#include <stdio.h> // printf
#include <stdlib.h> // EXIT_SUCCESS
#include <math.h> // M_PI, fabs
#include <omp.h>

#include "pi_shared.h"

int main(int argc, char* argv[]) {
    int seed;
    int N;
    int hits;
    double my_pi;
    int number_of_workers;
    int N_per_worker;

    // get parameters from command line arguments
    read_arguments(&seed, &N, argc, argv);

    // create OMP_NUM_THREADS threads
    // every thread will have its own hits variable (reduction implies private keyword)
    // after the parallel section, a single hits variable will remain,
    // holding the sum of all thread private hits variables
#pragma omp parallel reduction(+:hits)
    {
        // divide the workload equally among threads
        number_of_workers = omp_get_num_threads();
        N_per_worker = N / number_of_workers;
        N = N_per_worker * number_of_workers;

        // generate N_per_worker random coordinates, get number of hits
        // (see count_hits() function for details)
        // please note that every thread will start with a different seed value
        hits = count_hits(seed + omp_get_thread_num(), N_per_worker);
    } 

    // estimate PI based on following assumptions
    //   area of circle                = PI * r^2   ~= hits
    //   area of square around circle  = (2 * r)^2  ~= N
    //   r = 1
    my_pi = 4. * ((double) hits  / (double) N);

    // print seed, N, number_of_workers, result, error
    printf("%d\t%d\t%d\t%f\t%f\n", seed, N, number_of_workers, my_pi, fabs(my_pi-M_PI));

    return EXIT_SUCCESS;

}