This class provides an interface to SCOTCH-PT (parallel version)
More...
#include <SCOTCH.h>
|
static AdjacencyList< std::int32_t > | partition (const MPI_Comm mpi_comm, const int nparts, const AdjacencyList< SCOTCH_Num > &local_graph, const std::vector< std::size_t > &node_weights, std::int32_t num_ghost_nodes, bool ghosting) |
| Compute distributed graph partition. More...
|
|
static std::pair< std::vector< int >, std::vector< int > > | compute_gps (const AdjacencyList< std::int32_t > &graph, std::size_t num_passes=5) |
| Compute reordering (map[old] -> new) using Gibbs-Poole-Stockmeyer (GPS) re-ordering. More...
|
|
static std::pair< std::vector< int >, std::vector< int > > | compute_reordering (const AdjacencyList< std::int32_t > &graph, std::string scotch_strategy="") |
| Compute graph re-ordering. More...
|
|
This class provides an interface to SCOTCH-PT (parallel version)
◆ compute_gps()
std::pair< std::vector< int >, std::vector< int > > dolfinx::graph::SCOTCH::compute_gps |
( |
const AdjacencyList< std::int32_t > & |
graph, |
|
|
std::size_t |
num_passes = 5 |
|
) |
| |
|
static |
Compute reordering (map[old] -> new) using Gibbs-Poole-Stockmeyer (GPS) re-ordering.
- Parameters
-
[in] | graph | Input graph |
[in] | num_passes | Number of passes to use in GPS algorithm |
- Returns
- (map from old to new nodes, map from new to old nodes (inverse map))
◆ compute_reordering()
std::pair< std::vector< int >, std::vector< int > > dolfinx::graph::SCOTCH::compute_reordering |
( |
const AdjacencyList< std::int32_t > & |
graph, |
|
|
std::string |
scotch_strategy = "" |
|
) |
| |
|
static |
Compute graph re-ordering.
- Parameters
-
[in] | graph | Input graph |
[in] | scotch_strategy | (string) SCOTCH parameters |
- Returns
- (map from old to new nodes, map from new to old nodes (inverse map))
◆ partition()
graph::AdjacencyList< std::int32_t > dolfinx::graph::SCOTCH::partition |
( |
const MPI_Comm |
mpi_comm, |
|
|
const int |
nparts, |
|
|
const AdjacencyList< SCOTCH_Num > & |
local_graph, |
|
|
const std::vector< std::size_t > & |
node_weights, |
|
|
std::int32_t |
num_ghost_nodes, |
|
|
bool |
ghosting |
|
) |
| |
|
static |
Compute distributed graph partition.
- Parameters
-
mpi_comm | MPI Communicator |
nparts | Number of partitions to divide graph nodes into |
local_graph | Node connectivity graph |
node_weights | Weight for each node (optional) |
num_ghost_nodes | Number of graph nodes which are owned on other processes |
ghosting | Flag to enable ghosting of the output node distribution |
- Returns
- Destination rank for each input node
The documentation for this class was generated from the following files:
- /build/dolfinx-EGDDvI/dolfinx-2019.2.0~git20200723.696fbc0/cpp/dolfinx/graph/SCOTCH.h
- /build/dolfinx-EGDDvI/dolfinx-2019.2.0~git20200723.696fbc0/cpp/dolfinx/graph/SCOTCH.cpp