|
|
人生中第一次知道怎么写可持久化Treap……之前一直以为split是两个log的,后来发现split的过程中merge是O(1)的,这才想明白
|
|
.版权声明:本文为博主原创文章,未经博主允许不得转载。
目录(?)[+] 定义数据结构SBT树的性质旋转维护SBT性质Maintain插入删除获取最小值获取最大值前驱后继求第K小数求元素的秩定义数据结构 [cpp] view plain copy print? 01.struct SBT 02.{ 03. int key,left,right,size; 04.} tree[N]; struct SBT { int key,left,right,size; } tree[N];key:存储值,left,right:左右子树,size:保持平衡最终要的数据,表示子树的大小 SBT树的性质 定义一个节点x,同时满足下面两个条件 [cpp] view plain copy print? 01.(a)、x.left.size >= max(x.right.right.size, x.right.left.size) 02.(b)、x.right.size >= max(x.left.left.size, x.left.right.size) (a)、x.left.size >= max(x.right.right.size, x.right.left.size) (b)、x.right.size >= max(x.left.left.size, x.left.right.size)即每棵子树的大小不小于其兄弟子树的大小 结点L 和R 分别是结点T 的左右儿子。子树A;B;C;D 分别是结点L 和R 各自的左右子树。 即:R.size >= max(A.size,B.size) L.size >= max(C.size,D.size) 旋转 左旋伪代码: [cpp] view plain copy print? 01.Left-Rotate (t) 02. 03.1 k ← right[t] 04.2 right[t] ← left[k] 05.3 left[k] ← t 06.4 size[k] ← size[t] 07.5 size[t] ← size[left[t]] + size[right[t]] + 1 08.6 t ← k Left-Rotate (t) 1 k ← right[t] 2 right[t] ← left[k] 3 left[k] ← t 4 size[k] ← size[t] 5 size[t] ← size[left[t]] + size[right[t]] + 1 6 t ← k[cpp] view plain copy print? 01.void left_rot(int &x) 02.{ 03. int y = tree[x].right; 04. tree[x].right = tree[y].left; 05. tree[y].left = x; 06. tree[y].size = tree[x].size;//转上去的节点数量为先前此处节点的size 07. tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; 08. x = y; 09.} void left_rot(int &x) { int y = tree[x].right; tree[x].right = tree[y].left; tree[y].left = x; tree[y].size = tree[x].size;//转上去的节点数量为先前此处节点的size tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; x = y; } 右旋 [cpp] view plain copy print? 01.Right-Rotate(t) 02. 03.1 k ← left[t] 04.2 left[t] ← right[k] 05.3 right[k] ← t 06.4 size[k] ← size[t] 07.5 size[t] ← size[left[t]] + size[right[t]] + 1 08.6 t ← k Right-Rotate(t) 1 k ← left[t] 2 left[t] ← right[k] 3 right[k] ← t 4 size[k] ← size[t] 5 size[t] ← size[left[t]] + size[right[t]] + 1 6 t ← k[cpp] view plain copy print? 01.void right_rot(int &x) 02.{ 03. int y = tree[x].left; 04. tree[x].left = tree[y].right; 05. tree[y].right = x; 06. tree[y].size = tree[x].size; 07. tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; 08. x = y; 09.} void right_rot(int &x) { int y = tree[x].left; tree[x].left = tree[y].right; tree[y].right = x; tree[y].size = tree[x].size; tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; x = y; }旋转只需要理解如何维护子树size即可,旋转操作不能改变二叉查找树的基本性质 维护SBT性质(Maintain) 当我们插入或删除一个结点后,SBT的大小就发生了改变。这种改变有可能导致性质(a)或(b)被破坏。这时,我们需要用Maintain操作来修复这棵树。Maintain操作是SBT中最具活力的一个独特过程;Maintain(T)用于修复以T为根的 SBT。调用Maintain(T)的前提条件是T的子树都已经是SBT了。 我们需要讨论的有4种情况。由于性质a和性质b是对称的,所以我们仅仅详细的讨论性质a。 第一种情况:x.left.left.size > x.right.size 即:insert(T.left,key)后A.size > R.size 1、首先执行Right-Ratote(t),这个操作使上图变成下图; 2、在这之后,有时候这棵树还仍然不是一棵SBT,因为 s[C]>s[B] 或者 s[D]>s[B] 也是可能发生的。所以就有必要继续调用Maintain(T)。 3、结点L的右子树有可能被连续调整,因为有可能由于性质的破坏需要再一次运行Maintain(L)。 第二种情况:x.left.right.size > x.right.size 在执行完insert(T.left,key)后发生B.size > R.size,如图5,这种调整要比情况1复杂一些。我们可以执行下面的操作来修复: 1、执行一次左旋操作Left-Ratote(L)后,就会变成下图 2、接着执行一次右旋操作Right-Ratote(T),变成下图: 3、在经过两个巨大的旋转之后,整棵树就变得非常不可预料了。万幸的是,子树A;E; F;R 依然是容均树,所以要依次修复L 和T,Maintain(L),Maintain(T)。 