Surrogate optimization of variational quantum circuits.

Area of Science:

• Quantum computing
• Computational physics
• Optimization algorithms

Background:

• Variational quantum eigensolvers (VQE) are promising near-term quantum algorithms.
• Current VQE methods face challenges with optimization in noisy environments, limiting practical applications and quantum advantage claims.
• Improved convergence is crucial for accelerating the capabilities of near-term quantum hardware.

Purpose of the Study:

• To develop and demonstrate a novel surrogate optimization approach for variational quantum algorithms.
• To address the challenges of optimization convergence in the presence of noise for VQE.
• To enhance the efficiency and applicability of hybrid quantum-classical methods.

Main Methods:

• Utilized modern circuit simulation and stochastic classical optimization techniques.
• Developed a surrogate optimization approach combining classical (CPU/GPU) approximate state vector simulators with quantum processing units (QPUs).
• Employed an approximate Hessian calculated via classical simulation as input for quantum or exact circuit simulators, enabling parallelization across QPUs.

Main Results:

• Successfully implemented a surrogate optimization method for quantum circuits.
• Demonstrated the approach's effectiveness with and without sampling noise.
• Conducted a proof-of-principle demonstration on a 40-qubit quantum processing unit.

Conclusions:

• The surrogate optimization approach significantly improves convergence for variational quantum algorithms.
• This method offers a viable strategy to accelerate near-term quantum hardware capabilities for VQE and other hybrid optimization tasks.
• The parallelizable nature of the approach makes it suitable for scaling across multiple quantum processing units.