HDR – Danijela Marković

Danijela Marković will present her HDR entitled « Quantum neuromorphic computing with superconducting circuits » on 10th September 2026 at 2PM in the TRT auditorium.

Summary:

Quantum neuromorphic computing explores how analog quantum systems can learn to transform information through their dynamics. This approach raises two fundamental questions: how to encode information and which observables to measure in order to exploit non-linearities, quantum properties, and the effect of measurement, and how to drive the dynamics and measurement in order to obtain a quantum advantage in information processing.

To explore these questions, we used bosonic modes in superconducting circuits. In the context of computation with a reservoir, we have shown numerically and experimentally that measurements in the Fock basis produce non-linear characteristics even for a Gaussian dynamic, and that the Kerr effect enhances the computational capabilities. We then introduced parametric couplings between the modes in order to make their dynamics trainable, first from differentiable models and then by means of learning rules based directly on physical measurements, without classical simulation of the dynamics.

This work highlights the role of specifically quantum resources, such as coherences, squeezing encoding and non-Gaussian measurements, in the performance of these systems, and paves the way for trainable quantum detectors.