4、子树都已经是容均树了,但B 可能还有问题,所以还要调用Maintain(B) 第三种情况:x.right.right.size > x.left.size 与第一种情况相反 第四种情况:x.right.left.size > x.left.size 与第二种情况相反 伪代码 [cpp] view plain copy print? 01.Maintain (t,flag) 02. 03.01 If flag=false then 04.02 If s[left[left[t]]>s[right[t]] then //case1 05.03 Right-Rotate(t) 06.04 Else 07.05 If s[right[left[t]]>s[right[t]] then //case2 08.06 Left-Rotate(left[t]) 09.07 Right-Rotate(t) 10.08 Else //needn’t repair 11.09 Exit 12.10 Else 13.11 If s[right[right[t]]>s[left[t]] then //case1' 14.12 Left-Rotate(t) 15.13 Else 16.14 If s[left[right[t]]>s[left[t]] then //case2' 17.15 Right-Rotate(right[t]) 18.16 Left-Rotate(t) 19.17 Else //needn’t repair 20.18 Exit 21.19 Maintain(left[t],false) //repair the left subtree 22.20 Maintain(right[t],true) //repair the right subtree 23.21 Maintain(t,false) //repair the whole tree 24.22 Maintain(t,true) //repair the whole tree Maintain (t,flag) 01 If flag=false then 02 If s[left[left[t]]>s[right[t]] then //case1 03 Right-Rotate(t) 04 Else 05 If s[right[left[t]]>s[right[t]] then //case2 06 Left-Rotate(left[t]) 07 Right-Rotate(t) 08 Else //needn’t repair 09 Exit 10 Else 11 If s[right[right[t]]>s[left[t]] then //case1' 12 Left-Rotate(t) 13 Else 14 If s[left[right[t]]>s[left[t]] then //case2' 15 Right-Rotate(right[t]) 16 Left-Rotate(t) 17 Else //needn’t repair 18 Exit 19 Maintain(left[t],false) //repair the left subtree 20 Maintain(right[t],true) //repair the right subtree 21 Maintain(t,false) //repair the whole tree 22 Maintain(t,true) //repair the whole tree[cpp] view plain copy print? 01.void maintain(int &x,bool flag) 02.{ 03. if(flag == false)//左边 04. { 05. if(tree[tree[tree[x].left].left].size > tree[tree[x].right].size)//左孩子的左子树大于右孩子 06. right_rot(x); 07. else if(tree[tree[tree[x].left].right].size > tree[tree[x].right].size)//右孩子的右子树大于右孩子 08. { 09. left_rot(tree[x].left); 10. right_rot(x); 11. } 12. else return; 13. } 14. else //右边 15. { 16. if(tree[tree[tree[x].right].right].size > tree[tree[x].left].size)//右孩子的右子树大于左孩子 17. left_rot(x); 18. else if(tree[tree[tree[x].right].left].size > tree[tree[x].left].size)//右孩子的左子树大于左孩子 19. { 20. right_rot(tree[x].right); 21. left_rot(x); 22. } 23. else return; 24. } 25. maintain(tree[x].left,false); 26. maintain(tree[x].right,true); 27. maintain(x,true); 28. maintain(x,false); 29.} void maintain(int &x,bool flag) { if(flag == false)//左边 { if(tree[tree[tree[x].left].left].size > tree[tree[x].right].size)//左孩子的左子树大于右孩子 right_rot(x); else if(tree[tree[tree[x].left].right].size > tree[tree[x].right].size)//右孩子的右子树大于右孩子 { left_rot(tree[x].left); right_rot(x); } else return; } else //右边 { if(tree[tree[tree[x].right].right].size > tree[tree[x].left].size)//右孩子的右子树大于左孩子 left_rot(x); else if(tree[tree[tree[x].right].left].size > tree[tree[x].left].size)//右孩子的左子树大于左孩子 { right_rot(tree[x].right); left_rot(x); } else return; } maintain(tree[x].left,false); maintain(tree[x].right,true); maintain(x,true); maintain(x,false); } 插入 在BST插入的基础上添加一个Maintain操作,用来对size的维护 [cpp] view plain copy print? 01.void insert(int &x,int key) 02.