Advantages and limitations of quantum routing

TL;DR

This paper investigates the potential advantages of quantum routing over classical methods, analyzing the circuit complexity and spectral properties of quantum architectures, and demonstrating scenarios with significant quantum speedups.

Contribution

It provides bounds on quantum routing complexity based on graph spectral properties and identifies conditions for quantum advantage over classical routing methods.

Findings

Quantum operations can outperform Swap gates in routing tasks.

Spectral properties of the architecture graph influence routing complexity.

Examples show quadratic and linear quantum speedups over classical routing.

Abstract

The Swap gate is a ubiquitous tool for moving information on quantum hardware, yet it can be considered a classical operation because it does not entangle product states. Genuinely quantum operations could outperform Swap for the task of permuting qubits within an architecture, which we call routing. We consider quantum routing in two models: (1) allowing arbitrary two-qubit unitaries, or (2) allowing Hamiltonians with norm-bounded interactions. We lower bound the circuit depth or time of quantum routing in terms of spectral properties of graphs representing the architecture interaction constraints, and give a generalized upper bound for all simple connected $n$-vertex graphs. In particular, we give conditions for a superpolynomial classical-quantum routing separation, which exclude graphs with a small spectral gap and graphs of bounded degree. Finally, we provide examples of a quadratic separation between gate-based and Hamiltonian routing models with a constant number of local ancillas per qubit and of an $\Omega(n)$ speedup if we also allow fast local interactions.