arXiv subjects

Yanlin Liu

Publications and source records attributed to Yanlin Liu.

At least 19 recordsLinked to original sources

arXiv2026

TOGEARI: Interaction-Space Preconditioning for Condensed Finite-Element Systems with IPC Contact

Condensed finite-element contact systems combine a large material-volumetric core with a thin factorized contact update. We formulate Truncated Operator-Gram Eigenspace Approximation of Relevant Interactions (TOGEARI), a factor-space compression of that update. It selects a low-dimensional space of contact combinations and installs their contribution as a Woodbury right preconditioner. The assembled Newton equation and independent residual test remain fixed. The inverse action requires an invertible core and reduced Woodbury system. Core-response selection additionally assumes a symmetric core and remains available when that core is indefinite. For a positive definite core, the response values have an energy ordering. The same analysis then gives the exact generalized spectrum, a scaled inverse-error identity, and the optimal worst omitted interaction at each dimension. A raw contact-factor selector provides a lower-setup empirical alternative. We derive the split for a quadratic-tetrahedral displacement formulation with locally condensed elementwise-constant pressure and a factorized positive-semidefinite IPC normal tangent. In a frozen system with 325,260 free coordinates, an eight-dimensional subspace of a 94-row contact factor with registered numerical rank 42 reduces the Arnoldi basis from 11 vectors to 5. The core-response and raw spaces are closely aligned on the primary and near-repeat extractions; row-norm and deterministic-random controls require 12 vectors. The retained-dimension sweep reduces the warm median from 2.849 s to 1.629 s. Post-load setup plus first solve changes from 53.885 s to 61.958 s. These results identify a compact interaction correction in one frozen, frictionless regime. Distinct Newton states, evolving contact, contact-rank and mesh scaling, friction, and nested pressure-space performance remain open experimental questions.

math.NA↗

arXiv2026

Quantitative Analysis of Exciton Composition and Dynamics in Y6 Films for Single-Component Solar Cells

Non-fullerene acceptors such as Y6 have enabled high-efficiency organic photovoltaic devices and motivated the development of single-component architectures; however, the microscopic mechanisms governing exciton transport and charge dissociation remain under active investigation. In particular, the interplay between Frenkel-charge-transfer excitations and their coupling to environmental fluctuations complicates the description of light absorption and subsequent exciton dynamics. Here, ultrafast transient absorption spectroscopy is used to probe exciton quenching dynamics in Y6 films interfaced with hole-transport layers. To interpret these measurements, we develop an analytical model based on hybrid Frenkel-charge-transfer states that enables direct extraction of intermolecular electronic couplings, charge-transfer character, and system-bath interaction strengths from experimental data. The analysis reveals a substantial charge-transfer admixture of 20-40% in the exciton states and identifies a transport regime characterized by delocalization-mediated exciton motion rather than purely diffusive hopping. Consistent with this interpretation, the corresponding quenching dynamics occur on a ~1 ps timescale within ~4 nm of the interface, suggesting a short-range injection mechanism facilitated by exciton delocalization. In addition to providing physical parameters for Y6, these results establish a quantitative framework that connects spectroscopic observables to microscopic transport mechanisms and can be generalized to other non-fullerene acceptors.

physics.chem-ph↗

arXiv2026

MoWorld: A Flash World Model

The future of World Models depends not only on scaling model capability, but also on scaling practicality and inference efficiency. High-frame-rate inference enables responsive perception, planning, and control in real-world autonomous systems. To this end, we present MoWorld, a cost-effective yet high-performance Flash World Model with an end-to-end framework spanning data generation, pre-training, distillation, and efficient inference, enabling up to 50 FPS real-time interaction with cinematic visual quality without the need of high-end GPUs. To enable large-scale real-world deployment, MoWorld jointly optimizes model capability and cost throughout the entire development pipeline. Specifically, unlike existing approaches that primarily rely on large-scale video corpora, MoWorld is built upon a scalable 3D-native data engine accumulated from our large-scale 3D vision and generative modeling pipeline, enabling the efficient construction of geometrically consistent training data across diverse real-world and synthetic environments. Based on this foundation, a curriculum cross-frame pre-training strategy for stable and scalable World Model learning, an efficient denoising-step distillation algorithm to reduce diffusion training cost, and a mixed-precision parallel inference framework for low-cost real-time deployment. MoWorld is the first real-time interactive World Model built on the Neural Processing Unit (NPU) and can achieves up to 50 FPS in such the devices, enabling practical and efficient deployment at scale. Comprehensive evaluations demonstrate that MoWorld achieves leading performance; notably, its average inference cost is only 30\%-50\% of that of existing World Models, providing a practical foundation for large-scale real-world applications of World Models. We also demonstrate diverse applications of MoWorld.

cs.CV↗

arXiv2026

GIFT: LLM-Guided State-Reward Interface for Financial Reinforcement Learning

Financial portfolio trading is naturally formulated as a reinforcement learning problem, where an agent sequentially rebalances assets under changing market conditions to balance return, risk, and transaction costs. Yet in non-stationary markets, raw OHLCV states and short-horizon return rewards often provide an under-specified learning interface, motivating large language models as a way to inject financial knowledge into state and reward design while constraining open-ended generation. To this end, we propose GIFT, an LLM-guided framework for state-reward interface design in PPO-based financial reinforcement learning. Rather than using the LLM to make trading decisions, GIFT uses Factor-guided State Enhancement to generate state features from financial-factor primitives, Risk-rule-guided Reward Shaping to generate auxiliary rewards from portfolio-risk rules, and Diagnostic-guided Refinement to revise candidate interfaces using PPO rollout diagnostics. After refinement, GIFT fixes the selected state-reward interface before evaluation, with no further LLM queries or interface updates at test time. Comprehensive rolling-window experiments across diverse market regimes and portfolio scenarios demonstrate that GIFT improves learning-signal quality and out-of-sample risk-adjusted portfolio performance over baselines. Code and data are available at: https://github.com/KAG778/GIFT .

cs.AI↗

arXiv2026

Robust Self-Training with Closed-loop Label Correction for Learning from Noisy Labels

Training deep neural networks with noisy labels remains a significant challenge, often leading to degraded performance. Existing methods for handling label noise typically rely on either transition matrix, noise detection, or meta-learning techniques, but they often exhibit low utilization efficiency of noisy samples and incur high computational costs. In this paper, we propose a self-training label correction framework using decoupled bilevel optimization, where a classifier and neural correction function co-evolve. Leveraging a small clean dataset, our method employs noisy posterior simulation and intermediate features to transfer ground-truth knowledge, forming a closed-loop feedback system that prevents error amplification. Theoretical guarantees underpin the stability of our approach, and extensive experiments on benchmark datasets like CIFAR and Clothing1M confirm state-of-the-art performance with reduced training time, highlighting its practical applicability for learning from noisy labels.

cs.LG↗

arXiv2026

Vision Also You Need: Navigating Out-of-Distribution Detection with Multimodal Large Language Model

Out-of-Distribution (OOD) detection is a critical task that has garnered significant attention. The emergence of CLIP has spurred extensive research into zero-shot OOD detection, often employing a training-free approach. Current methods leverage expert knowledge from large language models (LLMs) to identify potential outliers. However, these approaches tend to over-rely on knowledge in the text space, neglecting the inherent challenges involved in detecting out-of-distribution samples in the image space. In this paper, we propose a novel pipeline, MM-OOD, which leverages the multimodal reasoning capabilities of MLLMs and their ability to conduct multi-round conversations for enhanced outlier detection. Our method is designed to improve performance in both near OOD and far OOD tasks. Specifically, (1) for near OOD tasks, we directly feed ID images and corresponding text prompts into MLLMs to identify potential outliers; and (2) for far OOD tasks, we introduce the sketch-generate-elaborate framework: first, we sketch outlier exposure using text prompts, then generate corresponding visual OOD samples, and finally elaborate by using multimodal prompts. Experiments demonstrate that our method achieves significant improvements on widely used multimodal datasets such as Food-101, while also validating its scalability on ImageNet-1K.

