<div dir="ltr"><div>OK, thanks a lot Jed for your quick answer.<br></div><div><br></div><div>I have a second question on multigrid : <br></div><div>I define a matrix block size (equal to 3). Is this mandatory for any type of multigrid preconditioner (BoomerAMG, ML and GAMG) ? <br>
</div><div>Or is it only required with ML and GAMG ?<br><br></div><div>Best regards<br></div><div>Natacha <br></div><div> <br></div> <br></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Jul 18, 2014 at 3:07 PM, Jed Brown <span dir="ltr"><<a href="mailto:jed@jedbrown.org" target="_blank">jed@jedbrown.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="">Natacha BEREUX <<a href="mailto:natacha.bereux@gmail.com">natacha.bereux@gmail.com</a>> writes:<br>
<br>
> Dear all,<br>
><br>
> I am testing three multigrid preconditionners: BoomerAMG, ML and GAMG on a<br>
> 3D elasticity problem (a cube with prescribed displacement on upper and<br>
> lower faces).<br>
> I set the null space by calling MatNullSpaceCreateRigidBody and then<br>
> MatSetNearNullSpace).<br>
><br>
> I see a difference (about half iterations) for ML and GAMG but nothing<br>
> happens with BoomerAMG (same number of iterations with and without setting<br>
> the NearNullspace).<br>
<br>
</div>That's expected because BoomerAMG does not use any such information<br>
(though the algorithm might be extended in the future so that those<br>
vectors become seed vectors for Bootstrap AMG, for example; I don't know<br>
if the BoomerAMG developers have any interest in implementing this).<br>
</blockquote></div><br></div>