0.1)本文总结于 数据结构与算法分析, 但源代码均为原创,旨在实现 不相交集ADT的两个操作:合并集合union+查找集合find;
0.2) 不相交集ADT 的 Introduction , 参见 http://blog.csdn.net/PacosonSWJTU/article/details/497169051.1)大小求并法定义:上面的Union执行是相当任意的, 通过使第二棵树 成为第一棵树的子树而完成合并;对其的改进是借助任意的方法打破现有关系, 使得总让较小的树成为较大树的子树,我们把这种方法叫做 大小求并法 ;
1.2)可以看到,如果Union 操作都是按照大小求并的话,那么任何节点的深度均不会超过 logN;
1.3) 首先注意节点的深度为0, 然后它的深度随着一次 Union 的结果而增加的时候,该节点则被置于至少是 它以前所在树两倍大的一棵树上;因此,它的深度最多可以增加 logN次;1.3.2)Find 操作 的运行时间为 O(logN), 而连续M次操作则花费 O(MlogN);
1.3.3)下图指出在16次Union操作后可能得到这种最坏的树;而且如果所有的Union都对相等大小的树进行, 那么这样的树是会得到的;
1.4)为了实现这种方法, 我们需要记住每个树的大小。由于我们实际上只使用一个数组,因此可以让每个根的数组元素包含它 的树的大小的负值;
1.5)已经证明: 若使用按大小求并则连续 M次运算需要 O(M)平均时间, 这是因为 当随机的Union执行时, 整个算法一般只有一些很小的集合(通常是一个元素)与 大 集合 合并;1.6)souce code + printing
Source Code Statements:
S1)显然,本源代码只对元素的size进行了加,没有减;因为我想的话, 元素只能合并一次,也即是元素C起初合并到了集合A,就不能再次合并到集合B, 元素C就一直属于集合A的子集了;
S2)当然,这个代码只是 大致上实现了按大小求并的思想,如果元素C还可以再次合并到其他集合的话,这就涉及到集合根元素的size的加减问题了;需要的话,添加之即可;
1.6.2)souce code at a glance
#include <stdio.h> #include <malloc.h> #define ElementType int #define Error(str) printf("/n error: %s /n",str) struct UnionSet; typedef struct UnionSet* UnionSet; // we adopt the child-sibling expr struct UnionSet { int parent; int size; ElementType value; }; UnionSet makeEmpty(); UnionSet* initUnionSet(int size, ElementType* data); void printSet(UnionSet* set, int size); void printArray(ElementType data[], int size); int find(ElementType index, UnionSet* set); // initialize the union set UnionSet* initUnionSet(int size, ElementType* data) { UnionSet* set; int i; set = (UnionSet*)malloc(size * sizeof(UnionSet)); if(!set) { Error("out of space, from func initUnionSet"); return NULL; } for(i=0; i<size; i++) { set[i] = makeEmpty(); if(!set[i]) return NULL; set[i]->value = data[i]; } return set; } // allocate the memory for the single UnionSet and evaluate the parent and size -1 UnionSet makeEmpty() { UnionSet temp; temp = (UnionSet)malloc(sizeof(struct UnionSet)); if(!temp) { Error("out of space, from func makeEmpty!"); return NULL; } temp->parent = -1; temp->size = 1; return temp; } // merge set1 and set2 by size void setUnion(UnionSet* set, int index1, int index2) { //judge whether the index1 or index2 equals to -1 ,also -1 represents the root if(index1 != -1) index1 = find(index1, set); if(index2 != -1) index2 = find(index2, set); if(set[index1]->size > set[index2]->size) { set[index2]->parent = index1; set[index1]->size += set[index2]->size; } else { set[index1]->parent = index2; set[index2]->size += set[index1]->size; } } //find the root of one set whose value equals to given value int find(ElementType index, UnionSet* set) { UnionSet temp; while(1) { temp = set[index]; if(temp->parent == -1) break; index = temp->parent; } return index; } int main() { int size; UnionSet* unionSet; ElementType data[] = {110, 245, 895, 658, 321, 852, 147, 458, 469, 159, 347, 28}; size = 12; printf("/n/t====== test for union set by size ======/n"); //printf("/n/t=== the initial array is as follows ===/n"); //printArray(data, size); printf("/n/t=== the init union set are as follows ===/n"); unionSet = initUnionSet(size, data); // initialize the union set over printSet(unionSet, size); printf("/n/t=== after union(1,5) + union(2,5) + union(3,4) + union(4,5) ===/n"); setUnion(unionSet, 1, 5); setUnion(unionSet, 2, 5); setUnion(unionSet, 3, 4); setUnion(unionSet, 4, 5); printSet(unionSet, size); printf("/n/t=== after union(9,8) + union(7,6) + union(3,6) ===/n"); setUnion(unionSet, 9, 8); setUnion(unionSet, 7, 6); setUnion(unionSet, 3, 6); printSet(unionSet, size); return 0; } void printArray(ElementType data[], int size) { int i; for(i = 0; i < size; i++) printf("/n/t data[%d] = %d", i, data[i]); printf("/n/n"); } void printSet(UnionSet* set, int size) { int i; UnionSet temp; for(i = 0; i < size; i++) { temp = set[i]; printf("/n/t parent[%d] = %d", i, temp->parent); } printf("/n"); }
1.6.3)printing result
2.1)按高度求并定义:另外一种方法是按照高度求并(按照高度求并是按大小求并的简单修改) (推荐) ;
2.2)它同样保证所有的树的深入最多是 O(logN)。我们使得 浅的树 成为深 的树的子树,这是一种平缓算法, 因为只有当两颗相等深度的树求并时树的高度才会增加(此时树的高度增加1)。
2.3)source code + printing result2.3.1)download source code :
https://github.com/pacosonTang/dataStructure-algorithmAnalysis/blob/master/chapter8/p205_unionByHeight.c2.3.2)source code at a glance:
#include <stdio.h> #include <malloc.h> #define ElementType int #define Error(str) printf("/n error: %s /n",str) struct UnionSet; typedef struct UnionSet* UnionSet; // we adopt the depth-sibling expr struct UnionSet { int parent; int height; ElementType value; }; UnionSet makeEmpty(); UnionSet* initUnionSet(int depth, ElementType* data); void printSet(UnionSet* set, int depth); void printArray(ElementType data[], int depth); int find(ElementType index, UnionSet* set); // initialize the union set UnionSet* initUnionSet(int size, ElementType* data) { UnionSet* set; int i; set = (UnionSet*)malloc(size * sizeof(UnionSet)); if(!set) { Error("out of space, from func initUnionSet"); return NULL; } for(i=0; i<size; i++) { set[i] = makeEmpty(); if(!set[i]) return NULL; set[i]->value = data[i]; } return set; } // allocate the memory for the single UnionSet and evaluate the parent and depth -1 UnionSet makeEmpty() { UnionSet temp; temp = (UnionSet)malloc(sizeof(struct UnionSet)); if(!temp) { Error("out of space, from func makeEmpty!"); return NULL; } temp->parent = -1; temp->height = 0; return temp; } // merge set1 and set2 by depth void setUnion(UnionSet* set, int index1, int index2) { //judge whether the index1 or index2 equals to -1 ,also -1 represents the root if(index1 != -1) index1 = find(index1, set); if(index2 != -1) index2 = find(index2, set); if(set[index1]->height > set[index2]->height) set[index2]->parent = index1; else if(set[index1]->height < set[index2]->height) set[index1]->parent = index2; else { set[index1]->parent = index2; set[index2]->height++; // only if the height of both of subtrees is equal, the height increases one } } //find the root of one set whose value equals to given value int find(ElementType index, UnionSet* set) { UnionSet temp; while(1) { temp = set[index]; if(temp->parent == -1) break; index = temp->parent; } return index; } int main() { int size; UnionSet* unionSet; ElementType data[] = {110, 245, 895, 658, 321, 852, 147, 458, 469, 159, 347, 28}; size = 12; printf("/n/t====== test for union set by depth ======/n"); //printf("/n/t=== the initial array is as follows ===/n"); //printArray(data, depth); printf("/n/t=== the init union set are as follows ===/n"); unionSet = initUnionSet(size, data); // initialize the union set over printSet(unionSet, size); printf("/n/t=== after union(0, 1) + union(2, 3) + union(4, 5) + union(6, 7) + union(8, 9) + union(10 ,11) ===/n"); setUnion(unionSet, 0, 1); setUnion(unionSet, 2, 3); setUnion(unionSet, 4, 5); setUnion(unionSet, 6, 7); setUnion(unionSet, 8, 9); setUnion(unionSet, 10, 11); printSet(unionSet, size); printf("/n/t=== after union(1, 3) + union(5, 7) + union(9, 11) ===/n"); setUnion(unionSet, 1, 3); setUnion(unionSet, 5, 7); setUnion(unionSet, 9, 11); printSet(unionSet, size); printf("/n/t=== after union(3, 7) + union(7, 11) ===/n"); setUnion(unionSet, 3, 7); setUnion(unionSet, 7, 11); printSet(unionSet, size); return 0; } void printArray(ElementType data[], int size) { int i; for(i = 0; i < size; i++) printf("/n/t data[%d] = %d", i, data[i]); printf("/n/n"); } void printSet(UnionSet* set, int size) { int i; UnionSet temp; for(i = 0; i < size; i++) { temp = set[i]; printf("/n/t parent[%d] = %d", i, temp->parent); } printf("/n"); }
2.3.3)printing results: