1: \begin{abstract}

2:   We present and analyze a waveform relaxation version of the

3:   Dirichlet-Neumann and Neumann-Neumann methods for parabolic

4:   problems. Like the Dirichlet-Neumann method for steady problems, the

5:   method is based on a non-overlapping spatial domain decomposition,

6:   and the iteration involves subdomain solves with Dirichlet boundary

7:   conditions followed by subdomain solves with Neumann boundary

8:   conditions. For the Neumann-Neumann method, one step of the method

9:   consists of solving the subdomain problems using Dirichlet interface

10:   conditions, followed by a correction step involving Neumann

11:   interface conditions. However, each subdomain problem is now in

12:   space and time, and the interface conditions are also

13:   time-dependent.  Using Laplace transforms, we show for the heat

14:   equation that when we consider finite time intervals, the

15:   Dirichlet-Neumann and Neumann-Neumann methods converge superlinearly

16:   for an optimal choice of the relaxation parameter, similar to the

17:   case of Schwarz waveform relaxation algorithms. The convergence rate

18:   depends on the size of the subdomains as well as the length of the

19:   time window. For any other choice of the relaxation parameter,

20:   convergence is only linear. We illustrate our results with numerical

21:   experiments.

22: \end{abstract}

23: