Empirical Quantum Advantage in Constrained Optimization from Encoded Unitary Designs

TL;DR

This paper introduces CE-QAOA, a shallow, constraint-aware quantum algorithm that leverages encoded unitary designs to achieve quantum advantage in constrained optimization problems, demonstrated through traveling salesman problem simulations.

Contribution

The paper presents a novel, depth-efficient quantum ansatz with encoded unitary designs, enabling polynomial-time hybrid solvers and demonstrating quantum advantage over classical methods.

Findings

Achieves a Theta(n^r) reduction in shot complexity for fixing r locations.

Demonstrates an exponential minimax separation against classical baselines.

Recovers global optima in TSP instances with up to 10 locations using polynomial resources.

Abstract

We introduce the Constraint-Enhanced Quantum Approximate Optimization Algorithm (CE-QAOA), a shallow, constraint-aware ansatz that operates inside the one-hot product space [n]^m, where m is the number of blocks and each block is initialized in an n-qubit W_n state. We give an ancilla-free, depth-optimal encoder that prepares W_n using n-1 two-qubit rotations per block, and a two-local block-XY mixer that preserves the one-hot manifold and has a constant spectral gap on the one-excitation sector. At the level of expressivity, we establish per-block controllability, implying approximate universality per block. At the level of distributional behavior, we show that, after natural block and symbol permutation twirls, shallow CE-QAOA realizes an encoded unitary 1-design and supports approximate second-moment (2-design) behavior; combined with a Paley-Zygmund argument, this yields finite-shot anticoncentration guarantees. Algorithmically, we wrap constant-depth sampling with a deterministic feasibility checker to obtain a polynomial-time hybrid quantum-classical solver (PHQC) that returns the best observed feasible solution in O(S n^2) time, where S is a polynomial shot budget. We obtain two advantages. First, when CE-QAOA fixes r >= 1 locations different from the start city, we achieve a Theta(n^r) reduction in shot complexity even against a classical sampler that draws uniformly from the feasible set. Second, against a classical baseline restricted to raw bitstring sampling, we show an exp(Theta(n^2)) minimax separation. In noiseless circuit simulations of traveling salesman problem instances with n in {4,...,10} locations from the QOPTLib benchmark library, we recover the global optimum at depth p = 1 using polynomial shot budgets and coarse parameter grids defined by the problem size.