Optimal Quantum Walk Search on Kronecker Graphs with Dominant or Fixed Regular Initiators

TL;DR

This paper proves that certain classes of Kronecker graphs, generated from regular initiators with specific spectral properties, can be efficiently searched using continuous-time quantum walks in optimal or near-optimal time.

Contribution

It establishes sufficient conditions under which Kronecker graphs are optimally searchable by quantum walks, extending previous results to broader classes of graphs.

Findings

Quantum search on Kronecker graphs with dominant eigenvalue initiators is optimal.

Regular, non-bipartite, connected initiators enable optimal quantum search.

Bipartite initiators still allow faster-than-classical search, but not optimal.

Abstract

In network science, graphs obtained by taking the Kronecker or tensor power of the adjacency matrix of an initiator graph are used to construct complex networks. In this paper, we analytically prove sufficient conditions under which such Kronecker graphs can be searched by a continuous-time quantum walk in optimal $\Theta(\sqrt{N})$ time. First, if the initiator is regular and its adjacency matrix has a dominant principal eigenvalue, meaning its unique largest eigenvalue asymptotically dominates the other eigenvalues in magnitude, then the Kronecker graphs generated by this initiator can be quantum searched with probability 1 in $\pi\sqrt{N}/2$ time, asymptotically, and we give the critical jumping rate of the walk that enables this. Second, for any fixed initiator that is regular, non-bipartite, and connected, the Kronecker graphs generated by it are quantum searched in $\Theta(\sqrt{N})$ time. This greatly extends the number of Kronecker graphs on which quantum walks are known to optimally search. If the fixed, regular, connected initiator is bipartite, however, then search on its Kronecker powers is not optimal, but is still better than classical computer's $O(N)$ runtime if the initiator has more than two vertices.