1: \begin{abstract}

2: This paper presents estimates of the convergence rate and complexity of an 

3: algebraic multilevel preconditioner based on piecewise constant coarse vector spaces 

4: applied to the graph Laplacian. 

5: A bound is derived on the energy norm of the projection operator 

6: onto any piecewise constant vector space, 

7: which results in

8: an estimate of the two-level convergence rate 

9: where the coarse level graph is obtained by matching.  

10: %A notable aspect of this 

11: %convergence estimate is that it depends only on the maximum degree of the graphs in 

12: %the multilevel hierarchy

13: %and, hence, it directly accounts for the complexity of the algorithm.

14: The two-level convergence of the method is then used to establish 

15: the convergence of an Algebraic Multilevel Iteration that uses the two-level scheme recursively.

16: On structured grids, the method is proven to have 

17: convergence rate $\approx (1-1/\log n)$ 

18: and $O(n\log n)$ complexity for each cycle, where $n$ denotes the number of unknowns

19: in the given problem. 

20: Numerical results of the algorithm applied to various 

21: graph Laplacians are reported.

22: It is also shown that all the theoretical estimates derived for matching can be generalized to the case of

23:  aggregates containing more than two vertices.  

24: \end{abstract}

25: