Global Optimization of Damping Ring Designs Using A Multi-Objective Evolutionary Algorithm

Several damping ring designs for the International Linear Collider have been proposed recently. Some of the specifications, such as circumference and bunch train, are not fixed yet. Designers must make a choice anyway, select a geometry type (dog-bone or circular), an arc cell type (TME or FODO), and optimize linear and nonlinear parts of the optics. The design process includes straight-forward steps (usually the linear optics) and some steps not so straightforward (when nonlinear optics optimization is affected by the linear optics). A first attempt at automating this process for the linear optics is reported. We first recognize that the optics is defined by just a few primary parameters (e.g., phase advance per cell) that determine the rest (e.g., quadrupole strength). In addition to the exact specification of circumference, equilibrium emittance, and damping time, there are some other quantities, which could be optimized, that may conflict with each other. A multi-objective genetic optimizer solves this problem by producing a population of best-ranked solutions on a multi-dimensional surface from which one solution can be chosen by the designer. The application of the NSGA-II optimizer to a damping ring of FODO cells is presented.