/*----< find_nearest_cluster() >---------------------------------------------*/
__inline static
int find_nearest_cluster(int numClusters, /* no. clusters */
int numCoords, /* no. coordinates */
float *object, /* [numCoords] */
float **clusters) /* [numClusters][numCoords] */
{
int index, i, j;
float dist, min_dist;
/* find the cluster id that has min distance to object */
index = 0;
min_dist = euclid_dist_2(numCoords, object, clusters[0]);
dist = euclid_dist_2(numCoords, object, clusters[1]);
if (dist < min_dist) { /* find the min and its array index */
min_dist = dist;
index = 1;
}
dist = euclid_dist_2(numCoords, object, clusters[2]);
if (dist < min_dist) { /* find the min and its array index */
min_dist = dist;
index = 2;
}
return(index);
}
/*----< rms_err(): calculates RMSE of clustering
* >-------------------------------------*/
float rms_err(float **feature, /* [npoints][nfeatures] */
int nfeatures, int npoints,
float **cluster_centres, /* [nclusters][nfeatures] */
int nclusters) {
int i;
int nearest_cluster_index; /* cluster center id with min distance to pt */
float sum_euclid = 0.0; /* sum of Euclidean distance squares */
float ret; /* return value */
/* calculate and sum the sqaure of euclidean distance*/
#pragma omp parallel for shared(feature, cluster_centres) firstprivate( \
npoints, nfeatures, \
nclusters) private(i, nearest_cluster_index) schedule(static)
for (i = 0; i < npoints; i++) {
nearest_cluster_index = find_nearest_point(feature[i], nfeatures,
cluster_centres, nclusters);
sum_euclid += euclid_dist_2(
feature[i], cluster_centres[nearest_cluster_index], nfeatures);
}
/* divide by n, then take sqrt */
ret = sqrt(sum_euclid / npoints);
return (ret);
}
void *find_nearest_point(void *arg)
{
int i, j, l, start, end, len;
long offset;
offset=(long)arg;
len=data1.cluster_len;
start = offset*len;
if(offset==(NUM_THREADS - 1))
end = start + len + data1.rem;
else
end = start + len;
/* find the cluster center id with min distance to a point */
for (i=start; i<end; i++) {
float max_dist=FLT_MAX;
for(j=0; j<data1.nclusters; j++){
float dist;
dist = euclid_dist_2(data1.feature[i], data1.clusters[j], data1.nfeatures); /* no need square root */
if (dist < max_dist) {
max_dist = dist;
data1.membership[i] = j;
data1.new_centers_len[j]++; //the number of points contained by cluster with index 'j' is increased by '1'
for (l=0; l<data1.nfeatures; l++)
data1.new_centers[j][l] += data1.feature[i][l];
}
}
}
pthread_exit(NULL);
}
/*
* Function: find_nearest_cluster
* ------------------------------
* Function determining the cluster center which is closest to the given object.
* Returns the index of that cluster center.
*/
__inline static int find_nearest_cluster(int numClusters, int numCoords, double *object, double **clusters) {
int index, i;
double dist, min_dist;
/* find the cluster id that has min distance to object */
index = 0;
min_dist = euclid_dist_2(numCoords, object, clusters[0]);
for (i=1; i<numClusters; i++) {
dist = euclid_dist_2(numCoords, object, clusters[i]);
/* no need square root */
if (dist < min_dist) { /* find the min and its array index */
min_dist = dist;
index = i;
}
}
return index;
}
/*----< find_nearest_cluster() >---------------------------------------------*/
__inline static
int find_nearest_cluster(int numClusters,
int numCoords,
float *object, /* [numCoords] */
float **clusters) /* [numClusters][numCoords] */
{
int index, i;
float dist, min_dist;
index = 0;
min_dist = euclid_dist_2(numCoords, object, clusters[0]);
for (i=1; i<numClusters; i++) {
dist = euclid_dist_2(numCoords, object, clusters[i]);
/* no need square root */
if (dist < min_dist) { /* find the min and its array index */
min_dist = dist;
index = i;
}
}
return(index);
}
// ===========================================================================
// ===========================================================================
static inline int find_nearest_cluster(int num_clusters, float *object,
float *clusters) {
int index;
int i;
float dist;
float min_dist;
// Find the cluster id that has min distance to object
index = 0;
min_dist = euclid_dist_2(object, clusters);
for (i = 1; i < num_clusters; i++) {
dist = euclid_dist_2(object, clusters + i * COORDS);
// No need square root
if (dist < min_dist) { // Find the min and its array index
min_dist = dist;
index = i;
}
}
return index;
}
/*----< find_nearest_point() >-----------------------------------------------*/
__inline int find_nearest_point(float *pt, /* [nfeatures] */
int nfeatures,
float **pts, /* [npts][nfeatures] */
int npts) {
int index, i;
float max_dist = FLT_MAX;
/* find the cluster center id with min distance to pt */
for (i = 0; i < npts; i++) {
float dist;
dist = euclid_dist_2(pt, pts[i], nfeatures); /* no need square root */
if (dist < max_dist) {