{ 03. if(x == 0) 04. { 05. x = ++top; 06. tree[x].left = tree[x].right = 0; 07. tree[x].size = 1; 08. tree[x].key = key; 09. } 10. else 11. { 12. tree[x].size ++; 13. if(key < tree[x].key) insert(tree[x].left,key); 14. else insert(tree[x].right,key);//相同元素插入到右子树中 15. maintain(x, key >= tree[x].key);//每次插入把平衡操作压入栈中 16. } 17.} void insert(int &x,int key) { if(x == 0) { x = ++top; tree[x].left = tree[x].right = 0; tree[x].size = 1; tree[x].key = key; } else { tree[x].size ++; if(key < tree[x].key) insert(tree[x].left,key); else insert(tree[x].right,key);//相同元素插入到右子树中 maintain(x, key >= tree[x].key);//每次插入把平衡操作压入栈中 } } 删除 与普通维护size域的BST删除相同。 关于无需Maintain的说明by sqybi: 在删除之前,可以保证整棵树是一棵SBT。当删除之后,虽然不能保证这棵树还是SBT,但是这时整棵树的最大深度并没有改变,所以时间复杂度也不会增加。这时,Maintain就显得是多余的了。 下面给出两种删除方式,一种是找前驱替换,一种是找后继替换 //后继 [cpp] view plain copy print? 01.int remove(int &x,int key) 02.{ 03. tree[x].size --; 04. if(key > tree[x].key) 05. remove(tree[x].right,key); 06. else if(key < tree[x].key) 07. remove(tree[x].left,key); 08. else 09. { 10. //有左子树,无右子树 11. if(tree[x].left != 0 && tree[x].right == 0) 12. { 13. int temp = x; 14. x = tree[x].left; 15. return temp; 16. } 17. else if(tree[x].right !=0 && tree[x].left == 0) 18. { 19. int temp = x; 20. x = tree[x].right; 21. return temp; 22. } 23. //无左子树和右子树 24. else if(!tree[x].left && !tree[x].right) 25. { 26. int temp = x; 27. x = 0; 28. return temp; 29. } 30. //有右子树 31. else //找到x右子树中最小元素,也就是找后继元素 32. { 33. int temp = tree[x].right; 34. while(tree[temp].left) temp = tree[temp].left; 35. tree[x].key = tree[temp].key; 36. //tree[x].cnt = tree[temp].cnt; 37. remove(tree[x].right,tree[temp].key); 38. } 39. } 40.} int remove(int &x,int key) { tree[x].size --; if(key > tree[x].key) remove(tree[x].right,key); else if(key < tree[x].key) remove(tree[x].left,key); else { //有左子树,无右子树 if(tree[x].left != 0 && tree[x].right == 0) { int temp = x; x = tree[x].left; return temp; } else if(tree[x].right !=0 && tree[x].left == 0) { int temp = x; x = tree[x].right; return temp; } //无左子树和右子树 else if(!tree[x].left && !tree[x].right) { int temp = x; x = 0; return temp; } //有右子树 else //找到x右子树中最小元素,也就是找后继元素 { int temp = tree[x].right; while(tree[temp].left) temp = tree[temp].left; tree[x].key = tree[temp].key; //tree[x].cnt = tree[temp].cnt; remove(tree[x].right,tree[temp].key); } } }//前驱 [cpp] view plain copy print? 01.int remove(int &x,int key) 02.{ 03. int d_key; 04. //if(!x) return 0; 05. tree[x].size --; 06. if((key == tree[x].key)||(key < tree[x].key && tree[x].left == 0) || 07. (key>tree[x].key && tree[x].right == 0)) 08. { 09. d_key = tree[x].key; 10. if(tree[x].left && tree[x].right) 11. { 12. tree[x].key = remove(tree[x].left,tree[x].key+1); 13. } 14. else 15. { 16. x = tree[x].left + tree[x].right; 17. } 18. } 19. else if(key > tree[x].key) 20. d_key = remove(tree[x].right,key); 21. else if(key < tree[x].key) 22. d_key = remove(tree[x].left,key); 23. return d_key; 24.} int remove(int &x,int key) { int d_key; //if(!x) return 0; tree[x].size --; if((key == tree[x].key)||(key < tree[x].key && tree[x].left == 0) || (key>tree[x].key && tree[x].right == 0)) { d_key = tree[x].key; if(tree[x].left && tree[x].right) { tree[x].key = remove(tree[x].left,tree[x].key+1); } else { x = tree[x].left + tree[x].right; } } else if(key > tree[x].key) d_key = remove(tree[x].right,key); else if(key < tree[x].key) d_key = remove(tree[x].left,key); return d_key; } 获取最小值 直接向左子树找到最左叶子节点即可 [cpp] view plain copy print? 