Litcius/Paper detail

Learning arbitrary complex matrices by interlacing amplitude and phase masks with fixed unitary operations

Matthew Markowitz, Kevin Zelaya, Mohammad‐Ali Miri

2024Physical review. A/Physical review, A11 citationsDOI

Abstract

Programmable photonic integrated circuits are an emerging technology that amalgamates photonics and electronics, paving the way for light-based information processing at high speeds and low power consumption. Considering its wide range of applications as one of the most fundamental mathematical operations, there has been considerable interest in developing reliably implementable programmable circuit architectures that perform matrix-vector multiplication. Recently, it was shown that discrete unitary operations can be parameterized by interlacing fixed operators with diagonal phase parameters realized with phase shifter arrays. We show that these decompositions are a special case of a broader class of factorizations that enable parametrization of arbitrary complex matrices. The proposed representation of an $N\ifmmode\times\else\texttimes\fi{}N$ matrix is given by $N+1$ amplitude-and-phase-modulation layers interlaced with a fixed unitary layer that can be implemented, for example, via a coupled waveguide array. Thus, we introduce an architecture for physically implementing discrete linear operations, enabling the development of novel families of programmable photonic circuits for on-chip analog information processing.

Topics & Concepts

InterlacingUnitary stateUnitary matrixPhase (matter)AmplitudeMathematicsPure mathematicsAlgebra over a fieldComputer sciencePhysicsOpticsQuantum mechanicsComputer visionPolitical scienceLawNeural Networks and Reservoir ComputingOptical Network TechnologiesNeural Networks and Applications
Learning arbitrary complex matrices by interlacing amplitude and phase masks with fixed unitary operations | Litcius