On unitary reconstruction of linear optical networks

TL;DR

This paper introduces a novel method for reconstructing the unitary transformation of linear optical networks using only input-output data, significantly improving accuracy and robustness, especially in the presence of losses, demonstrated through simulations up to 14 modes.

Contribution

The paper presents a new approach to unitary reconstruction that outperforms existing methods, particularly in handling mode-dependent losses in large optical networks.

Findings

Improved reconstruction fidelity over existing methods.

Effective recovery of mode-dependent losses.

Scalability demonstrated up to 14 modes.

Abstract

Linear optical elements are pivotal instruments in the manipulation of classical and quantum states of light. The vast progress in integrated quantum photonic technology enables the implementation of large numbers of such elements on chip while providing interferometric stability. As a trade-off these structures face the intrinsic challenge of characterizing their optical transformation as individual optical elements are not directly accessible. Thus the unitary transformation needs to be reconstructed from a dataset generated with having access to the input and output ports of the device only. Here we present a novel approach to unitary reconstruction that significantly improves upon existing approaches. We compare its performance to several approaches via numerical simulations for networks up to 14 modes. We show that an adapted version of our approach allows to recover all mode-dependent losses and to obtain highest reconstruction fidelities under such conditions.