Adiabatically programmable Ising superpositions

Qu&Co comments on this publication:

Superpositions of bit strings (many-body spin configurations) have been recently proposed as a key to quantum machine learning applications. Adiabatic protocols may serve as an effective method to prepare such states. If the ground state of the final Hamiltonian in an adiabatic protocol is energetically degenerate, the final state of the protocol is a superposition of the configurations in the degenerate manifold. The challenge is to be able to control the dynamics of the protocol such that the amplitudes of the final state can be deterministically programmed. In this paper, Sieberer et al. present a framework to do precisely that. They apply an adiabatic protocol with controlled diabatic transitions to dynamically prepare programmable superpositions, where the control parameters can, even for large systems, be determined efficiently.