Finite element methods (FEMs) are a popular simulation technique for discretizing partial differential equations (PDEs) to an algebraic problem. For well-studied problems, theory shows which methods are optimal and can be applied to a wide variety of applications. Building a FEM model for a new or complex problem can be a labor intensive, error prone task. The difficulties have been somewhat alleviated by automation, but the tools often lag in performance to hand optimized, specialized code. Because of the ease of specifying new models and solver techniques in automated solvers, it is an ideal place to look at challenging problems and evaluate many methods. By developing and using automation techniques, we study a challenging class of problems: complex fluids. |