arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.01882cs.SInlin.AOphysics.soc-ph
收藏

固定规模动态无标度网络:建模、平稳性与鲁棒性

Fixed-Size Dynamic Scale-Free Networks: Modeling, Stationarity, and Resilience

College of Artificial Intelligence, Southwest University(西南大学人工智能学院);Faculty of Natural Sciences and Mathematics, University of Maribor(马里博尔大学自然科学与数学学院);Community Healthcare Center Dr. Adolf Drolc Maribor(马里博尔阿道夫·德罗尔克博士社区医疗中心);University College, Korea University(韩国大学综合研究生院);Department of Physics, Kyung Hee University(庆熙大学物理系);Potsdam Institute for Climate Impact Research(波茨坦气候影响研究所);Department of Physics, Humboldt University of Berlin(柏林洪堡大学物理系)

Yichao Yao, Minyu Feng, Matjaž Perc, Jürgen Kurths

AI总结:

本文提出一种固定规模动态无标度网络模型,通过状态依赖的随机游走过程描述节点度波动,证明其能稳定生成无标度网络,并通过实验证明该模型能准确复现实际数据。

摘要:

许多现实中的无标度网络,如神经网络和在线通信网络,由固定数量的节点组成,但表现出动态边波动。传统模型常忽视节点数恒定的场景,而优先考虑节点增长。本文突破节点数变化和偏好附着的假设,提出一种创新模型,将节点度波动视为状态依赖的随机游走过程,包含停滞和可变扩散系数。我们证明该模型能生成具有稳定无标度性质的随机动态网络。通过全面的理论和数值分析,我们证明在最低度状态不是吸收状态的情况下,度分布收敛于幂律分布。此外,我们研究了在有意识攻击后网络中最大分量的分数和平均最短路径长度的鲁棒性。通过三个现实网络验证,所提模型能准确复现实际数据。因此,该模型阐明了即使没有增长和偏好附着特征,网络仍可呈现幂律分布,并可用于模拟和研究受攻击固定规模无标度网络的鲁棒性。

英文摘要:

Many real-world scale-free networks, such as neural networks and online communication networks, consist of a fixed number of nodes but exhibit dynamic edge fluctuations. However, traditional models frequently overlook scenarios where the node count remains constant, instead prioritizing node growth. In this work, we depart from the assumptions of node number variation and preferential attachment to present an innovative model that conceptualizes node degree fluctuations as a state-dependent random walk process with stasis and variable diffusion coefficient. We show that this model yields stochastic dynamic networks with stable scale-free properties. Through comprehensive theoretical and numerical analyses, we demonstrate that the degree distribution converges to a power-law distribution, provided that the lowest degree state within the network is not an absorbing state. Furthermore, we investigate the resilience of the fraction of the largest component and the average shortest path length following deliberate attacks on the network. By using three real-world networks, we confirm that the proposed model accurately replicates actual data. The proposed model thus elucidates mechanisms by which networks, devoid of growth and preferential attachment features, can still exhibit power-law distributions and be used to simulate and study the resilience of attacked fixed-size scale-free networks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