arXiv subjects

Lei Yao

Publications and source records attributed to Lei Yao.

30 records · Page 2Linked to original sources

arXiv2025

The non-conservative compressible two-fluid system with common pressure: Global existence and sharp time asymptotics

This paper concerns the global-in-time evolution of a generic compressible two-fluid model in $\mathbb{R}^d$ ($d\geq3$) with the common pressure law. Due to the non-dissipative properties for densities and two different particle paths caused by velocities, the system lacks the usual symmetry structure and is partially dissipative in the sense that the Shizuta-Kawashima condition is violated, which makes it challenging to study its large-time stability. By developing a pure energy method in the framework of Besov spaces, we succeed in constructing a unique global classical solution to the Cauchy problem when the initial data are close to their constant equilibria. Compared to the previous related works, the main novelty lies in that our method is independent of the spectral analysis and does not rely on the $L^1$ smallness of the initial data. Furthermore, if additionally the initial perturbation is bounded in $\dot{B}^{σ_0}_{2,\infty}$ type spaces with lower regularity, the optimal time convergence rates are also obtained. In particular, the asymptotic convergence of the non-dissipative components toward their equilibrium states is first characterized.

math.AP↗

arXiv2024

A Survey of Embodied Learning for Object-Centric Robotic Manipulation

Embodied learning for object-centric robotic manipulation is a rapidly developing and challenging area in embodied AI. It is crucial for advancing next-generation intelligent robots and has garnered significant interest recently. Unlike data-driven machine learning methods, embodied learning focuses on robot learning through physical interaction with the environment and perceptual feedback, making it especially suitable for robotic manipulation. In this paper, we provide a comprehensive survey of the latest advancements in this field and categorize the existing work into three main branches: 1) Embodied perceptual learning, which aims to predict object pose and affordance through various data representations; 2) Embodied policy learning, which focuses on generating optimal robotic decisions using methods such as reinforcement learning and imitation learning; 3) Embodied task-oriented learning, designed to optimize the robot's performance based on the characteristics of different tasks in object grasping and manipulation. In addition, we offer an overview and discussion of public datasets, evaluation metrics, representative applications, current challenges, and potential future research directions. A project associated with this survey has been established at https://github.com/RayYoh/OCRM_survey.

cs.RO↗

arXiv2024

SGIFormer: Semantic-guided and Geometric-enhanced Interleaving Transformer for 3D Instance Segmentation

In recent years, transformer-based models have exhibited considerable potential in point cloud instance segmentation. Despite the promising performance achieved by existing methods, they encounter challenges such as instance query initialization problems and excessive reliance on stacked layers, rendering them incompatible with large-scale 3D scenes. This paper introduces a novel method, named SGIFormer, for 3D instance segmentation, which is composed of the Semantic-guided Mix Query (SMQ) initialization and the Geometric-enhanced Interleaving Transformer (GIT) decoder. Specifically, the principle of our SMQ initialization scheme is to leverage the predicted voxel-wise semantic information to implicitly generate the scene-aware query, yielding adequate scene prior and compensating for the learnable query set. Subsequently, we feed the formed overall query into our GIT decoder to alternately refine instance query and global scene features for further capturing fine-grained information and reducing complex design intricacies simultaneously. To emphasize geometric property, we consider bias estimation as an auxiliary task and progressively integrate shifted point coordinates embedding to reinforce instance localization. SGIFormer attains state-of-the-art performance on ScanNet V2, ScanNet200 datasets, and the challenging high-fidelity ScanNet++ benchmark, striking a balance between accuracy and efficiency. The code, weights, and demo videos are publicly available at https://rayyoh.github.io/sgiformer.

cs.CV↗

arXiv2024

A Concise but High-performing Network for Image Guided Depth Completion in Autonomous Driving

Depth completion is a crucial task in autonomous driving, aiming to convert a sparse depth map into a dense depth prediction. Due to its potentially rich semantic information, RGB image is commonly fused to enhance the completion effect. Image-guided depth completion involves three key challenges: 1) how to effectively fuse the two modalities; 2) how to better recover depth information; and 3) how to achieve real-time prediction for practical autonomous driving. To solve the above problems, we propose a concise but effective network, named CENet, to achieve high-performance depth completion with a simple and elegant structure. Firstly, we use a fast guidance module to fuse the two sensor features, utilizing abundant auxiliary features extracted from the color space. Unlike other commonly used complicated guidance modules, our approach is intuitive and low-cost. In addition, we find and analyze the optimization inconsistency problem for observed and unobserved positions, and a decoupled depth prediction head is proposed to alleviate the issue. The proposed decoupled head can better output the depth of valid and invalid positions with very few extra inference time. Based on the simple structure of dual-encoder and single-decoder, our CENet can achieve superior balance between accuracy and efficiency. In the KITTI depth completion benchmark, our CENet attains competitive performance and inference speed compared with the state-of-the-art methods. To validate the generalization of our method, we also evaluate on indoor NYUv2 dataset, and our CENet still achieve impressive results. The code of this work will be available at https://github.com/lmomoy/CHNet.

