Spectra of power hypergraphs and signed graphs via parity-closed walks

Lixiang Chen, Edwin R. van Dam, Changjiang Bu

Research output: Contribution to journalArticleScientificpeer-review

2 Downloads (Pure)

Abstract

The k-power hypergraph G(k) is the k-uniform hypergraph that is obtained by adding k−2 new vertices to each edge of a graph G, for k≥3. A parity-closed walk in G is a closed walk that uses each edge an even number of times. In an earlier paper, we determined the eigenvalues of the adjacency tensor of G(k) using the eigenvalues of signed subgraphs of G. Here, we express the entire spectrum (that is, we determine all multiplicities and the characteristic polynomial) of G(k) in terms of parity-closed walks of G. Moreover, we give an explicit expression for the multiplicity of the spectral radius of G(k). Our results are mainly obtained by exploiting the so-called trace formula to determine the spectral moments of G(k). As a side result, we show that the number of parity-closed walks of given length is the corresponding spectral moment averaged over all signed graphs with underlying graph G. We also extrapolate the characteristic polynomial of G(k) to k=2, thereby introducing a pseudo-characteristic function. Among other results, we show that this function is the geometric mean of the characteristic polynomials of all signed graphs on G and characterize when it is a polynomial. This supplements a result by Godsil and Gutman that the arithmetic mean of the characteristic polynomials of all signed graphs on G equals the matching polynomial of G.
Original languageEnglish
Article number105909
JournalJournal of Combinatorial Theory Series A
Volume207
DOIs
Publication statusPublished - Oct 2024

Keywords

  • power hypergraphs
  • adjacency tensor
  • signed graphs
  • spectral moments
  • spectral radius
  • closed walks
  • characteristic polynomial

Fingerprint

Dive into the research topics of 'Spectra of power hypergraphs and signed graphs via parity-closed walks'. Together they form a unique fingerprint.

Cite this