Optimal quantum control of Bose-Einstein condensates in magnetic microtraps: Comparison of GRAPE and Krotov optimization schemes

TL;DR

This paper compares two optimal control methods, GRAPE and Krotov, for manipulating Bose-Einstein condensates in microtraps, highlighting their differences in performance, smoothness, and convergence speed.

Contribution

It provides a systematic comparison of GRAPE and Krotov methods applied to Bose-Einstein condensate control, demonstrating their respective advantages and limitations.

Findings

GRAPE yields smoother control fields and is easier to use.

Krotov converges faster with appropriate step size.

Both methods effectively control condensate dynamics.

Abstract

We study optimal quantum control of the dynamics of trapped Bose-Einstein condensates: The targets are to split a condensate, residing initially in a single well, into a double well, without inducing excitation; and to excite a condensate from the ground to the first excited state of a single well. The condensate is described in the mean-field approximation of the Gross-Pitaevskii equation. We compare two optimization approaches in terms of their performance and ease of use, namely gradient ascent pulse engineering (GRAPE) and Krotov's method. Both approaches are derived from the variational principle but differ in the way the control is updated, additional costs are accounted for, and second order derivative information can be included. We find that GRAPE produces smoother control fields and works in a black-box manner, whereas Krotov with a suitably chosen step size parameter converges faster but can produce sharp features in the control fields.