From: bangerth Date: Mon, 22 Aug 2011 21:49:46 +0000 (+0000) Subject: Add a note. X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=153753262ae618b919b273883f257c92a7394c92;p=dealii-svn.git Add a note. git-svn-id: https://svn.dealii.org/trunk@24169 0785d39b-7218-0410-832d-ea1e28bc413d --- diff --git a/deal.II/include/deal.II/distributed/tria.h b/deal.II/include/deal.II/distributed/tria.h index 89a319c60c..888afd1d62 100644 --- a/deal.II/include/deal.II/distributed/tria.h +++ b/deal.II/include/deal.II/distributed/tria.h @@ -208,6 +208,29 @@ namespace parallel * check_for_distorted_cells * argument provided by the * base class. + * + * @note While it is possible to pass + * all of the mesh smoothing flags + * listed in the base class to + * objects of this type, it is not + * always possible to honor all of + * these smoothing options if they + * would require knowledge of + * refinement/coarsening flags on + * cells not locally owned by this + * processor. As a consequence, for + * some of these flags, the ultimate + * number of cells of the parallel + * triangulation may depend on the + * number of processors into which it + * is partitioned. On the other hand, + * if no smoothing flags are passed, + * if you always mark the same cells + * of the mesh, you will always get + * the exact same refined mesh + * independent of the number of + * processors into which the + * triangulation is partitioned. */ Triangulation (MPI_Comm mpi_communicator, const typename dealii::Triangulation::MeshSmoothing