De-correlated, Robust, End to End Design for Multiparameter Quantum Sensors

Quantum sensing platforms have emerged as a near-term application of quantum devices for precision measurements of physical quantities. Various protocols for designing quantum sensors exist for sensing both a single or multiple quantities of interest (for example, [1,2]). A current challenge, however, is engineering a quantum sensor to have maximum sensitivity to the parameters we wish to measure while being highly robust to noise in design parameters. We present a novel way of achieving both goals in a quantum sensor and de-correlating the sensitivities for different parameters. Our approach relies on recent advances in the theory of multiparameter classical and quantum fisher information [3,4], and can be implemented in an end-to-end fashion, where the sensor's design is globally optimized rather than component by component optimization.

As an illustrative use case, we demonstrate a robust shaken lattice atom interferometer design that is maximally sensitive to applied accelerations and insensitive to noise in lattice parameters. We optimize the interferometer using reinforcement learning (RL) methods[1,5] and show a multi-arm design that exceeds the maximal sensitivity obtained by Mach-Zehnder lattice atom interferometers.

[1] Liang-Ying Chih and Murray Holland, Phys. Rev. Research 3, 033279 – Published 27 September 2021

[2] Raphael Kaubruegger, Athreya Shankar, Denis V. Vasilyev, and Peter Zoller, PRX Quantum 4, 020333 – Published 1 June 2023

[3] Jing Liu et al 2020 J. Phys. A: Math. Theor. 53 023001

[4] Jarrod T. Reilly, John Drew Wilson, Simon B. Jäger, Christopher Wilson, and Murray J. Holland, Phys. Rev. Lett. 131, 150802 – Published 11 October 2023

[5] Le Desma et al. ,arXiv:2305.17603