Dr. Liudmila Zhukas
Duke Quantum Center
Abstract:
Trapped-ion processors utilize the qubit's long coherence and all-to-all connectivity mediated by shared motional modes to establish one of the leading platforms for scalable quantum information processing. In this talk, I describe the development of an application-driven system at Duke, detailing the transition from experimental basics—ion trapping, qubit control, and detection—to a software layer that automates scheduling and real-time recalibration to enable reproducible experiments at scale.
Then, I will focus on the digital regime, where universal gate sets enable programmable circuits and closed-loop hybrid quantum–classical optimization under finite-shot constraints. This framework supports diverse applications on the same hardware, including CAFQA-initialized VQE for molecular energies and quantum machine learning, where we perform feature mapping of diverse datasets utilizing contrastive training.
Moving beyond pairwise interactions, I discuss Hamiltonian learning of N-body terms, presenting a symmetry-protected signature that isolates genuine many-body interactions from unknown lower-order patterns. This information is accessed efficiently by probing a subspace that grows linearly with the number of qubits, offering a scalable alternative to full tomography. Finally, I will discuss how such methods are naturally suited to trapped-ion platforms due to their tunable long-range interactions. I contrast the digital regime with Hamiltonian-native analog operation, where engineered Hamiltonians deliver superior scaling for sensing and learning tasks compared to the universal gate approach