1: \begin{abstract}

2: Solving the normal equations corresponding to large sparse linear

3: least-squares problems is an important and challenging

4: problem. For very large problems, an iterative

5: solver is needed and, in general, a preconditioner is required

6: to achieve good convergence.

7: In recent years, a number of preconditioners have been proposed.

8: These are largely serial and reported results demonstrate that

9: none of the commonly used preconditioners for the normal equations

10: matrix is capable of solving all sparse least-squares problems.

11: Our interest is thus in 

12: designing new preconditioners for the normal equations that are efficient, robust,

13: and can be implemented in parallel. Our proposed preconditioners can be constructed

14: efficiently and algebraically without any knowledge of the problem and without

15: any assumption on the least-squares matrix except that it is sparse.

16: We exploit the structure of the symmetric positive definite normal equations matrix

17: and use the concept of algebraic local symmetric positive semi-definite splittings

18: to introduce  two-level Schwarz preconditioners for least-squares problems.

19: The condition number of the preconditioned normal equations is shown to be theoretically bounded

20: independently of the number of subdomains in the splitting. This upper bound can be adjusted using

21: a single parameter $\tau$ that the user can specify.

22: We discuss how the new preconditioners can be implemented on top of the PETSc library

23: using only 150 lines of Fortran, C, or Python code.

24: Problems arising from practical applications are used to compare the performance of the 

25: proposed new preconditioner with that of other preconditioners.

26: \end{abstract}

27: