Issues in the Direct Transcription of Optimal Control Problems to Sparse Nonlinear Programs

Abstract

Direct transcription of an optimal control problem into a sparse nonlinear programming problem requires analysis of the interaction between discretization, sparsity, and algorithm efficiency. The relative merits of various discretization techniques with regard to accuracy of the solution and efficiency of the sparse linear algebra and nonlinear program will be presented. Construction of gradient and Hessian information for the nonlinear program will be described. Issues affecting the nonlinear program algorithm strategy will be discussed.