Matrix optimization on universal unitary photonic devices

Abstract

Universal unitary photonic devices can apply arbitrary unitary transformations to a vector of input modes and provide a promising hardware platform for fast and energy-efficient machine learning using light. We simulate the gradient-based optimization of random unitary matrices on universal photonic devices composed of imperfect tunable interferometers. If device components are initialized uniform-randomly, the locally-interacting nature of the mesh components biases the optimization search space towards banded unitary matrices, limiting convergence to random unitary matrices. We detail a procedure for initializing the device by sampling from the distribution of random unitary matrices and show that this greatly improves convergence speed. We also explore mesh architecture improvements such as adding extra tunable beamsplitters or permuting waveguide layers to further improve the training speed and scalability of these devices.

arXiv comment18 pages, 2 tables, 14 figures, 6 videos (videos provided in Appendix via external URL link)

References in corpus (15)

• Deep Learning with Coherent Nanophotonic Circuits
• Quantum machine learning in feature Hilbert spaces
• Universal Linear Optics
• Boson Sampling on a Photonic Chip
• Quantum transport simulations in a programmable nanophotonic processor
• Training of photonic neural networks through in situ backpropagation
• Self-Configuring Universal Linear Optical Component
• Programmable unitary spatial modes manipulation
• All-optical mode unscrambling on a silicon photonic chip
• Self-aligning universal beam coupler
• Machine learning method for state preparation and gate synthesis on photonic quantum computers
• 60dB high-extinction auto-configured Mach--Zehnder interferometer
• Direct dialling of Haar random unitary matrices
• Benchmarking integrated photonic architectures
• Composite parameterization and Haar measure for all unitary and special unitary groups

Cited by in corpus (36)

• Reprogrammable Electro-Optic Nonlinear Activation Functions for Optical Neural Networks
• Experimentally realized in situ backpropagation for deep learning in nanophotonic neural networks
• Single chip photonic deep neural network with accelerated training
• Hardware error correction for programmable photonics
• Robust architecture for programmable universal unitaries
• Asymptotically Fault-Tolerant Programmable Photonics
• Cryogenic operation of silicon photonic modulators based on DC Kerr effect
• Inverse-design of high-dimensional quantum optical circuits in a complex medium
• Accurate Self-Configuration of Rectangular Multiport Interferometers
• Rapid classification of quantum sources enabled by machine learning
• Optimal design of error-tolerant reprogrammable multiport interferometers
• Addressing the programming challenges of practical interferometric mesh based optical processors
• Further Compactifying Linear Optical Unitaries
• Stability of Self-Configuring Large Multiport Interferometers
• Scalable and Robust Photonic Integrated Unitary Converter
• Developing of a photonic hardware platform for brain-inspired computing based on $5\times 5$ VCSEL arrays
• Universal programmable photonic architecture for quantum information processing
• Architecture agnostic algorithm for reconfigurable optical interferometer programming
• Universal linear optics by programmable multimode interference
• Unitary control of partially coherent waves. I. Absorption
• Realistic quantum photonic neural networks
• Lower-depth programmable linear optical processors
• Unitary control of partially coherent waves. II. Transmission or reflection
• On-chip Optical Phase Monitoring in Multi-Transverse-Mode Integrated Silicon-based Optical Processors
• Bayesian Photonic Accelerators for Energy Efficient and Noise Robust Neural Processing
• Fast reconstruction of programmable integrated interferometers
• Partially Unitary Learning
• Simulating photonic devices with noisy optical elements
• Quantum Channel Learning
• Near-optimal decomposition of unitary matrices using phase masks and the discrete Fourier transform
• Spatiotemporal coupled-mode equations for arbitrary pulse transformation
• Fast reconstruction of programmable interferometers with intensity-only measurements
• Reconfigurable quantum photonic circuits based on quantum dots
• Compactifying linear optical unitaries using multiport beamsplitters
• Optimal universal bounds for waves with varied coherence based on supremum and infimum coherence spectra
• Emergence of biased errors in imperfect photonic circuits