max_dist = dist;
index = i;
}
}
return (index);
}
void *calc_distance(void *arg)
{
int i, j, start, end, len;
float dist;
long offset;
offset=(long)arg;
len=data1.cluster_len;
start = offset*len;
if(offset==(NUM_THREADS - 1))
end = start + len + data1.rem;
else
end = start + len;
for (i=start; i<end; i++){
dist = euclid_dist_2(data1.feature[i], data1.clusters[data1.membership[i]], data1.nfeatures); // no need square root
pthread_mutex_lock(&mut1);
data1.total_sum+=dist;
pthread_mutex_unlock(&mut1);
}
pthread_exit(NULL);
}
/*----< find_nearest_cluster() >---------------------------------------------*/
__inline static
int find_nearest_cluster(int numClusters, /* no. clusters */
int numCoords, /* no. coordinates */
float *object, /* [numCoords] */
float **clusters) /* [numClusters][numCoords] */
{
int index, i;
float dist, min_dist;
/* find the cluster id that has min distance to object */
index = 0;
min_dist = euclid_dist_2(numCoords, object, clusters[0]);
//for (i=1; i<numClusters; i++) {
// dist = euclid_dist_2(numCoords, object, clusters[i]);
// /* no need square root */
// if (dist < min_dist) { /* find the min and its array index */
// min_dist = dist;
// index = i;
// }
//}
return(index);
}
int mpi_kmeans(float **objects, /* in: [numObjs][numCoords] */
int numCoords,
int numObjs,
int numClusters,
float threshold, /* % objects change membership */
int *membership, /* out: [numObjs] membership of points with parent cluster */
float **clusters, /* out: [numClusters][numCoords] */
MPI_Comm comm) /* MPI communicator */
{
int i, j, rank, index, loop=0, total_numObjs , done=1;
int *newClusterSize; /* [numClusters]: no. objects assigned in each
new cluster */
int *clusterSize; /* [numClusters]: temp buffer for Allreduce */
float no_of_changes; /* % of objects change their clusters */
float no_of_changes_tmp;
float **newClusters; /* [numClusters][numCoords] */
extern int _debug;
float **temp1;
float result = 0.0;
int k=0;
if (_debug) MPI_Comm_rank(comm, &rank);
/* initialize membership[] */
for (i=0; i<numObjs; i++) membership[i] = -1;
/* need to initialize newClusterSize and newClusters[0] to all 0 */
newClusterSize = (int*) calloc(numClusters, sizeof(int));
assert(newClusterSize != NULL);
clusterSize = (int*) calloc(numClusters, sizeof(int));
assert(clusterSize != NULL);
newClusters = (float**) malloc(numClusters * sizeof(float*));
assert(newClusters != NULL);
newClusters[0] = (float*) calloc(numClusters * numCoords, sizeof(float));
assert(newClusters[0] != NULL);
for (i=1; i<numClusters; i++)
newClusters[i] = newClusters[i-1] + numCoords;
MPI_Allreduce(&numObjs, &total_numObjs, 1, MPI_INT, MPI_SUM, comm);
if (_debug) printf("%2d: numObjs=%d total_numObjs=%d numClusters=%d numCoords=%d\n",rank,numObjs,total_numObjs,numClusters,numCoords);
do {
done = 1;
double curT = MPI_Wtime();
no_of_changes = 0.0;
for (i=0; i<numObjs; i++) {
/* find the array index of nearest cluster center */
index = find_nearest_cluster(numClusters, numCoords, objects[i],
clusters);
/*$$$$$$ if membership changes, increase no_of_changes by 1 */
if (membership[i] != index) no_of_changes += 1.0;
/* assign the membership to object i */
membership[i] = index;
/* update new cluster centers : sum of objects located within */
newClusterSize[index]++;
for (j=0; j<numCoords; j++)
newClusters[index][j] += objects[i][j];
}
temp1 = (float**) malloc(numClusters * sizeof(float*));
for(i=0;i<5;i++)
{
temp1[i] = (float*) calloc(numClusters * numCoords, sizeof(float));
temp1[i][0]=clusters[i][0];
temp1[i][1]=clusters[i][1];
}
/* sum all data objects in newClusters */
MPI_Allreduce(newClusters[0], clusters[0], numClusters*numCoords,
MPI_FLOAT, MPI_SUM, comm);
MPI_Allreduce(newClusterSize, clusterSize, numClusters, MPI_INT,
MPI_SUM, comm);
/* average the sum and replace old cluster centers with newClusters */
for (i=0; i<numClusters; i++) {
for (j=0; j<numCoords; j++) {
if (clusterSize[i] > 1)
clusters[i][j] /= clusterSize[i];
newClusters[i][j] = 0.0; /* set back to 0 */
}
newClusterSize[i] = 0; /* set back to 0 */
}
for( k=0;k<5 && done == 1;k++)
{
result = euclid_dist_2(2,temp1[k],clusters[k]);
if(result > 0.04)
done = 0;
}
MPI_Allreduce(&no_of_changes, &no_of_changes_tmp, 1, MPI_FLOAT, MPI_SUM, comm);
no_of_changes = no_of_changes_tmp / total_numObjs;
if (_debug) {
double maxTime;
curT = MPI_Wtime() - curT;
MPI_Reduce(&curT, &maxTime, 1, MPI_DOUBLE, MPI_MAX, 0, comm);
if (rank == 0) printf("%2d: loop=%d time=%f sec\n",rank,loop,curT);
}
free(temp1);
}while((done == 0) && (loop++ < 10000));
//printf("\nresult is %f and done is %d\n",result,loop);
if (_debug && rank == 0) printf("%2d: no_of_changes=%f threshold=%f loop=%d\n",rank,no_of_changes,threshold,loop);
free(newClusters[0]);
free(newClusters);
free(newClusterSize);
free(clusterSize);
return 1;
}