ABSTRACT
As Moore’s Law winds down, optical computing has been reignited as a promising alternative to electronics, for its ultra-high speed and power efficient on-chip optical interconnects and computing. Based on the computation paradigms, optical computing can be classified into digital and analog computing. For the first category, intensive study has been made on both optical logic units and generic logic synthesis algorithms. For the second category, research efforts have been made on matrix multiplication, and optical neural networks (ONNs), evolving from the former. In this chapter, we introduce the principles of the two computing paradigms as well as recent breakthrough towards automated design for better scalability and robustness.