01.int getmin() 02.{ 03. int x; 04. for(x = root ; tree[x].left; x = tree[x].left); 05. return tree[x].key; 06.} int getmin() { int x; for(x = root ; tree[x].left; x = tree[x].left); return tree[x].key; } 获取最大值 [cpp] view plain copy print? 01.int getmax() 02.{ 03. int x; 04. for(x = root ; tree[x].right; x = tree[x].right); 05. return tree[x].key; 06.} int getmax() { int x; for(x = root ; tree[x].right; x = tree[x].right); return tree[x].key; } 前驱 [cpp] view plain copy print? 01.int pred(int &x,int y,int key)//前驱 小于 02.{ 03. if(x == 0) return y; 04. if(tree[x].key < key)//加上等号,就是小于等于 05. return pred(tree[x].right,x,key); 06. else return pred(tree[x].left,y,key); 07.}//pred(root,0,key) int pred(int &x,int y,int key)//前驱 小于 { if(x == 0) return y; if(tree[x].key < key)//加上等号,就是小于等于 return pred(tree[x].right,x,key); else return pred(tree[x].left,y,key); }//pred(root,0,key) 注解:if(tree[x].key < key) 则说明前驱(小于key中所有元素最大的那个)在右子树,并且设定当前节点x为其前驱 if(tree[x].key >= key)说明前驱在左子树,当前节点x的key值也不是其前驱,所以设定其前驱仍为y 后继 [cpp] view plain copy print? 01.int succ(int &x,int y,int key)//后继 大于 02.{ 03. if(x == 0) return y; 04. if(tree[x].key > key) 05. return succ(tree[x].left,x,key); 06. else return succ(tree[x].right,y,key); 07.} int succ(int &x,int y,int key)//后继 大于 { if(x == 0) return y; if(tree[x].key > key) return succ(tree[x].left,x,key); else return succ(tree[x].right,y,key); }同前驱。 求第K小数 [cpp] view plain copy print? 01.int select(int &x,int k)//求第k小数 02.{ 03. int r = tree[tree[x].left].size + 1; 04. if(r == k) return tree[x].key; 05. else if(r < k) return select(tree[x].right,k - r); 06. else return select(tree[x].left,k); 07.} int select(int &x,int k)//求第k小数 { int r = tree[tree[x].left].size + 1; if(r == k) return tree[x].key; else if(r < k) return select(tree[x].right,k - r); else return select(tree[x].left,k); }注解:首先该源码如果插入的数有重复数据(即树中会出现两个或多个27,这样的数据),此SBT是可以建树的,同样在查询第K小数上述代码也不会产生错误,但是这里需要消耗更多的存储空间(这里应该很容易明白),如果我们在数据结构中加上一个字段cnt,专门用来记录重复数据的个数,这样的话树中就没有重复数据,因为它们已经被合并了,这里需要修改insert函数和remove函数和旋转操作,如果删除操作每次删除的是整个节点而不是cnt>2就仅仅将cnt--而是整个删除,这样就会对size造成很大的影响 ,这种情况的remove函数我暂时没有想好如何去写,首先可以确定思路,如果删除节点是x,它的直接或间接父亲节点的size都需要减去x.cnt,但是我们是用的替换删除,这里怎么操作?还请哪位大牛指教,能够写出这样的remove函数。 还是先解释上面的代码,首先求出x的左子树size,再加上自己本身,如果这时r == k 则说明x.key就是第K小数,如果r < k,就说明第K小数在以x为根的右子树上,此时只需要在右子树上求第k-r小数即可,如果r > k那么说明第K小数在以x为根的左子树上。 求元素的秩 就是求这个元素排第几 [cpp] view plain copy print? 01.int rank(int &x,int key)//求key排第几 02.{ 03. if(key < tree[x].key) 04. return rank(tree[x].left,key); 05. else if(key > tree[x].key) 06. return rank(tree[x].right,key) + tree[tree[x].left].size + 1; 07. return tree[tree[x].left].size + 1; 08.} int rank(int &x,int key)//求key排第几 { if(key < tree[x].key) return rank(tree[x].left,key); else if(key > tree[x].key) return rank(tree[x].right,key) + tree[tree[x].left].size + 1; return tree[tree[x].left].size + 1; }这里就类似select的逆操作,代码就不解释了。 下面给出整个C++源码 [cpp] view plain copy print? 01.#include <iostream> 02.#include <cstdio> 03.#include <cstdlib> 04.