Nodal Domains of Eigenvectors for $1$-Laplacian on Graphs

Abstract

The eigenvectors for graph $1$-Laplacian possess some sort of localization property: On one hand, any nodal domain of an eigenvector is again an eigenvector with the same eigenvalue; on the other hand, one can pack up an eigenvector for a new graph by several fundamental eigencomponents and modules with the same eigenvalue via few special techniques. The Courant nodal domain theorem for graphs is extended to graph $1$-Laplacian for strong nodal domains, but for weak nodal domains it is false. The notion of algebraic multiplicity is introduced in order to provide a more precise estimate of the number of independent eigenvectors. A positive answer is given to a question raised in [{\sl K.~C. Chang, Spectrum of the $1$-Laplacian and Cheeger constant on graphs, J. Graph Theor., DOI: 10.1002/jgt.21871}], to confirm that the critical values obtained by the minimax principle may not cover all eigenvalues of graph $1$-Laplacian.

arXiv commentThis manuscript has been already submitted for publication on November 2, 2015 and the current status is under review

References in corpus (1)

• An Inverse Power Method for Nonlinear Eigenproblems with Applications in 1-Spectral Clustering and Sparse PCA

Cited by in corpus (7)

• Submodular Hypergraphs: p-Laplacians, Cheeger Inequalities and Spectral Clustering
• A simple iterative algorithm for maxcut
• Dual Cheeger Constants, Signless 1-Laplacians and Maxcut
• Homological eigenvalues of graph $p$-Laplacians
• The Pseudo-orthogonality for Graph $1$-Laplacian Eigenvectors and Applications to Higher Cheeger Constants and Data Clustering
• The $1$-Yamabe equation on graph
• Spectrum of signless 1-Laplacian on simplicial complexes