1: \begin{abstract}

2:   Approximations of the Dirac delta distribution are commonly used to

3:   create sequences of smooth functions approximating nonsmooth

4:   (generalized) functions, via convolution. In this work we show

5:   a-priori rates of convergence of this approximation process in

6:   standard Sobolev norms, with minimal regularity assumptions on the

7:   approximation of the Dirac delta distribution. The application of

8:   these estimates to the numerical solution of elliptic problems with

9:   singularly supported forcing terms allows us to provide sharp $H^1$

10:   and $L^2$ error estimates for the corresponding regularized

11:   problem. As an application, we show how finite element

12:   approximations of a regularized immersed interface method result in

13:   the same rates of convergence of its non-regularized counterpart,

14:   provided that the support of the Dirac delta approximation is set to

15:   a multiple of the mesh size, at a fraction of the implementation

16:   complexity. Numerical experiments are provided to support our

17:   theories.

18: \end{abstract}

19: