1: \begin{abstract}

2: In this paper we study Lipschitz regularity of elliptic PDEs on

3: geometric graphs, constructed from random data points. The data points

4: are sampled from a distribution supported on a smooth manifold. The family of equations

5: that we study arises in data analysis in the context of graph-based

6: learning and contains, as important examples, the equations satisfied

7: by graph Laplacian eigenvectors. In particular, we prove high

8: probability interior and global Lipschitz estimates for solutions of

9: graph Poisson equations. Our results can be used to show that graph

10: Laplacian eigenvectors are, with high probability, essentially

11: Lipschitz regular with constants depending explicitly on their

12: corresponding eigenvalues. Our analysis relies on a probabilistic

13: coupling argument of suitable random walks at the continuum level, and

14: an interpolation method for extending functions on random point

15: clouds to the continuum manifold. As a byproduct of our general

16: regularity results, we obtain high probability $L^\infty$ and

17: approximate $\mathcal{C}^{0,1}$ convergence rates for the convergence of graph

18: Laplacian eigenvectors towards eigenfunctions of the corresponding

19: weighted Laplace-Beltrami operators. The convergence rates we obtain

20: scale like the $L^2$-convergence rates established in \cite{calder2019improved}. 

21: \end{abstract}

22: