Abstract We present a direct comparison between QAOA (Quantum Alternating Operator Ansatz), and QA (Quantum Annealing) on 127 qubit problem instances. QAOA with p = 1, 2 rounds is executed on the 127 qubit heavy-hex graph gate-model quantum computer ibm_washington, using on-device grid-searches for angle finding, and QA is executed on two Pegasus-chip D-Wave quantum annealers. The problems are random Ising models whose connectivity matches heavy-hex graphs and the Pegasus graph connectivity, and optionally include hardware-compatible cubic terms ( Z Z Z terms). The QAOA circuits are heavily optimized and of extremely short depth, with a CNOT depth of 6 per round, which allows whole chip usage of the heavy-hex lattice. QAOA and QA are both compared against simulated annealing and the optimal solutions are computed exactly using CPLEX. The noiseless mean QAOA expectation values for p = 1, 2 are computed using classical light-cone based simulations. We find QA outperforms QAOA on the evaluated devices.
Abstract We present a direct comparison between QAOA (Quantum Alternating Operator Ansatz), and QA (Quantum Annealing) on 127 qubit problem instances. QAOA with p = 1, 2 rounds is executed on the 127 qubit heavy-hex graph gate-model quantum computer ibm_washington, using on-device grid-searches for angle finding, and QA is executed on two Pegasus-chip D-Wave quantum annealers. The problems are random Ising models whose connectivity matches heavy-hex graphs and the Pegasus graph connectivity, and optionally include hardware-compatible cubic terms ( Z Z Z terms). The QAOA circuits are heavily optimized and of extremely short depth, with a CNOT depth of 6 per round, which allows whole chip usage of the heavy-hex lattice. QAOA and QA are both compared against simulated annealing and the optimal solutions are computed exactly using CPLEX. The noiseless mean QAOA expectation values for p = 1, 2 are computed using classical light-cone based simulations. We find QA outperforms QAOA on the evaluated devices.
Abstract We show that the quantum approximate optimization algorithm (QAOA) for higher-order, random coefficient, heavy-hex compatible spin glass Ising models has strong parameter concentration across problem sizes from 16 up to 127 qubits for p = 1 up to p = 5, which allows for computationally efficient parameter transfer of QAOA angles. Matrix product state (MPS) simulation is used to compute noise-free QAOA performance. Hardware-compatible short-depth QAOA circuits are executed on ensembles of 100 higher-order Ising models on noisy IBM quantum superconducting processors with 16, 27, and 127 qubits using QAOA angles learned from a single 16-qubit instance using the JuliQAOA tool. We show that the best quantum processors find lower energy solutions up to p = 2 or p = 3, and find mean energies that are about a factor of two off from the noise-free distribution. We show that p = 1 QAOA energy landscapes remain very similar as the problem size increases using NISQ hardware gridsearches with up to a 414 qubit processor.
We investigate and compare the performance of two quantum optimization approaches, the Quantum Approximate Optimization Algorithm (QAOA) and quantum annealing, applied to the Minimum Vertex Cover (MVC) problem. The problem is encoded as an and Ising model, and experiments are conducted on IBM’s GenericBackendV2 noisy superconducting qubit simulator and the D-Wave Advantage2 quantum annealer. Performance is evaluated in terms of solution quality, measurement probability, and proportion of valid solutions. The results we obtained show that, within our experimental setting, quantum annealing consistently outperforms its classical counterpart on small instances, while QAOA, though currently limited by simulation constraints, shows promising behavior that improves with increasing circuit depth. As problem size grows, both approaches exhibit sensitivity to parameter choices such as the penalty term and graph density, underscoring the need for careful tuning. These findings suggest that while both paradigms hold potential for combinatorial optimization, further advances in hardware capabilities and parameter calibration will be necessary to achieve reliable performance on larger instances.
We investigate and compare the performance of two quantum optimization approaches, the Quantum Approximate Optimization Algorithm (QAOA) and quantum annealing, applied to the Minimum Vertex Cover ( MVC ) problem. The problem is encoded as an Ising Hamiltonian, and experiments are conducted on the ibm_kingston gate-based quantum machine, accessed via the IBM Quantum Platform , and on the D-Wave Advantage2 quantum annealer, accessed via the D-Wave Leap quantum cloud service. A classical 2-approximation algorithm is included as a reference baseline. The variational parameters used in the QAOA implementation are selected using the Qiskit AerSimulator with calibration data from the ibm_kingston quantum processor. For quantum annealing, the parameters are determined testing multiple configurations of simulated annealing performed with the SimulatedAnnealingSampler of the dwave-samplers package, and retaining those yielding the best solution quality. Performance is evaluated in terms of solution quality, measurement probability, and proportion of valid solutions. Our results show that, within our experimental setting, quantum annealing consistently outperforms QAOA on instances with up to 10 vertices, and in some cases recovers the exact minimum vertex cover, while the classical 2-approximation baseline serves as a reference point for solution quality. On larger instances, both paradigms exhibit reduced stability and yield fewer valid solutions, with performance becoming increasingly sensitive to parameter choices such as the penalty term.
Recently, a Hamiltonian dynamics simulation was performed on a kicked ferromagnetic 2D transverse field Ising model with a connectivity graph native to the 127 qubit heavy-hex IBM Quantum architecture using ZNE quantum error mitigation. We demonstrate that one of the observables in this Trotterized Hamiltonian dynamics simulation, namely magnetization, can be efficiently simulated on current superconducting qubit-based programmable quantum annealing computers. We show this using two distinct methods: reverse quantum annealing and h-gain state encoding. This simulation is possible because the 127 qubit heavy-hex connectivity graph can be natively embedded onto the D-Wave Pegasus quantum annealer hardware graph and because there exists a direct equivalence between the energy scales of the two types of quantum computers. We derive equivalent anneal pauses in order to simulate the Trotterized quantum circuit dynamics for varying Rx rotations θh∈(0,2π], using quantum annealing processors. Multiple disjoint instances of the Ising model of interest can be embedded onto the D-Wave Pegasus hardware graph, allowing for parallel quantum annealing. We report equivalent magnetization dynamics using quantum annealing for time steps of 20, 50 up to 10,000, which we find are consistent with exact classical 27 qubit heavy-hex Trotterized circuit magnetization dynamics, and we observe reasonable, albeit noisy, agreement with the existing simulations for single site magnetization at 20 Trotter steps. The quantum annealers are able to simulate equivalent magnetization dynamics for thousands of time steps, significantly out of the computational reach of the digital quantum computers on which the original Hamiltonian dynamics simulations were performed.
Quantum Approximate Optimization Algorithm (QAOA) is a prospective candidate for providing quantum advantage in finding approximate solutions to optimization problems using near-term quantum devices. In this paper, the goal is to reduce the number of CNOT gates and the depth of quantum ansatz circuits for QAOA. First, we present a generalized QAOA formulation for any Hamiltonian that involves upto two-body interactions, as a graph. The circuit realization of the depth-p QAOA requires 2mp CNOT gates where the graph for the Hamiltonian has n vertices and m edges. Presently, a CNOT gate is one of the primary sources of error. We propose a graph-theoretic algorithm which can eliminate n−1 CNOTs from the QAOA ansatz while retaining functional equivalence with the original circuit. This improves upon previously proposed Depth First Search (DFS) method by restricting the increase in depth of the circuit, which was a drawback of the method, by ≃84.8% for Erdős-Renyi graphs. Finally, if the underlying connectivity (hardware coupling map) and the initial qubit placement for a hardware are known, then we present how to extend the greedy heuristic method in order to obtain a functionally equivalent circuit with ≃5%; fewer SWAP gates for Erdős-Renyi graphs.
The emergent practical applicability of the Quantum Approximate Optimization Algorithm (QAOA) for approximate combinatorial optimization is a subject of considerable interest. One of the primary limitations of QAOA is the task of finding a set of good parameters, which is usually done using a variational optimization loop. Parameter transfer, or parameter concentration, is a phenomenon where QAOA angles trained on problem instances that are self-similar tend to perform well for other problem instances from that similar class. This suggests a potentially highly efficient and scalable nonvariational learning method for QAOA angle finding. In this work, we systematically study QAOA parameter transferability from small problem sizes (16 and 27 decision variables) onto large problem instances (up to 156 qubits) for heavy-hex graph Ising models with geometrically local higher-order terms using the Julia based QAOA simulation tool to perform classical angle finding for up to 49 QAOA layers ( p ) . Parameter transfer of the fixed angles is validated using a combination of full state vector, projected entangled pair states, matrix product state, and LOWESA numerical simulations. We find that the QAOA parameter transfer from single instances applied to other (unseen) problem instances does not in general provide monotonically improving performance as a function of p —there are many cases where the performance temporarily decreases as a function of p —but despite this the transferred angles have a general trend of improved expectation value as the QAOA depth increases, in many cases converging close to the true ground-state energy of the 100 + qubit instances. We also sample the hardware-compatible Ising models using the ensemble of transfer-learned QAOA parameters on several superconducting qubit IBM quantum processors with 127, 133, and 156 qubits. We find continuous solution quality improvement of the hardware-compatible QAOA circuits run on the IBM noisy intermediate-scale quantum processors up to p = 5 on , up to p = 9 on , and up to p = 10 on .