Quantum point spread function for imaging trapped few-body systems with a quantum gas microscope

TL;DR

This paper introduces a quantum point spread function to correct distortions in imaging trapped quantum systems with gas microscopes, enabling accurate reconstruction of original density distributions through machine learning deconvolution.

Contribution

It develops a quantum point spread function model and applies machine learning to deconvolve images, improving the accuracy of density distribution reconstruction in quantum gas microscopy.

Findings

Quantum point spread function accurately models imaging distortions.

Machine learning deconvolution restores original density distributions.

Method applicable to various experimental setups.

Abstract

Quantum gas microscopes, which image the atomic occupations in an optical lattice, have opened a new avenue to the exploration of many-body lattice systems. Imaging trapped systems after freezing the density distribution by ramping up a pinning lattice leads, however, to a distortion of the original density distribution, especially when its structures are on the scale of the pinning lattice spacing. We show that this dynamics can be described by a filter, which we call in analogy to classical optics a quantum point spread function. Using a machine learning approach, we demonstrate via several experimentally relevant setups that a suitable deconvolution allows for the reconstruction of the original density distribution. These findings are both of fundamental interest for the theory of imaging and of immediate importance for current quantum gas experiments.