Constrained optimization of a zoom lens with CMA-ES algorithm

In the present paper we investigate how optimization algorithm can be tailored to improve the lens design process. We replaced gradient-based optimization methods by the Covariance Matrix Adaptation Evolution Strategy (CMA-ES). This stochastic algorithm is considered more robust and is well suited to avoid local optima often found in optical design. In addition, the algorithm is paired with an augmented Lagrangian method to incorporate constraints handling inside the computation framework. Performances are illustrated on a photographic zoom lens.