Investigating graph isomorphism in cospectral graphs via multiparticle quantum walk in fermionic basis and entanglement entropy

TL;DR

This paper demonstrates that multiparticle quantum walks and entanglement entropy can effectively distinguish non-isomorphic cospectral graphs, advancing methods for graph isomorphism detection using quantum and entanglement-based techniques.

Contribution

It introduces a novel approach combining multiparticle quantum walks in fermionic basis and entanglement entropy to identify non-isomorphic cospectral graphs.

Findings

Quantum walk distinguishes non-isomorphic cospectral graphs.

Entanglement entropy differentiates graph pairs.

Analytical methods applied to specific graph families.

Abstract

We investigate the graph isomorphism (GI) in some cospectral networks. Two graph are isomorphic when they are related to each other by a relabeling of the graph vertices. We want to investigate the GI in two scalable (n + 2)-regular graphs G4(n; n + 2) and G5(n; n + 2), analytically by using the multiparticle quantum walk. These two graphs are a pair of non-isomorphic connected cospectral regular graphs for any positive integer n. In order to investigation GI in these two graphs, we rewrite the adjacency matrices of graphs in the antisymmetric fermionic basis and show that they are different for thesepairs of graphs. So the multiparticle quantum walk is able to distinguish pairs of non- isomorph graphs. Also we construct two new graphs T4(n; n + 2) and T5(n; n + 2) and repeat the same process of G4 and G5 to study the GI problem by using multiparticle quantum walk. Then we study GI by using the entanglement entropy. To this aim, we calculate entanglement entropy between two parts of network. In our model the nodes are considered as identical quantum harmonic oscillators. The entanglement entropy between two special parts of G4(n; n+2) and G5(n; n+2) are calculated analytically. It is shown that the entanglement entropy can distinguish pairs of non-isomorphic cospectral graphs too.