Simulation study demonstrates modularized inverse design for scalable optical matrix operations, indicating a rapid route to compact photonic computing architectures.
Key Points
To establish and validate a modularized inverse design framework that rapidly generates scalable and compact optical matrix processors.
Applied singular value decomposition to decompose arbitrary 10×10 matrices into unitary and diagonal components.
Decomposed the resulting 10×10 unitary matrix into 45 2×2 unitary matrices and a diagonal matrix to determine beamsplitting ratios.
Generated the complete optical matrix architecture layout and simulated optical performance to compute matrix fidelity.
Generated the complete 10×10 optical matrix architecture in 6.5 minutes on a standard desktop workstation.
Successfully mapped arbitrary unitary transformations into modular 2×2 optical splitting building blocks for fidelity verification.