The series of IQIS Conferences brings together researchers in the field of Quantum Information Science from Italy and from around the world, fostering active collaborations among senior and junior researchers, postdocs and students. In this edition of the conference, our 3rd-year PhD student Danilo Zia has been awarded with the best poster prize. Danilo’s contribution was about the papers “Dynamical learning of a photonics quantum-state engineering process” and “Regression of high-dimensional angular momentum states of light“, here we report the abstract submitted for the conference:
The engineering of arbitrary high-dimensional quantum states (qudit) is a pivotal task in quantum information science. In particular, the ability to produce the desired states with high fidelity and independently of the environmental situation is a highly demanded feature. Several strategies and platforms have been proposed to accomplish this task. However, the inevitable presence of noises, losses, and the lack of accurate characterization of the apparatus significantly decreases the performance. In this work, we focused on the generation of qudits encoded in the Orbital Angular Momentum (OAM) of photons through a quantum walk-based experimental platform. We addressed the engineering with an adaptive optimization protocol that, working in a fully black-box scenario, tunes the relevant parameters relying only on the measured fidelity, without needing a description of the generation setup and automatically accounting for experimental noises. We also studied the certification of the produced states with a machine learning-based approach. Here the measurement is treated as a regression task in which we retrieve the coefficients of arbitrary OAM superpositions, in this way reducing the losses present in interferometric and diffractive measurement approaches. The high values of the fidelities obtained in a 4-dimensional Hilbert space showcase the performance of both approaches.
Danilo Zia won best poster award at IQIS 23023!
