Leveraging Structure in Physical Systems for Retrieval of Unseen Quantities

Extracting meaningful insight from experiments lies at the heart of scientific discovery. However, in many systems critical properties are often hidden or entirely inaccessible to direct measurement. These challenges are often amplified in quantum many-body systems, where the exponential growth of the state space with system size makes both simulation and interpretation increasingly difficult. Focusing on cold atom experiments, I will present two projects that target such challenges through
computational approaches. In the first, I will present a machine learning framework for identification of phase transitions in quantum many-body systems directly from raw, incomplete, measurements, without relying on any labels, models, or prior assumptions. We detect transitions such as many-body localization and Mott-to-superfluid phases in experiments, even where conventional detection heuristics fail. Our method offers a new tool for exploring quantum phenomena in large-scale quantum simulators. In the second project, I will describe a novel algorithmic method for reconstructing the phase of atomic matter waves from density-only measurements, bypassing the need for complex atom interferometry. I will showcase the first experimental demonstration of such a technique and will discuss subtle fundamental issues arising from the nature of the quantum measurements.

Ph.D. Student of Distinguished Professor Moti Segev.