cs.CV↗

arXiv2023

SAI: Solving AI Tasks with Systematic Artificial Intelligence in Communication Network

In the rapid development of artificial intelligence, solving complex AI tasks is a crucial technology in intelligent mobile networks. Despite the good performance of specialized AI models in intelligent mobile networks, they are unable to handle complicated AI tasks. To address this challenge, we propose Systematic Artificial Intelligence (SAI), which is a framework designed to solve AI tasks by leveraging Large Language Models (LLMs) and JSON-format intent-based input to connect self-designed model library and database. Specifically, we first design a multi-input component, which simultaneously integrates Large Language Models (LLMs) and JSON-format intent-based inputs to fulfill the diverse intent requirements of different users. In addition, we introduce a model library module based on model cards which employ model cards to pairwise match between different modules for model composition. Model cards contain the corresponding model's name and the required performance metrics. Then when receiving user network requirements, we execute each subtask for multiple selected model combinations and provide output based on the execution results and LLM feedback. By leveraging the language capabilities of LLMs and the abundant AI models in the model library, SAI can complete numerous complex AI tasks in the communication network, achieving impressive results in network optimization, resource allocation, and other challenging tasks.

cs.AI↗

arXiv2022

Global Stability and Non-Vanishing Vacuum States of 3D Compressible Navier-Stokes Equations

We investigate the global stability and non-vanishing vacuum states of large solutions to the compressible Navier-Stokes equations on the torus $\mathbb{T}^3$, and the main novelty of this work is three-fold: First, under the assumption that the density $ρ({\bf{x}}, t)$ verifies $\sup_{t\geq 0}\|ρ(t)\|_{L^\infty}\leq M$, it is shown that the solutions converge to equilibrium state exponentially in $L^2$-norm. Second, by employing some new thoughts, we also show that the density converges to its equilibrium state exponentially in $L^\infty$-norm if additionally the initial density $ρ_0({\bf{x}})$ satisfies $\inf_{\bf{x}\in\mathbb{T}^3}ρ_0({\bf{x}})\geq c_0>0$. Finally, we prove that the vacuum states will not vanish for any time provided that the vacuum states are present initially. This phenomenon is totally new and somewhat surprising, and particularly is in contrast to the previous work of [H. L. Li et al., Commun. Math. Phys., 281 (2008), 401-444], where the authors showed that the vacuum states must vanish within finite time for the 1D compressible Navier-Stokes equations with density-dependent viscosity $μ(ρ)=ρ^α$ with $α>1/2$.

math.AP↗

arXiv2022

On instability of a generic compressible two-fluid model in $\mathbb R^3$

We are concerned with the instability of a generic compressible two-fluid model in the whole space $\mathbb{R}^3$, where the capillary pressure $f(α^-ρ^-)=P^+-P^-\neq 0$ is taken into account. For the case that the capillary pressure is a strictly decreasing function near the equilibrium, namely, $f'(1)<0$, Evje-Wang-Wen established global stability of the constant equilibrium state for the three-dimensional Cauchy problem under some smallness assumptions. Recently, Wu-Yao-Zhang proved global stability of the constant equilibrium state for the case $P^+=P^-$ (corresponding to $f'(1)=0$). In this work, we investigate the instability of the constant equilibrium state for the case that the capillary pressure is a strictly increasing function near the equilibrium, namely, $f'(1)>0$. First, by employing Hodge decomposition technique and making detailed analysis of the Green's function for the corresponding linearized system, we construct solutions of the linearized problem that grow exponentially in time in the Sobolev space $H^k$, thus leading to a global instability result for the linearized problem. Moreover, with the help of the global linear instability result and a local existence theorem of classical solutions to the original nonlinear system, we can then show the instability of the nonlinear problem in the sense of Hadamard by making a delicate analysis on the properties of the semigroup. Therefore, our result shows that for the case $f'(1)>0$, the constant equilibrium state of the two-fluid model is linearly globally unstable and nonlinearly locally unstable in the sense of Hadamard, which is in contrast to the cases $f'(1)<0$ and $P^+=P^-$ (corresponding to $f'(1)=0$) where the constant equilibrium state of the two--fluid model is nonlinearly globally stable.

math.AP↗

arXiv2022

Global well--posedness and large time behavior of classical solutions to a generic compressible two-fluid model

In this paper, we investigate a generic compressible two-fluid model with common pressure ($P^+=P^-$) in $\mathbb{R}^3$. Under some smallness assumptions, Evje-Wang-Wen [Arch Rational Mech Anal 221:1285--1316, 2016] obtained the global solution and its optimal decay rate for the 3D compressible two-fluid model with unequal pressures $P^+\neq P^-$. More precisely, the capillary pressure $f(α^-ρ^-)=P^+-P^-\neq 0$ is taken into account, and is assumed to be a strictly decreasing function near the equilibrium. As indicated by Evje-Wang-Wen, this assumption played an key role in their analysis and appeared to have an essential stabilization effect on the model. However, global well-posedness of the 3D compressible two-fluid model with common pressure has been a challenging open problem due to the fact that the system is partially dissipative and its nonlinear structure is very terrible. In the present work, by exploiting the dissipation structure of the model and making full use of several key observations, we establish global existence and large time behavior of classical solutions to the 3D compressible two-fluid model with common pressure. One of key observations here is that to closure the higher-order energy estimates of non-dissipative variables (i.e, fraction densities $α_{\pm}ρ_\pm$), we will introduce the linear combination of two velocities ($u^\pm$): $v=(2μ^++λ^+)u^+-(2μ^-+λ^-)u^-$ and explore its good regularity, which is particularly better than ones of two velocities themselves.

math.AP↗

arXiv2019

GRID: a Student Project to Monitor the Transient Gamma-Ray Sky in the Multi-Messenger Astronomy Era

The Gamma-Ray Integrated Detectors (GRID) is a space mission concept dedicated to monitoring the transient gamma-ray sky in the energy range from 10 keV to 2 MeV using scintillation detectors onboard CubeSats in low Earth orbits. The primary targets of GRID are the gamma-ray bursts (GRBs) in the local universe. The scientific goal of GRID is, in synergy with ground-based gravitational wave (GW) detectors such as LIGO and VIRGO, to accumulate a sample of GRBs associated with the merger of two compact stars and study jets and related physics of those objects. It also involves observing and studying other gamma-ray transients such as long GRBs, soft gamma-ray repeaters, terrestrial gamma-ray flashes, and solar flares. With multiple CubeSats in various orbits, GRID is unaffected by the Earth occultation and serves as a full-time and all-sky monitor. Assuming a horizon of 200 Mpc for ground-based GW detectors, we expect to see a few associated GW-GRB events per year. With about 10 CubeSats in operation, GRID is capable of localizing a faint GRB like 170817A with a 90% error radius of about 10 degrees, through triangulation and flux modulation. GRID is proposed and developed by students, with considerable contribution from undergraduate students, and will remain operated as a student project in the future. The current GRID collaboration involves more than 20 institutes and keeps growing. On August 29th, the first GRID detector onboard a CubeSat was launched into a Sun-synchronous orbit and is currently under test.

astro-ph.IM↗

arXiv2017

Existence of global weak solutions for the Navier-Stokes-Vlasov-Boltzmann equations

A moderately thick spray can be described by a coupled system of equations consisting of the incompressible Navier-Stokes equations and the Vlasov-Boltzmann equation. We investigate this kind of mathematical model in this paper. In particular, we study the initial value problem for the Navier-Stokes-Vlasov-Boltzmann equations. The existence of global weak solutions is established by a weak convergence method. The interesting point of our main result is to handle the model with some breakup effects while the velocity of particles is in the whole space.

math.AP↗

arXiv2012

Global spherically symmetric classical solution to compressible Navier-Stokes equations with large initial data and vacuum

In this paper, we obtain a result on the existence and uniqueness of global spherically symmetric classical solutions to the compressible isentropic Navier-Stokes equations with vacuum in a bounded domain or exterior domain Ω of Rn(n >= 2). Here, the initial data could be large. Besides, the regularities of the solutions are better than those obtained in [H.J. Choe and H. Kim, Math. Methods Appl. Sci., 28 (2005), pp. 1-28; Y. Cho and H. Kim, Manuscripta Math., 120 (2006), pp. 91-129; S.J. Ding, H.Y.Wen, and C.J. Zhu, J. Differential Equations, 251 (2011), pp. 1696-1725]. The analysis is based on some new mathematical techniques and some new useful energy estimates. This is an extension of the work of Choe and Kim, Cho and Kim, and Ding, Wen, and Zhu, where the global radially symmetric strong solutions, the local classical solutions in three dimensions, and the global classical solutions in one dimension were obtained, respectively. This paper can be viewed as the first result on the existence of global classical solutions with large initial data and vacuum in higher dimension

math.AP↗

arXiv2003

Evolution of population with sexual and asexual reproduction in changing environment

Using a lattice model based on Monte Carlo simulations, we study the role of the reproduction pattern on the fate of an evolving population. Each individual is under the selection pressure from the environment and random mutations. The habitat ("climate") is changing periodically. Evolutions of populations following two reproduction patterns are compared, asexual and sexual. We show, via Monte Carlo simulations, that sexual reproduction by keeping more diversified populations gives them better chances to adapt themselves to the changing environment. However, in order to obtain a greater chance to mate, the birth rate should be high. In the case of low birth rate and high mutation probability there is a preference for the asexual reproduction.

cond-mat.stat-mech↗