Aspects of illumination system optimization

This paper focuses on the facets of illumination system optimization, in particular parameterization of objects, the number of rays that must be traced to sample properly its properties, and the optimization algorithm with the associated merit function designation. Non-interference ensures that the parameterized objects do not erroneously intersect each other or leave gaps during the steps of the optimization procedure. The required number of rays is based on a model developed for television cameras during their initial days of development. Using signal to noise ratio, it provides the number of rays based on the desired contrast, feature size, and allowed error probability. A lightpipe is used to highlight the nuances of this model. The utility of using system symmetry to increase ray count is also discussed. A modified simplex method of optimization is described. This algorithm provides quicker convergence than the standard simplex method, while it is also robust, accurate, and convergent. A previous example using a compound parabolic concentrator highlights the utility of this improvement.