#include <cstring> 05.#include <cmath> 06.#include <algorithm> 07.#include <set> 08.#include <map> 09.#include <vector> 10.#include <queue> 11.#include <ctime> 12.using namespace std; 13.#define LL long long 14.const int N = 10005; 15.const int INF=0x7FFFFFFF; 16. 17.struct SBT 18.{ 19. int key,left,right,size; 20.} tree[N]; 21. 22.int root,top; 23. 24.void left_rot(int &x) 25.{ 26. int y = tree[x].right; 27. tree[x].right = tree[y].left; 28. tree[y].left = x; 29. tree[y].size = tree[x].size;//转上去的节点数量为先前此处节点的size 30. tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; 31. x = y; 32.} 33. 34.void right_rot(int &x) 35.{ 36. int y = tree[x].left; 37. tree[x].left = tree[y].right; 38. tree[y].right = x; 39. tree[y].size = tree[x].size; 40. tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; 41. x = y; 42.} 43. 44.void maintain(int &x,bool flag) 45.{ 46. if(flag == false)//左边 47. { 48. if(tree[tree[tree[x].left].left].size > tree[tree[x].right].size)//左孩子的左子树大于右孩子 49. right_rot(x); 50. else if(tree[tree[tree[x].left].right].size > tree[tree[x].right].size)//右孩子的右子树大于右孩子 51. { 52. left_rot(tree[x].left); 53. right_rot(x); 54. } 55. else return; 56. } 57. else //右边 58. { 59. if(tree[tree[tree[x].right].right].size > tree[tree[x].left].size)//右孩子的右子树大于左孩子 60. left_rot(x); 61. else if(tree[tree[tree[x].right].left].size > tree[tree[x].left].size)//右孩子的左子树大于左孩子 62. { 63. right_rot(tree[x].right); 64. left_rot(x); 65. } 66. else return; 67. } 68. maintain(tree[x].left,false); 69. maintain(tree[x].right,true); 70. maintain(x,true); 71. maintain(x,false); 72.} 73. 74./* 75.*insert没有合并相同的元素,如果出现相同的元素则把它放到右子树上,这样能保证求第k小数的时候对相同元素也能正确 76.*/ 77.void insert(int &x,int key) 78.{ 79. if(x == 0) 80. { 81. x = ++top; 82. tree[x].left = tree[x].right = 0; 83. tree[x].size = 1; 84. tree[x].key = key; 85. } 86. else 87. { 88. tree[x].size ++; 89. if(key < tree[x].key) insert(tree[x].left,key); 90. else insert(tree[x].right,key);//相同元素插入到右子树中 91. maintain(x, key >= tree[x].key);//每次插入把平衡操作压入栈中 92. } 93.} 94. 95.int del(int &p,int w) 96.{ 97. if (tree[p].key==w || (tree[p].left==0 && w<tree[p].key) || (tree[p].right==0 && w>tree[p].key)) 98. { 99. int delnum=tree[p].key; 100. if (tree[p].left==0 || tree[p].right==0) p=tree[p].left+tree[p].right; 101. else tree[p].key=del(tree[p].left,INF); 102. return delnum; 103. } 104. if (w<tree[p].key) return del(tree[p].left,w); 105. else return del(tree[p].right,w); 106.} 107. 108.int remove(int &x,int key) 109.{ 110. int d_key; 111. //if(!x) return 0; 112. tree[x].size --; 113. if((key == tree[x].key)||(key < tree[x].key && tree[x].left == 0) || 114. (key>tree[x].key && tree[x].right == 0)) 115. { 116. d_key = tree[x].key; 117. if(tree[x].left && tree[x].right) 118. { 119. tree[x].key = remove(tree[x].left,tree[x].key+1); 120. } 121. else 122. { 123. x = tree[x].left + tree[x].right; 124. } 125. } 126. else if(key > tree[x].key) 127. d_key = remove(tree[x].right,key); 128. else if(key < tree[x].key) 129. d_key = remove(tree[x].left,key); 130. return d_key; 131.} 132. 133.int getmin() 134.{ 135. int x; 136. for(x = root ; tree[x].left; x = tree[x].left); 137. return tree[x].key; 138.} 139. 