I am required to simulate the propagation of a the wavefront produced by a laser light (about 633 nm) passing through a system of lens and undergoes a few reflection. What I did at the moment is to do ray tracing, record the optical path length of each ray, then connect every point of the same phase to approximate the wavefront. (The light ray will travel about 1 m) Clearly, this would introduce some inaccuracy as there are phenomenon like diffraction and etc. I think it will be more accurate to simulate this using physical optics basis, but how good would the improvement be? More importantly, there are quite some irregular boundaries involve here (lens, for example), which would make the whole computation much more complicated. 1) How can i know the order of discrepancy between geometry optics approximation and physical optics? My application would require error of not more than micrometer range. 2) What are the governing equations should i plan to do the simulation using physical optics?