Differential ray tracing for an efficient computation of the Jacobian in the damped least-squares algorithm

\nWe present a framework for the optimization of optical systems based up on the damped least-squares method and differential ray tracing. First, a mathematical description for the differentiation of the ray-surface intersection is introduced. It is observed that the structure of the optimization variables and operands can be exploited such that the Jacobian for the damped least-squares algorithm can be computed in the same computational complexity as the primal ray tracing for both forward and reverse mode algorithmic differentiation. To demonstrate the advantages of the method a freeform mirror system is analyzed.\n