cs.CV↗

arXiv2025

Constructing control landscape of non-convex optimal control problem governed by nonlinear elliptic equation

Non-convex optimal control arises from various applications but may contain multiple stationary points. Classical solvers usually perform a local search and therefore rely on good initial guesses to reach appropriate local optimal controls. In this work we introduce a novel solution strategy for the non-convex optimal control of an elliptic equation, the main idea of which is to construct the control landscape based on the functional high-index saddle dynamics (FHiSD) method. This method reduces the dependence on prescribed initial guesses by using the information of high-index saddle points, and depicts the macroscopic configuration of the space of the control variable. Then various minima could be systematically computed along transition pathways and the control strategy can then be selected among them. We prove the effectiveness of the FHiSD in locating saddle points, and then justify its applicability in non-convex optimal control. Numerical results not only indicate the effectiveness of the proposed method, but reveal unintuitive phenomena (e.g. the non-monotonicity of the values of the cost functional with respect to the Morse indices) that support the necessity of computing multiple solutions of high indices.

math.OC↗

arXiv2025

TACO: Think-Answer Consistency for Optimized Long-Chain Reasoning and Efficient Data Learning via Reinforcement Learning in LVLMs

DeepSeek R1 has significantly advanced complex reasoning for large language models (LLMs). While recent methods have attempted to replicate R1's reasoning capabilities in multimodal settings, they face limitations, including inconsistencies between reasoning and final answers, model instability and crashes during long-chain exploration, and low data learning efficiency. To address these challenges, we propose TACO, a novel reinforcement learning algorithm for visual reasoning. Building on Generalized Reinforcement Policy Optimization (GRPO), TACO introduces Think-Answer Consistency, which tightly couples reasoning with answer consistency to ensure answers are grounded in thoughtful reasoning. We also introduce the Rollback Resample Strategy, which adaptively removes problematic samples and reintroduces them to the sampler, enabling stable long-chain exploration and future learning opportunities. Additionally, TACO employs an adaptive learning schedule that focuses on moderate difficulty samples to optimize data efficiency. Furthermore, we propose the Test-Time-Resolution-Scaling scheme to address performance degradation due to varying resolutions during reasoning while balancing computational overhead. Extensive experiments on in-distribution and out-of-distribution benchmarks for REC and VQA tasks show that fine-tuning LVLMs leads to significant performance improvements.

cs.CV↗

arXiv2024

The optimal time-decay estimates for 2-D inhomogeneous Navier-Stokes equations

In this paper, we derive the optimal time-decay estimates for 2-D inhomogeneous Navier-Stokes equations. In particular, we prove that $\|u(t)\|_{\dot{B}^{\theta}_{p,1}({\mathop{\mathbb R\kern 0pt}\nolimits}^2)}={\mathcal O} (t^{\frac1p-\frac32-\frac{\theta}2})$ as $t\rightarrow\infty$ for any $p\in[2,\infty[,~\theta\in [0,2]$ if initially $\rho_0u_0\in \dot{B}^{-2}_{2,\infty}({\mathop{\mathbb R\kern 0pt}\nolimits}^2)$. This is optimal even for the classical homogeneous Navier-Stokes equations. Different with Schonbek and Wiegner's Fourier splitting device, our method here seems more direct, and can adapt to many other equations as well. Moreover, our method allows us to work in the $L^p$-based spaces.

math.AP↗

arXiv2024

Global stability of large Fourier mode for 3-D anisotropic Navier-Stokes equations in cylindrical domain

In this paper, we first establish the global existence and stability of solutions to 3-D classical Navier-Stokes equations $(NS)$ in an infinite cylindrical domain with large Fourier mode initial data. Then we extend similar result for 3-D anisotropic Navier-Stokes equations $(ANS).$ We remark that due to the loss of vertical viscosity in $(ANS),$ the construction of the energy functionals for $(ANS)$ is much more subtle than that of $(NS).$ Compared with our previous paper for $(NS)$, we improve the polynomial decay in $k$ for the Fourier coefficients of the solution to be exponential decay in $k$ here.

