Extrapolation method to optimize linear-ramp QAOA parameters: Evaluation of QAOA runtime scaling
read the original abstract
The Quantum Approximate Optimization Algorithm (QAOA) has been suggested as a promising candidate for the solution of combinatorial optimization problems. Yet, whether - or under what conditions - it may offer an advantage compared to classical algorithms remains to be proven. Using the standard variational form of QAOA requires a high number of circuit parameters that have to be optimized at a sufficiently large depth, which constitutes a bottleneck for achieving a potential scaling advantage. The linear-ramp QAOA (LR-QAOA) has been proposed to address this issue, as it relies on only two parameters which have to be optimized. Based on this, we develop a method to estimate suitable values for those parameters through extrapolation, starting from smaller problem sizes (number of qubits) towards larger problem sizes. We apply this method to several use cases such as portfolio optimization, feature selection, clustering and weighted maxcut. From results obtained on a noiseless quantum emulator, we evaluate the quantum runtime scaling for finding the optimal solution and compare it with that of classical methods. In the case of portfolio optimization, we demonstrate superior scaling compared to the classical runtime for the problem sizes of up to $28$ qubits that we consider in this work.
This paper has not been read by Pith yet.
Forward citations
Cited by 2 Pith papers
-
Nonvariational quantum optimisation approaches to pangenome-guided sequence assembly
Iterative-QAOA solves pangenome assembly instances on current quantum hardware by using a fixed-ramp QAOA schedule with warm-start updates and a new HUBO encoding that cuts variables from O(N^{2}) to O(N log N).
-
Hybrid Quantum-Classical Optimization Workflows for the Shipment Selection Problem
Hybrid Iterative-QAOA warm starts improve shipment delivery by up to 12% and cut drive distance by 6% on real logistics data when fed to a classical solver.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.