Dr. Neil Pomphrey
Princeton Plasma Physics Laboratory
Time: 11am, Monday, April 23
Location: Rm. 413, Bld. 11#, Yuquan Campus
Pareto Fronts: Why STELLOPT Needs to Address Them
The physics objective function components in STELLOPT are often competing functions, yet we desire a solution point in design space where the components are simultaneously attractive. STELLOPT combines all objective function components into a single scalar function. A model 2D problem with a non-convex Pareto front of solutions provides an interesting test-bed for the stellarator optimization code. The entire front is obtained exactly for the model by numerical quadrature. A comparison of solutions found by STELLOPT with those of the model indicate shortcomings of the present optimization method, this despite the many successes STELLOPT has had in designing attractive plasma configurations.