A small microring array that performs large complex-valued matrix-vector multiplication

• DOI: 10.1007/s12200-022-00009-4
• Publication Date: 2022-04-28T05:00:37Z
• Authors: Junwei Cheng; Yuhe Zhao; Wenkai Zhang; Hailong Zhou; Dongmei Huang; Qing Zhu; Yuhao Guo; Bo Xu; Jianji Dong; Xinliang Zhang
• Keywords: Matrix (chemical analysis)

As an important computing operation, photonic matrix-vector multiplication is widely used in neutral networks and signal processing. However, conventional incoherent focuses on real-valued operations, which cannot work well complex-valued neural discrete Fourier transform. In this paper, we propose a systematic solution to extend the matrix computation of microring arrays from field field, small-scale (i.e., 4 × 4) large-scale 16 16). Combining decomposition partition, our complex multiplier chip can support arbitrary computation. We further demonstrate Walsh-Hardmard transform, cosine image convolutional Our scheme provides path towards breaking limits accelerator optical architecture. More importantly, results reveal that integrated platform huge potential for large-scale, complex-valued, artificial intelligence