Optical Design Services

The algorithm described in the article has been implemented in the macro language of the commercial optical design program Code V. Here, we present several results obtained for merit function spaces of triplets. In our example the triplet forms an image of an object at infinity and we have used the first five curvatures as variables. The image plane was placed at its paraxial position and the curvature of the last surface has been solved to keep a constant effective focal length. The default Code V merit function (which is based on transverse ray aberrations) was used. The resulting network consists of 18 local minima and 20 saddle points. In this network, each node represents a point in the five-dimensional design space and the lines connecting nodes represent optimization paths that have been generated from the saddle points. As expected, the local minima form a network where they are all linked through optimization paths generated from the saddle points. This network, which has been detected in five dimensions, can be visualized in a two-dimensional graph. This allows us to examine the relationship between the various minima independently of the dimensionality of the merit function space.