#include <math.h>
#include <omp.h>
#include <stdio.h>
#include <stdlib.h>

#include "common.h"

int main(int argc, char **argv) {
  char *data_file_name = argv[1], *label_file_name = argv[2];
  int outer_iterations = atoi(argv[3]);
  int thread_count = atoi(argv[4]);

  omp_set_num_threads(8);

  struct data *data = read_data(data_file_name);
  struct labels *label = read_labels(label_file_name);
  printf("Running %d iteration(s) with %d thread(s).\n", outer_iterations,
         thread_count);

  FLOAT *w = calloc(data->dimensions, sizeof(FLOAT));

  for (int iter = 0; iter < outer_iterations; iter++) {
    double start_time = monotonic_seconds();
    FLOAT *new_w = calloc(data->dimensions, sizeof(FLOAT));

#pragma omp parallel for default(shared)
    for (int i = 0; i < data->dimensions; i++) {
      FLOAT *ouais = calloc(data->rows, sizeof(FLOAT));

#pragma omp parallel for default(shared)
      for (int j = 0; j < data->rows; j++) {
        FLOAT x_ni_w_ni = 0;

#pragma omp parallel for default(shared) reduction(+ : x_ni_w_ni)
        for (int i2 = 0; i2 < data->dimensions; i2++) {
          if (i2 == i)
            continue;

          x_ni_w_ni = data->buf[data->rows * i2 + j] * w[i2];
        }

        ouais[j] = label->buf[j] - x_ni_w_ni;
      }

      FLOAT numer = 0, denom = 0;

#pragma omp parallel for default(shared) reduction(+ : numer, denom)
      for (int j = 0; j < data->rows; j++) {
        FLOAT xij = data->buf[data->dimensions * i + j];
        numer = xij * ouais[j];
        denom = xij * xij;
      }

      free(ouais);
      new_w[i] = numer / denom;
    }

    printf("Done.\n");
    for (int idx = 0; idx < data->dimensions; idx++) {
      printf("%.3f ", new_w[idx]);
    }
    printf("\n");

    free(w);
    w = new_w;

    double end_time = monotonic_seconds();
    print_time(end_time - start_time);
  }

  free(w);

  return 0;
}