Using Differential Evolution to avoid local minima in Variational Quantum Algorithms

Daniel Faílde,José Daniel Viqueira,Mariamo Mussa Juane,Andrés Gómez
DOI: https://doi.org/10.1038/s41598-023-43404-3
2023-09-29
Abstract:Variational Quantum Algorithms (VQAs) are among the most promising NISQ-era algorithms for harnessing quantum computing in diverse fields. However, the underlying optimization processes within these algorithms usually deal with local minima and barren plateau problems, preventing them from scaling efficiently. Our goal in this paper is to study alternative optimization methods that can avoid or reduce the effect of these problems. To this end, we propose to apply the Differential Evolution (DE) algorithm to VQAs optimizations. Our hypothesis is that DE is resilient to vanishing gradients and local minima for two main reasons: (i) it does not depend on gradients, and (ii) its mutation and recombination schemes allow DE to continue evolving even in these cases. To demonstrate the performance of our approach, first, we use a robust local minima problem to compare state-of-the-art local optimizers (SLSQP, COBYLA, L-BFGS-B and SPSA) against DE using the Variational Quantum Eigensolver algorithm. Our results show that DE always outperforms local optimizers. In particular, in exact simulations of a 1D Ising chain with 14 qubits, DE achieves the ground state with a 100\% success rate, while local optimizers only exhibit around 40\%. We also show that combining DE with local optimizers increases the accuracy of the energy estimation once avoiding local minima. Finally, we demonstrate how our results can be extended to more complex problems by studying DE performance in a 1D Hubbard model.
Quantum Physics
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the problems of local minima and barren plateaus encountered in the optimization process of Variational Quantum Algorithms (VQAs). These problems lead to the inefficiency of VQAs in the optimization process, especially when the problem scale increases, the optimization difficulty increases significantly. Specifically, the goal of the paper is to study alternative optimization methods that can avoid or reduce the influence of local minima and barren plateaus, thereby improving the performance and scalability of VQAs. To achieve this goal, the author proposes to use the Differential Evolution (DE) algorithm to optimize VQAs. The DE algorithm is characterized by being gradient - independent, and its mutation and recombination mechanisms can continue to evolve when encountering local minima and barren plateaus. By comparing with several existing local optimizers (such as SLSQP, COBYLA, L - BFGS - B and SPSA), the author shows the superior performance of the DE algorithm in optimizing the Variational Quantum Eigensolver (VQE). The experimental results show that the DE algorithm can not only successfully avoid local minima, but also maintain a high optimization success rate on larger - scale problems. In addition, the paper also explores the possibility of using the DE algorithm in combination with other optimization methods to further improve the accuracy and convergence speed of optimization. For example, by first using the DE algorithm to avoid local minima and then using a gradient - based optimizer (such as L - BFGS - B) for fine - tuning optimization, the quality of the final optimization result can be significantly improved. Finally, the author also applies the DE algorithm to the more complex 1D Hubbard model, verifying its potential in dealing with strongly correlated systems.