<div dir="ltr"><div>It sounds like AMG is used on the whole system. I've done this for a different saddle point system but it required enough algorithm extension that I wrote a paper on it. These methods are not in GAMG.<br><br></div><div>I would at least not use smoothed aggregation, use plane aggregation (-pc_gamg_agg_nsmooths 0). It might work if it has a stable MG smoother.<br></div><div><br></div>Mark<br><br>@Article{,
<br> author = {Adams, M.~F.},
<br> title = {Algebraic multrigrid methods for constrained
linear systems with applications to contact problems in solid mechanics},
<br> journal = {Numerical Linear Algebra with Applications},
<br> year = {2004},
<br> volume = {11},
<br> number = {2-3},
<br> pages = {141-153}
<br>}
<br></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Mar 5, 2015 at 5:33 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"><span class="">Fabian Gabel <<a href="mailto:gabel.fabian@gmail.com">gabel.fabian@gmail.com</a>> writes:<br>
<br>
> AMG has apparently been used (successfully?) for fully-coupled<br>
> finite-volume discretizations of Navier-Stokes:<br>
><br>
> <a href="http://dx.doi.org/10.1080/10407790.2014.894448" target="_blank">http://dx.doi.org/10.1080/10407790.2014.894448</a><br>
> <a href="http://dx.doi.org/10.1016/j.jcp.2008.08.027" target="_blank">http://dx.doi.org/10.1016/j.jcp.2008.08.027</a><br>
<br>
</span>These papers effectively bring a pressure projection or SIMPLE-type<br>
algorithm into a coupled system. It's not a saddle-point formulation<br>
and is not uniformly valid across parameters (I think; methods of this<br>
sort normally aren't, but I haven't analyzed it). The multigrid<br>
algorithm is not described and might not be purely algebraic.<br>
<br>
Note that the 2014 paper copies liberally from the 2009 paper.<br>
<br>
You can try formulating Navier-Stokes this way and test existing solvers<br>
if you like, but I predict it doesn't come without tradeoffs.<br>
</blockquote></div><br></div>