Optimization of sensor micro-structures: a general and improved first order method

An improved first order version of the moving asymptotes method is presented within the optimization of sensor micro-structures. In each iteration of the optimization process, a convex approximating subproblem is generated and solved. The subproblem generation is made based on so called “moving asymptotes”. Here, the authors propose a new scheme for calculating the moving asymptotes. It proceeds with curvature normalization of approximation functions between two successive iterations