High-Performance Computing Implementation of Grey Wolf Optimizer (GWO)

Overview

This project provides different implementations of the Grey Wolf Optimizer (GWO) algorithm, focusing on high-performance computing (HPC) approaches. It includes:

• A serial implementation of GWO.
• Two parallel implementations where the population is split across multiple processors.
• A hybrid parallel implementation where both the population and dimensions are split across processors.
• Bash scripts to execute each implementation using the Portable Batch System (PBS) for job scheduling on an HPC cluster.

File Structure

📂 project_directory
│-- gwo_serial.c                    # Serial implementation of GWO
│-- gwo_parallel_master-slave.c     # Parallel implementation with master-slave communication pattern
│-- gwo_parallel_distributed.c		# Parallel implementation with fully-distributed communication pattern
│-- gwo_parallel_hybrid.c           # Parallel implementation splitting both population and dimensions
│-- gwo_serial.sh             # PBS script for serial implementation
│-- gwo_parallel_master-slave.sh # PBS script for master-slave implementation
│-- gwo_parallel_distributed.sh  # PBS script for fully-distributed implementation
│-- gwo_parallel_hybrid.sh       # PBS script for hybrid implementation
│-- README.md                       # This documentation file

Dependencies

To compile and run the parallel implementations, ensure you have:

• A C compiler (e.g., gcc)
• MPI (Message Passing Interface) installed (mpicc for compilation)
• Access to an HPC cluster with PBS job scheduling

Compilation

Compile the implementations using the following commands:

# Serial version
gcc -o gwo_serial gwo_serial.c -lm

# Parallel versions (MPI required)
mpicc -o gwo_parallel_master-slave gwo_parallel_master-slave.c -lm
mpicc -o gwo_parallel_population2 gwo_parallel_distributed.c -lm
mpicc -o gwo_parallel_hybrid gwo_parallel_hybrid.c -lm

Running the Implementations

Running Locally

For the serial implementation:

./gwo_serial

For the parallel implementations:

mpirun -np <num_processes> ./gwo_parallel_master-slave
mpirun -np <num_processes> ./gwo_parallel_distributed
mpirun -np <num_processes> ./gwo_parallel_hybrid

Running on an HPC Cluster (PBS)

Use the bash script associated with your desired version to submit a job:

qsub gwo_serial.sh
qsub gwo_parallel_master-slave.sh
qsub gwo_parallel_distributed.sh
qsub gwo_parallel_hybrid.sh