Quantum-classical tradeoffs and multi-controlled quantum gate decompositions in variational algorithms

Quantum algorithms for unconstrained optimization problems, such as the Quantum Approximate Optimization Algorithm (QAOA), have been proposed as interesting near-term algorithms which operate under a hybrid quantum-classical execution model. Recent work has shown that the QAOA can also be applied to constrained combinatorial optimization problems by incorporating the problem constraints within the design of the variational ansatz - often resulting in quantum circuits containing many multi-controlled gate operations. This paper investigates potential resource tradeoffs for the QAOA when applied to the particular constrained optimization problem of Maximum Independent Set. We consider three variants of the QAOA which make different tradeoffs between the number of classical parameters, quantum gates, and iterations of classical optimization. We also study the quantum cost of decomposing the QAOA circuits on hardware which may support different qubit technologies and native gate sets, and compare the different algorithms using the gate decomposition score which combines the fidelity of the gate operations with the efficiency of the decomposition into a single metric. We find that all three QAOA variants can attain similar performance but the classical and quantum resource costs may vary greatly between them.Comment: 17 pages, 8 figures, 5 table