math.AP↗

arXiv2023

On the one time-varying component regularity criteria for 3-D Navier-Stokes equations

In this paper, we consider the one time-varying component regularity criteria for local strong solution of 3-D Navier-Stokes equations. Precisely, if $\beta(t)$ is a piecewise $H^1$ unit vector from $[0,T] $ to $\Bbb{S}^2$ with finitely many jump discontinuities, we prove that if $\int_0^{T}\|u(t)\cdot \beta(t)\|_{\dot{H}^{\frac32}(\mathbb{R}^3)}^2\,dt<\infty,$ then the solution $u$ can be extended beyond the time $T.$ Compared with the previous results concerning one-component regularity criteria, here the unit vector $\beta(t)$ varies with time variable.

math.AP↗

arXiv2023

On the global stability of large Fourier mode for 3-D Navier-Stokes equations

In this paper, we first prove the global existence of strong solutions to 3-D incompressible Navier-Stokes equations with solenoidal initial data, which writes in the cylindrical coordinates is of the form: $A(r,z)\cos N\theta +B(r,z)\sin N\theta,$ provided that $N$ is large enough. In particular, we prove that the corresponding solution has almost the same frequency $N$ for any positive time. The main idea of the proof is first to write the solution in trigonometrical series in $\theta$ variable and estimate the coefficients separately in some scale-invariant spaces, then we handle a sort of weighted sum of these norms of the coefficients in order to close the a priori estimate of the solution. Furthermore, we shall extend the above well-posedness result for initial data which is a linear combination of axisymmetric data without swirl and infinitely many large mode trigonometric series in the angular variable.

math.AP↗

arXiv2023

On the existence and structures of almost axisymmetric solutions to 3-D Navier-Stokes equations

In this paper, we consider 3-D Navier-Stokes equations with almost axisymmetric initial data, which means that by writing $u_0 =u^r_0 e_r+u^\theta_0 e_\theta+u^z_0 e_z$ in the cylindrical coordinates, then $\partial_\theta u^r_0,\,\partial_\theta u^\theta_0$ and $\partial_\theta u^z_0$ are small in some sense (recall axisymmetric means these three quantities vanish). Then with additional smallness assumption on $u^\theta_0$, we prove the global existence of a unique strong solution $u$, and this solution keeps close to some axisymmetric vector field. We also establish some refined estimates for the integral average in $\theta$ variable for $u$. Moreover, as $u^r_0,\,u^\theta_0$ and $u^z_0$ here depend on $\theta$, it is natural to expand them into Fourier series in $\theta$ variable. And we shall consider one special form of $u_0$, with some small parameter $\varepsilon$ to measure its swirl part and oscillating part. We study the asymptotic expansion of the corresponding solution, and the influences between different profiles in the asymptotic expansion. In particular, we give some special symmetric structures that will persist for all time. These phenomena reflect some features of the nonlinear terms in Navier-Stokes equations.

math.AP↗

arXiv2023

Simple Combinatorial Construction of the $k^{o(1)}$-Lower Bound for Approximating the Parameterized $k$-Clique