140.int getmax() 141.{ 142. int x; 143. for(x = root ; tree[x].right; x = tree[x].right); 144. return tree[x].key; 145.} 146. 147.int select(int &x,int k)//求第k小数 148.{ 149. int r = tree[tree[x].left].size + 1; 150. if(r == k) return tree[x].key; 151. else if(r < k) return select(tree[x].right,k - r); 152. else return select(tree[x].left,k); 153.} 154. 155.int rank(int &x,int key)//求key排第几 156.{ 157. if(key < tree[x].key) 158. return rank(tree[x].left,key); 159. else if(key > tree[x].key) 160. return rank(tree[x].right,key) + tree[tree[x].left].size + 1; 161. return tree[tree[x].left].size + 1; 162.} 163. 164.int pred(int &x,int y,int key)//前驱 小于 165.{ 166. if(x == 0) return y; 167. if(tree[x].key < key) 168. return pred(tree[x].right,x,key); 169. else return pred(tree[x].left,y,key); 170.} 171. 172.int succ(int &x,int y,int key)//后继 大于 173.{ 174. if(x == 0) return y; 175. if(tree[x].key > key) 176. return succ(tree[x].left,x,key); 177. else return succ(tree[x].right,y,key); 178.} 179. 180.void inorder(int &x) 181.{ 182. if(x==0) return; 183. else 184. { 185. inorder(tree[x].left); 186. cout<<x<<" "<<tree[x].key<<" "<<" "<<tree[x].size<<" "<<tree[tree[x].left].key<<" "<<tree[tree[x].right].key<<endl; 187. inorder(tree[x].right); 188. } 189.} 190. 191.int main() 192.{ 193. root = top = 0; 194. char ch; 195. int x,tmp; 196. while(scanf("%c %d",&ch,&x)) 197. { 198. switch(ch) 199. { 200. case 'I': 201. insert(root,x); 202. break; 203. case 'D': 204. remove(root,x); 205. break; 206. case 'K': 207. tmp=select(root,x); 208. printf("%d\n",tmp); 209. break; 210. case 'C': 211. printf("%d\n",rank(root,x)); 212. break; 213. case 'P': 214. tmp = pred(root,0,x); 215. printf("%d\n",tree[tmp].key); 216. break; 217. case 'S': 218. tmp = succ(root,0,x); 219. printf("%d\n",tree[tmp].key); 220. break; 221. case 'L': 222. inorder(root); 223. break; 224. } 225. 226. } 227. return 0; 228.} #include <iostream> #include <cstdio> #include <cstdlib> #include <cstring> #include <cmath> #include <algorithm> #include <set> #include <map> #include <vector> #include <queue> #include <ctime> using namespace std; #define LL long long const int N = 10005; const int INF=0x7FFFFFFF; struct SBT { int key,left,right,size; } tree[N]; int root,top; void left_rot(int &x) { int y = tree[x].right; tree[x].right = tree[y].left; tree[y].left = x; tree[y].size = tree[x].size;//转上去的节点数量为先前此处节点的size tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; x = y; } void right_rot(int &x) { int y = tree[x].left; tree[x].left = tree[y].right; tree[y].right = x; tree[y].size = tree[x].size; tree[x].size = tree[tree[x].left].size + tree[tree[x].right].size + 1; x = y; } void maintain(int &x,bool flag) { if(flag == false)//左边 { if(tree[tree[tree[x].left].left].size > tree[tree[x].right].size)//左孩子的左子树大于右孩子 right_rot(x); else if(tree[tree[tree[x].left].right].size > tree[tree[x].right].size)//右孩子的右子树大于右孩子 { left_rot(tree[x].left); right_rot(x); } else return; } else //右边 { if(tree[tree[tree[x].right].right].size > tree[tree[x].left].size)//右孩子的右子树大于左孩子 left_rot(x); else if(tree[tree[tree[x].right].left].size > tree[tree[x].left].size)//右孩子的左子树大于左孩子 { right_rot(tree[x].right); left_rot(x); } else return; } maintain(tree[x].left,false); maintain(tree[x].right,true); maintain(x,true); maintain(x,false); } /* *insert没有合并相同的元素,如果出现相同的元素则把它放到右子树上,这样能保证求第k小数的时候对相同元素也能正确 */ void insert(int &x,int key) { if(x == 0) { x = ++top; tree[x].left = tree[x].right = 0; tree[x].size = 1; tree[x].key = key; } else { tree[x].size ++; if(key < tree[x].key) insert(tree[x].left,key); else insert(tree[x].right,key);//相同元素插入到右子树中 maintain(x, key >= tree[x].key);//每次插入把平衡操作压入栈中 } } int del(int &p,int w) { if (tree[p].key==w || (tree[p].left==0 && w<tree[p].key) || (tree[p].right==0 && w>tree[p].key)) { int delnum=tree[p].key; if (tree[p].left==0 || tree[p].right==0) p=tree[p].left+tree[p].right; else tree[p].key=del(tree[p].left,INF); return delnum; } if (w<tree[p].key) return del(tree[p].left,w); else return del(tree[p].right,w); } int remove(int &x,int key) { int d_key; //if(!x) return 0; tree[x].size --; if((key == tree[x].key)||(key < tree[x].key && tree[x].left == 0) || (key>tree[x].key && tree[x].right == 0)) { d_key = tree[x].key; if(tree[x].left && tree[x].right) { tree[x].key = remove(tree[x].left,tree[x].key+1); } else { x = tree[x].left + tree[x].right; } } else if(key > tree[x].key) d_key = remove(tree[x].right,key); else if(key < tree[x].key) d_key = remove(tree[x].left,key); return d_key; } int getmin() { int x; for(x = root ; tree[x].left; x = tree[x].left); return tree[x].key; } int getmax() { int x; for(x = root ; tree[x].right; x = tree[x].right); return tree[x].key; } int select(int &x,int k)//求第k小数 { int r = tree[tree[x].left].size + 1; if(r == k) return tree[x].key; else if(r < k) return select(tree[x].right,k - r); else return select(tree[x].left,k); } int rank(int &x,int key)//求key排第几 { if(key < tree[x].key) return rank(tree[x].left,key); else if(key > tree[x].key) return rank(tree[x].right,key) + tree[tree[x].left].size + 1; return tree[tree[x].left].size + 1; } int pred(int &x,int y,int key)//前驱 小于 { if(x == 0) return y; if(tree[x].key < key) return pred(tree[x].right,x,key); else return pred(tree[x].left,y,key); } int succ(int &x,int y,int key)//后继 大于 { if(x == 0) return y; if(tree[x].key > key) return succ(tree[x].left,x,key); else return succ(tree[x].right,y,key); } void inorder(int &x) { if(x==0) return; else { inorder(tree[x].left); cout<<x<<" "<<tree[x].key<<" "<<" "<<tree[x].size<<" "<<tree[tree[x].left].key<<" "<<tree[tree[x].right].key<<endl; inorder(tree[x].right); } } int main() { root = top = 0; char ch; int x,tmp; while(scanf("%c %d",&ch,&x)) { switch(ch) { case 'I': insert(root,x); break; case 'D': remove(root,x); break; case 'K': tmp=select(root,x); printf("%d\n",tmp); break; case 'C': printf("%d\n",rank(root,x)); break; case 'P': tmp = pred(root,0,x); printf("%d\n",tree[tmp].key); break; case 'S': tmp = succ(root,0,x); printf("%d\n",tree[tmp].key); break; case 'L': inorder(root); break; } } return 0; } 总结:SBT,Treap,红黑树,都很类似,其中SBT和Treap代码很好写,SBT只需要正确理解旋转操作,和如何维护size域,另外就是Maintain函数,这个是重点,其他的操作都和BST一样了,SBT的高度是O(logn),Maintain是O(1),所有主要操作都是O(logn)。 最后还请大神解决下重复数据的remove操作函数如何写(具体见求第K小数下的注解),欢迎指教。
题目 2314 [HZOI 2015] Persistable Editor
2016-05-16 19:20:33
|