1: \begin{abstract}

2: 	This paper studies general geometric condition for developing numerical analysis on discretized manifolds, especially when the solutions involve directly the differential structure of the manifolds. In the literature, it is quite standard to ask for exact geometric information to implement a priori or a posteriori error analysis of numerical solutions, which, however, is often violated in practice.  Such gap is filled using the superconvergence of differential structure on manifolds. We illustrate the idea using two concrete exemplary case study. In the first case, we deal with deviated manifolds, where we propose a concept called geometric supercloseness. It is shown that this condition is capable of proving the superconvergence of gradient recovery on deviated discretized manifolds. This closes the open questions proposed by Wei, Chen, and Huang in SIAM J. Numer. Anal., 48(2010), pp. 1920--1943. For this matter,  an algorithmic framework for gradient recovery without asking  exact geometric information are proposed. A second example shows that the superconvergence of differential structure enables us to develop optimal convergent numerical method for solving vector Laplacians on discretized manifold without superparametric element or exact normal fields, which are required by approaches from literature.  Numerical results are documented for verifying the theoretical discoveries.

3: 

4: 	\vskip .3cm

5: 	{\bf AMS subject classifications.} \ {41A25, 65N15, 65N30}

6: 	\vskip .3cm

7: 	

8: 	{\bf Key words.} \ {Discretized manifolds, deviated discretization, differential structure, geometric supercloseness, superconvergence, gradient recovery, vector Laplace-Beltrami problem, tangent vector field.}

9: \end{abstract}

10: