05_decomposition_parallel.cpp File Reference

Comprehensive demonstration of OpenPFC Decomposition API. More...

#include <iomanip>
#include <iostream>
#include <mpi.h>
#include <openpfc/core/decomposition.hpp>
#include <openpfc/core/world.hpp>
#include <sstream>

Include dependency graph for 05_decomposition_parallel.cpp:

Functions
std::string	rank_prefix (int rank)
void	print_sync (int rank, int size, const std::string &message)
void	scenario_manual_grid ()
void	scenario_automatic_grid ()
void	scenario_coordinate_mapping ()
void	scenario_properties ()
void	scenario_load_balance ()
void	scenario_aspect_ratios ()
int	main (int argc, char **argv)

Detailed Description

Comprehensive demonstration of OpenPFC Decomposition API.

This example showcases domain decomposition for distributed-memory parallelism:

• Manual grid specification (explicit layout)
• Automatic grid selection (optimal communication)
• Rank-local vs global coordinate mapping
• Subdomain queries and properties
• Integration with MPI parallel execution

The Decomposition class partitions a global World into non-overlapping subdomains, each owned by one MPI rank. This enables scalable parallel simulations on HPC clusters.

Requirements: MPI library (must be run with mpirun/mpiexec)

Usage:

# Compile (assuming build system configured)
cd build && make 05_decomposition_parallel
 
# Run with 4 MPI ranks
mpirun -np 4 ./05_decomposition_parallel