Machine learning aided dimensionality reduction toward a resource efficient projective quantum eigensolver: Formal...

Area of Science:

• Quantum Computing
• Computational Chemistry
• Machine Learning

Background:

• Hybrid quantum-classical algorithms are crucial for molecular simulations on Noisy Intermediate-Scale Quantum (NISQ) devices.
• Current methods require extensive quantum measurements for parameter optimization, leading to long runtimes.
• Reducing quantum hardware dependency is essential for practical applications.

Purpose of the Study:

• To develop a method for drastically reducing quantum measurement requirements in hybrid quantum-classical algorithms.
• To enhance the efficiency of the Projective Quantum Eigensolver (PQE) for calculating ground state energies.
• To create a noise-resilient approach for NISQ devices.

Main Methods:

• An interdisciplinary approach combining quantum computation and supervised machine learning.
• Perceiving nonlinear parameter optimization as a dynamic interplay of fast and slow modes.
• Employing an on-the-fly supervised machine learning protocol to reduce the optimization subspace.
• Tuning the machine learning model to capture noisy NISQ device data.

Main Results:

• A significant reduction in the number of quantum measurements needed for parameter updates.
• Maintained accuracy in calculating ground state energies.
• Demonstrated analytical and numerical validation of the proposed methodology.
• The machine learning model shows resilience to noise inherent in NISQ devices.

Conclusions:

• The proposed machine learning-enhanced approach drastically reduces quantum measurement overhead for hybrid algorithms.
• This method accelerates the computation of molecular ground state energies on NISQ hardware.
• The approach is accurate and robust against noise, paving the way for more efficient quantum simulations.