Quantum Approximate and Quantum Walk Optimization Approaches to Set Balancing
AI Breakdown
Get a structured breakdown of this paper — what it's about, the core idea, and key takeaways for the field.
Abstract
We explore the application of variational quantum algorithms to the NP-hard set balancing problem, a critical challenge in clinical trial design and experimental scheduling. The problem is mapped to an Ising model, with tailored Quadratic Unconstrained Binary Optimization (QUBO) formulations and cost Hamiltonians expressed in Pauli-Z form. We implement both the Quantum Approximate Optimization Algorithm (QAOA) and the Quantum Walk Optimization Algorithm (QWOA), evaluating them in separate experimental settings. For QAOA, we perform a comparative analysis of six mixer Hamiltonians (X, XY, Full-SWAP, Ring-SWAP, Grover, and Warm-Started), employing scaled-exponential Pauli-string realizations of the mixer unitaries, which yield superior performance over conventional circuit decompositions. Additionally, we introduce a Shannon-entropy-based post-processing technique that refines solutions by maximizing feature-distribution uniformity across partitions. These results underscore the importance of mixer choice and circuit implementation in enhancing QAOA performance for combinatorial optimization.