Quantum optimal control with quantum computers: an hybrid algorithm featuring machine learning optimization

TL;DR

This paper presents a hybrid quantum-classical algorithm that uses quantum computers to simulate molecular wavefunctions and machine learning to optimize laser pulses, demonstrating improved control over molecular states.

Contribution

It introduces a novel hybrid quantum-classical approach combining quantum simulation and machine learning for optimal molecular control, with practical implementation insights.

Findings

Outperforms downhill simplex optimization in molecular control tasks.

Achieves performance comparable to advanced algorithms like Rabitz's.

Assesses scalability and current technological limitations on NISQ devices.

Abstract

We develop an hybrid quantum-classical algorithm to solve an optimal population transfer problem for a molecule subject to a laser pulse. The evolution of the molecular wavefunction under the laser pulse is simulated on a quantum computer, while the optimal pulse is iteratively shaped via a machine learning (evolutionary) algorithm. A method to encode on the quantum computer the n-electrons wavefunction is discussed, the circuits accomplishing its quantum simulation are derived and the scalability in terms of number of operations is discussed. Performance on Noisy Intermediate-Scale Quantum devices (IBM Q X2) is provided to assess the current technological gap. Furthermore the hybrid algorithm is tested on a quantum emulator to compare performance of the evolutionary algorithm with standard ones. Our results show that such algorithms are able to outperform the optimization with a downhill simplex method and provide performance comparable to more advanced (but quantum-computer unfriendly) algorithms such as Rabitz's.