Points-connecting neural network ray tracing

Unsupervised neural network ray tracing (NNRT) to calculate a light ray path connecting given points in a gradient-index medium is proposed here. If two points are given, the NNRT can provide a light ray path passing through these points without knowledge of the light ray direction. Maxwell's fisheye lens having a spherical gradient-index is used to demonstrate how well the NNRT works. Light rays calculated using the NNRT are shown to trace an ideal path passing through given points. (C) 2021 Optical Society of America

Automated summary

Unsupervised neural network ray tracing (NNRT) is proposed to calculate light ray paths connecting given points in a gradient-index medium. Using Maxwell's fisheye lens as an example, NNRT is shown to accurately trace ideal light ray paths passing through given points.