K-means-parallel-and-sequential

Problem Statement

Implement a parallel K-means clustering algorithm. The assignment is to be submitted as three versions of the code – one sequential, one with Pthreads, and one using OpenMP. Lab1_suite was provided by the TAs.

The code that adheres to the format provided in the suite is present in Omp.cpp, pthread.cpp and Seq.cpp. Codes Omp_global_centroid_sum, Seq_global_array.cpp and Seq_global_vectors adhere to more generic formats and maybe suitable depending on the data size. The input format can be seen from dataset_10000_4.txt Use Lab1_suite/dataset.py to generate datasets.

Running the code

1. To compile

g++ main_omp.c lab1_io.h lab1_omp.h lab1_io.c Omp.cpp -lm -fopenmp
g++ main_sequential.c lab1_io.h lab1_sequential.h lab1_io.c Seq.cpp -lm
g++ main_pthread.c lab1_io.h lab1_pthread.h lab1_io.c pthread.cpp -lm -pthread -fopenmp

2. To run arg1: K (no of clusters) arg2: no of threads arg3: input filename (data points) arg4: output filename (data points & cluster) arg5: output filename (centroids of each iteration)

./a.out 150 4 test.txt o1.txt o2.txt

3. To visualise the output given by k-means and the dataset points

python3 visualise.py o1.txt