]> https://gitweb.dealii.org/ - dealii.git/commitdiff
step-21:
authorWolfgang Bangerth <bangerth@math.tamu.edu>
Wed, 6 Apr 2011 19:54:38 +0000 (19:54 +0000)
committerWolfgang Bangerth <bangerth@math.tamu.edu>
Wed, 6 Apr 2011 19:54:38 +0000 (19:54 +0000)
If the size of a linear system is small, i.e. when the mesh is very
coarse, then it is sometimes not sufficient to set a maximum of
<code>src.size()</code> CG iterations before the solver in the
<code>vmult()</code> function converges. (This is, of course, a result
of numerical round-off, since we know that on paper, the CG method
converges in at most <code>src.size()</code> steps.) As a consequence,
we set the maximum number of iterations equal to the maximum of the
size of the linear system and 200.

git-svn-id: https://svn.dealii.org/trunk@23563 0785d39b-7218-0410-832d-ea1e28bc413d

deal.II/doc/news/changes.h

index bac4a0b86ec8ff1fed4f19b04e1e46fd37ca3c75..f62a76a807b3bb0262a9ae97df0fe24cf608dd1b 100644 (file)
@@ -95,6 +95,14 @@ should be fixed now.
 <h3>Specific improvements</h3>
 
 <ol>
+<li> Fixed: In step-21, the inner iteration would sometimes not converge for
+very coarse meshes because of numerical roundoff. This is now fixed by allowing
+more than <code>rhs.size()</code> CG iterations if the number of degrees of freedom
+is very small.
+<br>
+(Jichao Yin, WB, 2011/04/06)
+</li>
+
 <li> New: There is now a new function ConditionalOStream::get_stream().
 <br>
 (WB, 2011/03/09)

In the beginning the Universe was created. This has made a lot of people very angry and has been widely regarded as a bad move.

Douglas Adams


Typeset in Trocchi and Trocchi Bold Sans Serif.