1: \begin{abstract}

2: \commentout{Direct transcription with collocation polynomials on Legendre-Gauss-Radau points together with a sparse non-linear programming method is a widely used approach for numerically solving optimal control problems, which can be addressed to three reasons: First, the simplicity of implementing the collocation method. Second, the capabilities of the Radau discretization when the problem is stiff. And finally because the collocation polynomials can be chosen of arbitrary order, which yields high-order consistency. However it is known from examples that this method does not converge in the general case because it can be unstable. In particular it can diverge for bang-singular arcs and for differential-algebraic problems with a high index. Further it is undesirable that all states and controls must be discretized on the same mesh, especially when some states are much smoother or, e.g., constant.}

3: 

4: In this paper we present a finite element method for the direct transcription of constrained non-linear optimal control problems. 

5: 

6: We prove that our method converges of high order under mild assumptions. Our analysis uses a regularized penalty-barrier functional. The convergence result is obtained from local strict convexity and Lipschitz-continuity of this functional in the finite-element space.

7: 

8: The method is very flexible. Each component of the numerical solution can be discretized with a different mesh. General differential-algebraic constraints of arbitrary index can be treated easily with this new method. 

9: 

10: From the discretization results an unconstrained non-linear programming problem (NLP) with penalty- and barrier-terms. The derivatives of the NLP functions have a sparsity pattern that can be analysed and tailored in terms of the chosen finite-element bases in an easy way. We discuss how to treat the resulting NLP in a practical way with general-purpose software for constrained non-linear programming.

11: 

12: \end{abstract}

13: