Quantum Optimal Control in a Chopped Basis: Applications in Control of Bose-Einstein Condensates

TL;DR

This paper compares various quantum optimal control methods for manipulating Bose-Einstein condensates, demonstrating that the GROUP approach outperforms traditional methods in convergence speed and control quality.

Contribution

It introduces a combined gradient optimization method called GROUP that incorporates control electronics bandwidth and shows superior performance over existing techniques.

Findings

GROUP converges faster than Nelder-Mead with CRAB

GROUP achieves better control results than GRAPE and Krotov

Gradient-based methods outperform derivative-free methods in this context

Abstract

We discuss quantum optimal control of Bose-Einstein condensates trapped in magnetic microtraps. The objective is to transfer a condensate from the ground state to the first-excited state. This type of control problem is typically solved using derivative-based methods in a high-dimensional control space such as gradient-ascent pulse engineering (GRAPE) and Krotov's method or derivative-free methods in a reduced control space such as Nelder-Mead with a chopped random basis (CRAB). We discuss how these methods can be combined in gradient optimization using parametrization (GROUP) including the finite bandwidth of the control electronics. We compare these methods and find that GROUP converges much faster than Nelder-Mead with CRAB and achieves better results than GRAPE and Krotov's method on the control problem presented here.