In the parameterized $k$-clique problem, or $k$-Clique for short, we are given a graph $G$ and a parameter $k\ge 1$. The goal is to decide whether there exist $k$ vertices in $G$ that induce a complete subgraph (i.e., a $k$-clique). This problem plays a central role in the theory of parameterized intractability as one of the first W[1]-complete problems. Existing research has shown that even an FPT-approximation algorithm for $k$-Clique with arbitrary ratio does not exist, assuming the Gap-Exponential-Time Hypothesis (Gap-ETH) [Chalermsook et al., FOCS'17 and SICOMP]. However, whether this inapproximability result can be based on the standard assumption of $\mathrm{W} 1\ne \mathrm{FPT}$ remains unclear. The recent breakthrough of Bingkai Lin [STOC'21] and subsequent works by Karthik C.S. and Khot [CCC'22], and by Lin, Ren, Sun Wang [ICALP'22] give a technique that bypasses Gap-ETH, thus leading to the inapproximability ratio of $O(1)$ and $k^{o(1)}$ under $\mathrm{W}[1]$-hardness (the first two) and ETH (for the latter one). All the work along this line follows the framework developed by Lin, which starts from the $k$-vector-sum problem and requires some involved algebraic techniques. This paper presents an alternative framework for proving the W[1]-hardness of the $k^{o(1)}$-FPT-inapproximability of $k$-Clique. Using this framework, we obtain a gap-producing self-reduction of $k$-Clique without any intermediate algebraic problem. More precisely, we reduce from $(k,k-1)$-Gap Clique to $(q^k, q^{k-1})$-Gap Clique, for any function $q$ depending only on the parameter $k$, thus implying the $k^{o(1)}$-inapproximability result when $q$ is sufficiently large. Our proof is relatively simple and mostly combinatorial. At the core of our construction is a novel encoding of $k$-element subset stemming from the theory of "network coding" and a "Sidon set" representation of a graph.

cs.CC↗

arXiv2021

The influence of viscous coefficients on the lifespan of 3-D anisotropic Navier-Stokes system

The anisotropic Navier-Stokes system arises in geophysical fluid dynamics, which is derived by changing $-\nu\D$ in the classical Navier-Stokes system to $-(\nu_1\partial^2_1+\nu_2\partial^2_2+\nu_3\partial^2_3)$. Here $\nu_1,\,\nu_2\,,\nu_3$ are the viscous coefficients, which can be different from each other. This reflects that the fluid can behave differently in each direction. The purpose of this paper is to derive some lower bound estimates on the lifespan to such anisotropic Navier-Stokes system. We not only investigate the case when $\nu_1,\,\nu_2\,,\nu_3$ are all positive, but also the more sophisticated cases when one or two of them vanish. We find that in these lower bound estimates, the weights of $\nu_1,\,\nu_2\,,\nu_3$ are not equal. A detailed study of this problem can also help us to have a better understanding of the nonlinear structure in the classical Navier-Stokes system.

math.AP↗

arXiv2021

Latest Studies of the SM Higgs Boson Couplings to Fermions at ATLAS

This proceeding presents the latest studies on the Yukawa couplings of the Standard Model Higgs boson with 139 fb$^{-1}$ data collected using the ATLAS detector at a center-of-mass energy of 13 TeV. A first direct probe of $CP$ violation in the top-quark Yukawa coupling using events where the Higgs boson is produced in association with top quarks ($t\bar{t}H$ and $tH$), and decays into two photons ($H \rightarrow γγ$ ) is discussed. The latest results on the Higgs boson production in association with a $W$ or $Z$ boson ($VH$) in the $H \rightarrow b \bar{b}$ channel are depicted as well. Finally, the searches for Higgs boson decays into two muons ($H \rightarrow μ^{+} μ^{-}$) and two electrons ($H \rightarrow e^{+} e^{-}$) are presented.

hep-ex↗

arXiv2021

Long-time asymptotics of $3$-D axisymmetric Navier-Stokes equations in critical spaces

We show that any unique global solution (here we do not require any smallness condition beforehand) to 3-D axisymmetric Navier-Stokes equations in some scaling invariant spaces must eventually become a small solution. In particular, we show that the limits of $\|\omega^\theta(t)/r\|_{L^1}$ and $\|u^\theta(t)/\sqrt r\|_{L^2}$ are all $0$ as $t$ tends to infinity. And by using this, we can refine some decay estimates for the axisymmetric solutions.

math.AP↗

arXiv2020

Relative Equivariant Coarse Index Theorem and Relative $L^2$-Index Theorem

In this paper, we give a definition of the relative equivariant coarse index for proper actions and derive a relative equivariant coarse index theorem connecting this index with the localized equivariant coarse indices. This is an equivariant version of Roe's relative coarse index theorem in arXiv:arch-ive/1210.6100. Furthermore, we present a definition of the relative $L^2$-index and prove a relative $L^2$-index theorem which is a relative version of Atiyah's $L^2$-index theorem.

math.OA↗