the survey examines proactive measures for resilience, which include distinctive aspects of verification, testing, and redundancy. Fourth, it explores recovery mechanisms, moving beyond traditional fault models to design "just good enough" recovery strategies that prioritize safety-critical functions during perturbations. Finally, it highlights the central role of the human, focusing on the different levels of human intervention, the necessity of trust calibration, and the requirement for explainable AI to support human-CPS teaming. These themes are illustrated through representative application domains, primarily Connected and Autonomous Transportation Systems (CATS) and Medical CPS (MCPS). By integrating the five interconnected themes, this survey provides a systematic roadmap for achieving the resilient CPS in increasingly complex and adversarial environments.

cs.CR↗

arXiv2026

A scalable and real-time neural decoder for topological quantum codes

Fault-tolerant quantum computing will require error rates far below those achievable with physical qubits. Quantum error correction (QEC) bridges this gap, but depends on decoders being simultaneously fast, accurate, and scalable. This combination of requirements remains unmet by a machine-learning decoder, nor by any decoder for promising resource-efficient codes such as the color code. Here we introduce AlphaQubit 2, a neural-network decoder that achieves near-optimal logical error rates for both surface and color codes at scale under realistic noise. For the color code, it is orders of magnitude faster than other high-accuracy decoders. We demonstrate real-time decoding faster than 1μs per cycle on commercial accelerators: for the surface code to distance 11, with better accuracy than leading real-time decoders; and the first real-time decoding of the color code to distance 9. These results support the practical application of a wider class of promising QEC codes, and establish a credible path towards high-accuracy, real-time neural decoding at the scales required for fault-tolerant quantum computation.

quant-ph↗

arXiv2026

Slow neutrinos: non-linearity and momentum-space emulation

Recent cosmological bounds on the sum of neutrino masses, M_nu = sum m_nu, are in tension with laboratory oscillation experiments, making cosmological tests of neutrino free-streaming imperative. In order to study the scale-dependent clustering of massive neutrinos, we develop a fast linear response method, FAST-nu f, applicable to neutrinos and other non-relativistic hot dark matter. Using it as an accurate linear approximation to help us reduce the dynamic range of emulator training data, based upon a non-linear perturbation theory for massive neutrinos, we improve the emulator's accuracy at small M_nu and length scales by a factor of two. We significantly sharpen its momentum resolution for the slowest neutrinos, which, despite their small mass fraction, dominate small-scale clustering. Furthermore, we extend the emulator from the degenerate to the normal and inverted mass orderings. Applying this new emulator, Cosmic-Enu-II, to large halos in N-body simulations, we show that non-linear perturbation theory can reproduce the neutrino density profile in the halo outskirts, 2R_vir < r < 10R_vir , to better than 10%.

astro-ph.CO↗

arXiv2026

ReGLA: Efficient Receptive-Field Modeling with Gated Linear Attention Network

Balancing accuracy and latency on high-resolution images is a critical challenge for lightweight models, particularly for Transformer-based architectures that often suffer from excessive latency. To address this issue, we introduce \textbf{ReGLA}, a series of lightweight hybrid networks, which integrates efficient convolutions for local feature extraction with ReLU-based gated linear attention for global modeling. The design incorporates three key innovations: the Efficient Large Receptive Field (ELRF) module for enhancing convolutional efficiency while preserving a large receptive field; the ReLU Gated Modulated Attention (RGMA) module for maintaining linear complexity while enhancing local feature representation; and a multi-teacher distillation strategy to boost performance on downstream tasks. Extensive experiments validate the superiority of ReGLA; particularly the ReGLA-M achieves \textbf{80.85\%} Top-1 accuracy on ImageNet-1K at $224px$, with only \textbf{4.98 ms} latency at $512px$. Furthermore, ReGLA outperforms similarly scaled iFormer models in downstream tasks, achieving gains of \textbf{3.1\%} AP on COCO object detection and \textbf{3.6\%} mIoU on ADE20K semantic segmentation, establishing it as a state-of-the-art solution for high-resolution visual applications.

cs.CV↗

arXiv2026

One latent to fit them all: a unified representation of baryonic feedback on matter distribution

Accurate and parsimonious quantification of baryonic feedback on matter distribution is of crucial importance for understanding both cosmology and galaxy formation from observational data. This is, however, challenging given the large discrepancy among different models of galaxy formation simulations, and their distinct subgrid physics parameterizations. Using 5,072 simulations from 4 different models covering broad ranges in their parameter spaces, we find a unified 2D latent representation. Compared to the simulations and other phenomenological models, our representation is independent of both time and cosmology, much lower-dimensional, and disentangled in its impacts on the matter power spectra. The common latent space facilitates the comparison of parameter spaces of different models and is readily interpretable by correlation with each. The two latent dimensions provide a complementary representation of baryonic effects, linking black hole and supernova feedback to distinct and interpretable impacts on both the matter power spectrum, and field, level. Our approach enables developing robust and economical analytic models for optimal gain of physical information from data, and is generalizable to other fields with significant modeling uncertainty.

astro-ph.CO↗

arXiv2025

Decomposing LLM Self-Correction: The Accuracy-Correction Paradox and Error Depth Hypothesis

Large Language Models (LLMs) are widely believed to possess self-correction capabilities, yet recent studies suggest that intrinsic self-correction--where models correct their own outputs without external feedback--remains largely ineffective. In this work, we systematically decompose self-correction into three distinct sub-capabilities: error detection, error localization, and error correction. Through cross-model experiments on GSM8K-Complex (n=500 per model, 346 total errors) with three major LLMs, we uncover a striking Accuracy-Correction Paradox: weaker models (GPT-3.5, 66% accuracy) achieve 1.6x higher intrinsic correction rates than stronger models (DeepSeek, 94% accuracy)--26.8% vs 16.7%. We propose the Error Depth Hypothesis: stronger models make fewer but deeper errors that resist self-correction. Error detection rates vary dramatically across architectures (10% to 82%), yet detection capability does not predict correction success--Claude detects only 10% of errors but corrects 29% intrinsically. Surprisingly, providing error location hints hurts all models. Our findings challenge linear assumptions about model capability and self-improvement, with important implications for the design of self-refinement pipelines.

cs.AI↗

arXiv2025

Rethinking Chain-of-Thought Reasoning for Videos

Chain-of-thought (CoT) reasoning has been highly successful in solving complex tasks in natural language processing, and recent multimodal large language models (MLLMs) have extended this paradigm to video reasoning. However, these models typically build on lengthy reasoning chains and large numbers of input visual tokens. Motivated by empirical observations from our benchmark study, we hypothesize that concise reasoning combined with a reduced set of visual tokens can be sufficient for effective video reasoning. To evaluate this hypothesis, we design and validate an efficient post-training and inference framework that enhances a video MLLM's reasoning capability. Our framework enables models to operate on compressed visual tokens and generate brief reasoning traces prior to answering. The resulting models achieve substantially improved inference efficiency, deliver competitive performance across diverse benchmarks, and avoid reliance on manual CoT annotations or supervised fine-tuning. Collectively, our results suggest that long, human-like CoT reasoning may not be necessary for general video reasoning, and that concise reasoning can be both effective and efficient. Our code will be released at https://github.com/LaVi-Lab/Rethink_CoT_Video.

cs.CV↗