Super-compact scalable optical matrix operations by modularized inverse design

Supporting code for ["Super-compact scalable optical matrix operations by modularized inverse design"] Execution Logic of .m File Code : 1. SVD Decomposition ( SVD.m ) Execute the SVD.m script to: - Generate an arbitrary 10×10 random matrix. - Perform Singular Value Decomposition (SVD) into three matrices: - U (unitary matrix) - S (diagonal matrix of singular values) - V (unitary matrix). - Construct an optical matrix using the U matrix as an example. - Save the U matrix to the directory: *"..\data\Target\10\10_U\"*. 2. Matrix Decomposition ( matrix_decomposition.m ) Run the matrix_decomposition.m script to: - Decompose the 10×10 U matrix (from Step 1) into: - 45 unitary 2×2 matrices (the beamsplitting ratios of these matrices are saved to *"..\data\Target\10\10_ratio\"*) - 1 diagonal matrix D (saved to *"..\data\Target\10\10_D\"*) 3. Optical Structure Construction ( m_dxf.m ) Run m_dxf.m to **build the optical matrix architecture**. **Runtime**: 6.5 minutes on a desktop workstation. **Save** the generated structure to *"..model\model0_10\"*. 4. Fidelity Evaluation ( data_process.m ) Use data_process.m to: * Simulate the optical behavior of the 10×10 matrix. * Calculate and report the matrix fidelity. Citation If you use this code for your research, please cite: [M. Jiang, L. He, and X. Zhang. Super-compact scalable optical matrix operations by modularized inverse design.]