Photonic Nonlinear Synthesis by Vector-Spatial Dimension Expansion

The limited diversity and reconfigurability greatly hinder the application and development of optical nonlinearity. To address this issue, we propose a photonic nonlinear synthesizer (PNS) by vector-spatial dimension expansion. By virtue of the linear transformation, the PNS can realize arbitrary nonlinearity of the discrete input signal theoretically. We experimentally demonstrate an integrated PNS with eight nonlinear channels and utilize it to realize various discrete nonlinear functions, including basic logic gates (AND, OR, and XOR), the 2-bit decimal to binary converter, and the state machine to predict the next day of the week. Furthermore, two continuous nonlinear functions of sigmoid and sine are implemented by curve fitting of the PNS. We believe the proposed method will enrich the diversity of optical nonlinearity and prompt the development of optical nonlinear computing.