Physical ray-tracing method for anisotropic optical media in GPGPU

In this paper we discuss the development of a fast ray-tracing solver for complex anisotropic uniaxial optical media based on heterogeneous supercomputing in GPGPU using PyOpenCl. This solver simulates both the propagation of ordinary and extraordinary rays, while taking into account the polarization rotation introduced by position dependent modulations of the optical axis of the medium. We demonstrate the application of this solver by simulating the generation of polarization caustics in random uniaxial optical media.