<div class="gmail_quote">On Fri, Dec 3, 2010 at 19:43, Vijay S. Mahadevan <span dir="ltr"><<a href="mailto:vijay.m@gmail.com">vijay.m@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div id=":4fi">Are there any added advantages in terms of<br>
performance/memory footprint and such between the two ?</div></blockquote></div><br><div>Generally,</div><div><br></div><div>ML takes less memory and it returns coarse level operators to PETSc so you have a lot more flexibility (you can use all of PETSc's preconditioners as smoothers, you can control each level independently, and you have lots of options to solve the coarse-level problem). ML needs fewer levels and has lower setup costs. BoomerAMG usually produces a more robust hierarchy so it works for some problems that ML does not. It is basically a black box so you only have the flexibility that they specifically provided (rather than everything in PETSc plus whatever you might want to do).</div>
<div><br></div><div>Jed</div>