Merge pull request #13 from Sitaras/debugCluster2

Debug/Test cluster

Author: giannhskp

Clustering/clustering.c

@@ -244,7 +244,7 @@ void clusteringLSH(List vecList,int numOfClusters,int l,FILE* fptr){
  }
  props = calloc(numOfVecs,sizeof(double));
 
- hashTableSize=numOfVecs/16;
+ hashTableSize=numOfVecs/32;
  LSH lsh = initializeLSH(l);
  for(int i=0;i<numOfVecs;i++){
  initializeClusterInfo(vectors[i]);
@@ -366,7 +366,7 @@ void reverseAssignmentHypercube(HyperCube cube,Vector *vectors,Vector *clusters,
  List confList=initializeList();
  for(int i=0;i<numOfClusters;i++){
  radiusNeigborHypercubeClustering(cube,clusters[i],clustersHt[i],radius,probes,m,i,&confList,&assignCounter,iteration);
- printf("******* ---- ASSINGED ITEMS = %d\n",assignCounter);
+ printf("******* ---- ASSINGED ITEMS = %d | Cluster %d\n",assignCounter,i+1);
  }
  listSolveRangeConflicts(confList,clustersHt,clusters,numOfClusters,d,iteration);
  printf("---- ASSINGED ITEMS = %d\n",assignCounter);
@@ -467,6 +467,7 @@ void clusteringHypercube(List vecList,int numOfClusters,int m,int probes,FILE* f
 
  fprintf(fptr, "clustering_time: %f seconds\n",cluster_time);
  fflush(fptr);
+ fflush(stdout);
 
  printf("- COMPUTING SILHOUETTES FOR CLUSTERS\n");
  double stotal = 0.0;

cluster.conf

@@ -1,6 +1,6 @@
 number_of_clusters: 4
-number_of_vector_hash_tables: 6
+number_of_vector_hash_tables: 5
 number_of_vector_hash_functions: 6
-max_number_M_hypercube: 100
-number_of_hypercube_dimensions: 5
-number_of_probes: 15
+max_number_M_hypercube: 500
+number_of_hypercube_dimensions: 4
+number_of_probes: 16

hashTable/hashTable.c

@@ -86,7 +86,7 @@ void htPrint(const HashTable ht){
 void htPrintClustering(const HashTable ht,FILE* fptr){
  for (int i=0;i<ht->buckets;i++){
  listPrintClusteringInFile(ht->table[i].head,fptr);
- if(i!=ht->buckets-1)
+ if(i!=ht->buckets-1 && (ht->table[i].head!=NULL))
  fprintf(fptr,", ");
  }
  printf("\n\n");

hashTable/hashTableList/hashTableList.c

@@ -121,7 +121,7 @@ void listPrint(List list){
 }
 
 void listPrintClusteringInFile(List list,FILE* fptr){
- if(list==NULL){ printf("List Empty!\n"); return;}
+ if(list==NULL){ return;}
  List temp=list;
  while(temp!=NULL){
  printVectorIdInFileNoNewline(temp->v,fptr);
@@ -426,7 +426,8 @@ void listFindNeighborsInRadiusClustering(List list,int centroidIndex,List* confL
  if(id==(temp->vector_ID)){
  if(assignedToCluster(temp->v) && (getAssignedIteration(temp->v)==iteration)){
  int assignedCluster = getAssignedCluster(temp->v);
- if(assignedCluster==centroidIndex || (((int)getAssignedAtRadius(temp->v))!=radius)){
+ // if(assignedCluster==centroidIndex || (((int)getAssignedAtRadius(temp->v))!=radius)){
+ if(assignedCluster==centroidIndex || (((int)getAssignedAtRadius(temp->v))<-1)){
  temp=temp->next;
  continue;
  }else{
@@ -470,7 +471,8 @@ void listFindNeighborsInRadiusClusteringCube(List list,int centroidIndex,List* c
  // then
  // check if vector has already been assigned at the some cluster (check the centroid index)
  // or if vector has already been assigned previously in cluster at a search with different radius
- if(assignedCluster==centroidIndex || (getAssignedAtRadius(temp->v)!=radius)){
+ if(assignedCluster==centroidIndex || (getAssignedAtRadius(temp->v)<-1.0 )){
+ // if(assignedCluster==centroidIndex || (getAssignedAtRadius(temp->v)!=radius )){
  // ok, then skip it
  temp=temp->next;
  continue;
@@ -484,10 +486,10 @@ void listFindNeighborsInRadiusClusteringCube(List list,int centroidIndex,List* c
  }
  }
  else{
- htRangeInsert(storeNeighbors,temp->v,temp->vector_ID,d);
  setAssignedCluster(temp->v,centroidIndex);
  setAssignedIteration(temp->v,iteration);
  setAssignedAtRadius(temp->v,radius);
+ htRangeInsert(storeNeighbors,temp->v,temp->vector_ID,d);
  (*assignCounter)++;
  }
  }