Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userFinding nonlinear system equations and complex network structures from data: a sparse optimization approach
123
0
0.0
(
0
)
Authors
Ying-Cheng Lai
Ask ChatGPT about the research
No Arabic abstract
In applications of nonlinear and complex dynamical systems, a common situation is that the system can be measured but its structure and the detailed rules of dynamical evolution are unknown. The inverse problem is to determine the system equations and structure based solely on measured time series. Recently, methods based on sparse optimization have been developed. For example, the principle of exploiting sparse optimization such as compressive sensing to find the equations of nonlinear dynamical systems from data was articulated in 2011 by the Nonlinear Dynamics Group at Arizona State University. This article presents a brief review of the recent progress in this area. The basic idea is to expand the equations governing the dynamical evolution of the system into a power series or a Fourier series of a finite number of terms and then to determine the vector of the expansion coefficients based solely on data through sparse optimization. Examples discussed here include discovering the equations of stationary or nonstationary chaotic systems to enable prediction of dynamical events such as critical transition and system collapse, inferring the full topology of complex networks of dynamical oscillators and social networks hosting evolutionary game dynamics, and identifying partial differential equations for spatiotemporal dynamical systems. Situations where sparse optimization is effective and those in which the method fails are discussed. Comparisons with the traditional method of delay coordinate embedding in nonlinear time series analysis are given and the recent development of model-free, data driven prediction framework based on machine learning is briefly introduced.
rate research
Read More
Stars and cycles are basic structures in network construction. The former has been well studied in network analysis, while the latter attracted rare attention. A node together with its neighbors constitute a neighborhood star-structure where the basic assumption is two nodes interact through their direct connection. A cycle is a closed loop with many nodes who can influence each other even without direct connection. Here we show their difference and relationship in understanding network structure and function. We define two cycle-based node characteristics, namely cycle number and cycle ratio, which can be used to measure a nodes importance. Numerical analyses on six disparate real networks suggest that the nodes with higher cycle ratio are more important to network connectivity, while cycle number can better quantify a node influence of cycle-based spreading than the common star-based node centralities. We also find that an ordinary network can be converted into a hypernetwork by considering its basic cycles as hyperedges, meanwhile, a new matrix called the cycle number matrix is captured. We hope that this paper can open a new direction of understanding both local and global structures of network and its function.
We study the self-organization of the consonant inventories through a complex network approach. We observe that the distribution of occurrence as well as cooccurrence of the consonants across languages follow a power-law behavior. The co-occurrence network of consonants exhibits a high clustering coefficient. We propose four novel synthesis models for these networks (each of which is a refinement of the earlier) so as to successively match with higher accuracy (a) the above mentioned topological properties as well as (b) the linguistic property of feature economy exhibited by the consonant inventories. We conclude by arguing that a possible interpretation of this mechanism of network growth is the process of child language acquisition. Such models essentially increase our understanding of the structure of languages that is influenced by their evolutionary dynamics and this, in turn, can be extremely useful for building future NLP applications.
Spatio-temporally extended nonlinear systems often exhibit a remarkable complexity in space and time. In many cases, extensive datasets of such systems are difficult to obtain, yet needed for a range of applications. Here, we present a method to generate synthetic time series or fields that reproduce statistical multi-scale features of complex systems. The method is based on a hierarchical refinement employing transition probability density functions (PDFs) from one scale to another. We address the case in which such PDFs can be obtained from experimental measurements or simulations and then used to generate arbitrarily large synthetic datasets. The validity of our approach is demonstrated at the example of an experimental dataset of high Reynolds number turbulence.
Sun et al. provided an insightful comment arXiv:1108.5739v1 on our manuscript entitled Controllability of Complex Networks with Nonlinear Dynamics on arXiv. We agree on their main point that linearization about locally desired states can be violated in general by the breakdown of local control of the linearized complex network with nonlinear state. Therefore, we withdraw our manuscript. However, other than nonlinear dynamics, our claim that a single-node-control can fully control the general bidirectional/undirected linear network with 1D self-dynamics is still valid, which is similar to (but different from) the conclusion of arXiv:1106.2573v3 that all-node-control with a single signal can fully control any direct linear network with nodal-dynamics (1D self-dynamics).
We discuss the problem of extending data mining approaches to cases in which data points arise in the form of individual graphs. Being able to find the intrinsic low-dimensionality in ensembles of graphs can be useful in a variety of modeling contexts, especially when coarse-graining the detailed graph information is of interest. One of the main challenges in mining graph data is the definition of a suitable pairwise similarity metric in the space of graphs. We explore two practical solutions to solving this problem: one based on finding subgraph densities, and one using spectral information. The approach is illustrated on three test data sets (ensembles of graphs); two of these are obtained from standard graph generating algorithms, while the graphs in the third example are sampled as dynamic snapshots from an evolving network simulation. We further incorporate these approaches with equation free techniques, demonstrating how such data mining approaches can enhance scientific computation of network evolution dynamics.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
