Convex optimization using quantum oracles

TL;DR

This paper explores how quantum algorithms can significantly accelerate convex optimization by reducing the number of oracle queries needed, achieving exponential speed-ups over classical methods.

Contribution

It introduces quantum algorithms for implementing separation and optimization oracles with improved query complexities, including exponential speed-ups and new lower bounds.

Findings

Quantum algorithms implement separation oracles with $ ilde{O}(1)$ queries.

Quantum methods compute approximate subgradients efficiently.

Quantum query complexity for convex optimization is significantly reduced.

Abstract

We study to what extent quantum algorithms can speed up solving convex optimization problems. Following the classical literature we assume access to a convex set via various oracles, and we examine the efficiency of reductions between the different oracles. In particular, we show how a separation oracle can be implemented using $\tilde{O}(1)$ quantum queries to a membership oracle, which is an exponential quantum speed-up over the $\Omega(n)$ membership queries that are needed classically. We show that a quantum computer can very efficiently compute an approximate subgradient of a convex Lipschitz function. Combining this with a simplification of recent classical work of Lee, Sidford, and Vempala gives our efficient separation oracle. This in turn implies, via a known algorithm, that $\tilde{O}(n)$ quantum queries to a membership oracle suffice to implement an optimization oracle (the best known classical upper bound on the number of membership queries is quadratic). We also prove several lower bounds: $\Omega(\sqrt{n})$ quantum separation (or membership) queries are needed for optimization if the algorithm knows an interior point of the convex set, and $\Omega(n)$ quantum separation queries are needed if it does not.