A Minimax Fitting Algorithm for Ultra-Precision Aspheric Surfaces

Aspheric lenses show significant superiority over traditional spherical ones. The peak-to-valley form deviation is an important criterion for surface qualities of optical lenses. The peak-to-valley errors obtained using traditional methods are usually greater than the actual values, which in turn causing unnecessary rejections. In this paper the form errors of aspheric surfaces are evaluated in the sense of minimum zone, i.e. to directly minimize the peak-to-valley deviation of the data points with respect to the nominal surface. A powerful heuristic optimization algorithm, called differential evolution (DE) is adopted. The control parameters are obtained by meta-optimization. Normally the number of data points is very large, which makes the optimization program unacceptably slow. To improve the efficiency, alpha-shapes are employed to decrease the number of data points involved in the DE optimization. Finally numerical examples are presented to validate this minimum zone evaluation method and compare its results with other algorithms.