ANL, Argonne
Optimization tools are needed in every step of an accelerator project, from the design to commissioning to operations. However, different phases have different optimization needs that may require different optimization algorithms. For example, a global optimizer is more appropriate in the design phase to map the whole parameter space whereas a local optimizer with a shorter path to solution is more adequate during operations to find the next best operating point. Different optimization algorithms are being used in accelerator physics, we mention in particular standard algorithms based on least square minimization and evolutionary algorithms such as genetic optimizers. Over the years, we have developed several optimization tools for beam tracking codes to include 3D fields and SC effects. Including particle tracking in the optimization process calls for parallel computing. We will review the different algorithms and their implementation and present few highlight applications.
