Is laplace operator well-conditioned?

63 views
Skip to first unread message

yy.wayne

unread,
Mar 14, 2023, 2:22:54 AM3/14/23
to deal.II User Group
Hello, 

I‘m kind of lost in the nature of laplace operator.
Poisson equations are easy to solve numerically, and it has
<grad, grad> type weak form.
However, in step-22 it says laplace operator is ill-conditioned
and expensive to invert with iterative method.

This question might be silly. Is laplcae operator a positive definite
system and easy to solve? Does it become ill-conditioned when
mesh size and mesh quality reduce?

Wolfgang Bangerth

unread,
Mar 14, 2023, 12:19:43 PM3/14/23
to dea...@googlegroups.com

> This question might be silly. Is laplcae operator a positive definite
> system and easy to solve? Does it become ill-conditioned when
> mesh size and mesh quality reduce?

"ill-conditioned" and "difficult to solve" are two different things. The
condition number of the discretized Laplace operator (i.e., the Laplace
matrix) grows like O(1/h^2), so it becomes quite large on fine meshes. But
linear systems with this matrix are relatively easy to solve if you use
multigrid methods.

Best
W.

--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@colostate.edu
www: http://www.math.colostate.edu/~bangerth/


yy.wayne

unread,
Mar 14, 2023, 7:59:30 PM3/14/23
to deal.II User Group
I get it now. Thanks for your rely.

Best,
Wayne

Reply all
Reply to author
Forward
0 new messages