Generative models based on eigendecomposition for dense ray tracing.

Area of Science:

• Computational physics
• Geometric optics
• Numerical methods

Background:

• Ray tracing is crucial for simulating wave propagation.
• Existing methods can be computationally intensive, especially in complex media.
• Accurate emulation of ray trajectories is needed for efficient simulation.

Purpose of the Study:

• To develop an algorithm for emulating ray trajectories based on the least action principle.
• To create a method agnostic to the underlying physics of raypath generation.
• To efficiently populate scenarios with dense arrays of emulated rays.

Main Methods:

• Spectral decomposition of geometric shapes derived from raypaths.
• Shape analysis independent of specific ray tracing techniques.
• Algorithm trained on a set of representative raypaths.

Main Results:

• The algorithm efficiently emulates ray trajectories in mildly heterogeneous media.
• Emulated rays show close agreement with actual rays.
• The method effectively identifies regions with high ray variation and potential shadow zones.

Conclusions:

• The proposed algorithm offers an efficient way to emulate ray trajectories.
• It provides a versatile tool applicable across different ray tracing methods.
• The algorithm aids in identifying complex wave propagation phenomena like shadow zones.