Title: The many layers of control and parameterization: from closed-form solutions to machine learning

Speaker: Stefan Krastanov

Abstract: Since the 90s our community has been developing methods for the complete control of various quantum states. Initially we started with explicit constructions prescribing how to compile an arbitrary unitary operation out of the available control fields, but over the years optimal control, and its more recent reincarnation — machine learning, have become popular, providing much lower resource overhead. In this talk we will discuss the evolution of these techniques, including in-depth examples from cavity-QED and nonlinear optics. At the end we will touch upon how we can keep some of the benefits of the machine learning approaches, without sacrificing the interpretability of closed-form control constructs.

references:

– (1996) Arbitrary Control of a Quantum Electromagnetic Field

10.1103/PhysRevLett.76.1055

– Universal Control of an Oscillator with Dispersive Coupling to a Qubit

arXiv:1502.08015

– Implementing a Universal Gate Set on a Logical Qubit Encoded in an Oscillator arXiv:1608.02430

– Room-Temperature Photonic Logical Qubits via Second-Order Nonlinearities arXiv:2002.07193

– Efficient cavity control with SNAP gates arXiv:2004.14256

– Stochastic Estimation of Dynamical Variables arXiv:1812.05120

– Unboxing Quantum Black Box Models: Learning Non-Markovian Dynamics

arXiv:2009.03902