1: \begin{abstract}

2: The dynamic Laplace operator arises from extending problems of isoperimetry from fixed manifolds to manifolds evolved by general nonlinear dynamics.  Eigenfunctions of this operator are used to identify and track finite-time coherent sets, which physically manifest in fluid flows as jets, vortices, and more complicated structures.

3: Two robust and efficient finite-element discretisation schemes for numerically computing the dynamic Laplacian were proposed in \cite{froylandjunge18}.

4: In this work we consider higher-order versions of these two numerical schemes and analyse them experimentally. 

5: We also prove the numerically computed eigenvalues and eigenvectors converge to the true objects for both schemes under certain assumptions.

6: %We find that the first scheme, which requires the repeated evaluation of a Cauchy-Green-type tensor, second-order elements lead to higher convergence orders than first-order elements. 

7: %The second scheme, which does not require computation of spatial derivatives, and instead uses transfer operator

8: % and  roposed in  for the discretization of the dynamic Laplacian and   While for the first scheme, , this is not true for the second scheme which is based on an explicit approximation of the transfer operator.  Still, convergence can be proven and we give numerical evidence that higher convergence orders can indeed not be expected for this scheme.  

9: We provide an efficient implementation of the higher-order element schemes in an accompanying Julia package.

10: \end{abstract}

11: