1: \begin{abstract}

2:   The paper considers standard iterative methods for solving the

3:   generalized Stokes problem arising from the time and space

4:   approximation of the time-dependent incompressible Navier-Stokes

5:   equations. Various preconditioning techniques are considered

6:   (Cahouet\&Chabard and augmented Lagrangian), and one investigates

7:   whether these methods can compete with traditional

8:   pressure-correction and velocity-correction methods in terms of CPU

9:   time per degree of freedom and per time step.  Numerical tests on

10:   fine unstructured meshes (68 millions degrees of freedoms)

11:   demonstrate convergence rates that are independent of the mesh size

12:   and improve with the Reynolds number.  Three conclusions are drawn

13:   from the paper: (1) Although very good parallel scalability is

14:   observed for the augmented Lagrangian method, thorough tests on

15:   large problems reveal that the overall CPU time per degree of

16:   freedom and per time step is best for the standard Cahouet\&Chabard

17:   preconditioner. (2) Whether solving the pressure Schur complement

18:   problem or solving the full couple system at once does not make any

19:   significant difference in term of CPU time per degree of freedom and

20:   per time step.  (3) All the methods tested in the paper, whether

21:   matrix-free or not, are on average 30 times slower than traditional

22:   pressure-correction and velocity-correction methods. Hence, although

23:   all these methods are very efficient for solving steady state

24:   problems, they are not yet competitive for solving time-dependent

25:   problems.

26: \end{abstract}

27: