From: Marc Fehling Date: Thu, 26 Nov 2020 02:55:51 +0000 (-0700) Subject: Update documentation to avoid deprecated functions ... X-Git-Tag: v9.3.0-rc1~849^2 X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=refs%2Fpull%2F11253%2Fhead;p=dealii.git Update documentation to avoid deprecated functions ... ... DoFHandler::locally_owned_dofs_per_processor(), DoFHandler::n_locally_owned_dofs_per_processor(), DoFHandler::locally_owned_mg_dofs_per_processor(). --- diff --git a/include/deal.II/dofs/dof_handler.h b/include/deal.II/dofs/dof_handler.h index a535398f20..363b3f7eaf 100644 --- a/include/deal.II/dofs/dof_handler.h +++ b/include/deal.II/dofs/dof_handler.h @@ -1198,11 +1198,12 @@ public: * @deprecated As of deal.II version 9.2, we do not populate a vector with * the index sets of all processors by default any more due to a possibly * large memory footprint on many processors. As a consequence, this - * function needs to call `Utilities::all_gather(comm, locally_owned_dofs())` - * upon the first invocation, including global communication. Use - * `Utilities::all_gather(comm, dof_handler.locally_owned_dofs())` instead if - * using up to a few thousands of MPI ranks or some variant involving local - * communication with more processors. + * function needs to call `Utilities::MPI::all_gather(comm, + * locally_owned_dofs())` upon the first invocation, including global + * communication. Use `Utilities::MPI::all_gather(comm, + * dof_handler.locally_owned_dofs())` instead if using up to a few thousands + * of MPI ranks or some variant involving local communication with more + * processors. */ DEAL_II_DEPRECATED const std::vector & locally_owned_dofs_per_processor() const; @@ -1216,9 +1217,9 @@ public: * @deprecated As of deal.II version 9.2, we do not populate a vector with * the numbers of dofs of all processors by default any more due to a * possibly large memory footprint on many processors. As a consequence, - * this function needs to call `Utilities::all_gather(comm, + * this function needs to call `Utilities::MPI::all_gather(comm, * n_locally_owned_dofs()` upon the first invocation, including global - * communication. Use `Utilities::all_gather(comm, + * communication. Use `Utilities::MPI::all_gather(comm, * dof_handler.n_locally_owned_dofs()` instead if using up to a few thousands * of MPI ranks or some variant involving local communication with more * processors. @@ -1233,9 +1234,9 @@ public: * @deprecated As of deal.II version 9.2, we do not populate a vector with * the index sets of all processors by default any more due to a possibly * large memory footprint on many processors. As a consequence, this - * function needs to call `Utilities::all_gather(comm, + * function needs to call `Utilities::MPI::all_gather(comm, * locally_owned_dofs_mg())` upon the first invocation, including global - * communication. Use `Utilities::all_gather(comm, + * communication. Use `Utilities::MPI::all_gather(comm, * dof_handler.locally_owned_dofs_mg())` instead if using up to a few * thousands of MPI ranks or some variant involving local communication with * more processors. diff --git a/include/deal.II/lac/affine_constraints.h b/include/deal.II/lac/affine_constraints.h index 03169d4b56..daf01e9eeb 100644 --- a/include/deal.II/lac/affine_constraints.h +++ b/include/deal.II/lac/affine_constraints.h @@ -1663,11 +1663,12 @@ public: * operation and will return @p true only if all processors are consistent. * * Please supply the owned DoFs per processor as returned by - * DoFHandler::locally_owned_dofs_per_processor() as @p locally_owned_dofs - * and the result of DoFTools::extract_locally_active_dofs() as - * @p locally_active_dofs. The - * former is used to determine ownership of the specific DoF, while the latter - * is used as the set of rows that need to be checked. + * Utilities::MPI::all_gather(MPI_Comm, DoFHandler::locally_owned_dofs()) as + * @p locally_owned_dofs and the result of + * DoFTools::extract_locally_active_dofs() as + * @p locally_active_dofs. The former is used to determine ownership of the + * specific DoF, while the latter is used as the set of rows that need to be + * checked. * * If @p verbose is set to @p true, additional debug information is written * to std::cout. diff --git a/include/deal.II/lac/sparsity_tools.h b/include/deal.II/lac/sparsity_tools.h index 162ff1fd83..dced4cd969 100644 --- a/include/deal.II/lac/sparsity_tools.h +++ b/include/deal.II/lac/sparsity_tools.h @@ -277,10 +277,11 @@ namespace SparsityTools * Communicate rows in a dynamic sparsity pattern over MPI, similar to the * one above but using a vector `rows_per_cpu` containing the number of * rows per CPU for determining ownership. This is typically the value - * returned by DoFHandler::n_locally_owned_dofs_per_processor -- given that - * the construction of the input to this function involves all-to-all - * communication, it is typically slower than the function above for more - * than a thousand of processes (and quick enough also for small sizes). + * returned by Utilities::MPI::all_gather(MPI_Comm, + * DoFHandler::locally_owned_dofs()) -- given that the construction of the + * input to this function involves all-to-all communication, it is typically + * slower than the function above for more than a thousand of processes (and + * quick enough also for small sizes). */ void distribute_sparsity_pattern(