van Emde Boas Trees

1. Predecessor search/ordered sets

predecessor: return the nearest left neighbor
successor: return the nearest right neighbor

2.Naive: O(∣U∣)O(|U|)O(∣U∣) space

predecessor/successor: if x∈Sx\in Sx∈S then it’s O(1)O(1)O(1);
insert: worst case: O∣U∣O|U|O∣U∣; compare with all elements; lucky case smaller than the min or bigger than the max.

3. Twolevel: O(∣U∣)O(|U|)O(∣U∣) space

predecessor/successor: worst case: loop over the cluster of one summary, and the length of the cluster is half of the length of the original number, which is w/2w/2w/2. And since for each position in the binary there could be 1/01/01/0, then there would be 2[w/2]2^{[w/2]}2[w/2] different elements in one cluster. Thus, the time: O(∣U∣)O(\sqrt{|U|})O(∣U∣).
insert: loop over the summary and loop over one of the whole cluster.

4. Recursive: O(∣U∣)O(|U|)O(∣U∣) space

Stop: when the length of the elements is 111;
Theorem: the recursion depth of the universe U=[2w]U=[2^w]U=[2w] is [log⁡2w]=O(loglog∣U∣)[\log_2w]=O(loglog|U|)[log2w]=O(loglog∣U∣);
Proof (intuitive): each recursion in binary we half the length of elements, and the recursion would stop when the length is 111, then we can know there are [log⁡2w][\log_2w][log2w] recursions in O(log⁡log⁡∣U∣)O(\log\log |U|)O(loglog∣U∣)

Time in operations

member: go through the recursion depth to check;
predecessor/successor: go through the recursion depth;
insert/delete: operation on both summary and cluster, then it would be double the depth of the recursion depth.

5. vEB: worst case O(log⁡log⁡∣U∣)O(\log\log|U|)O(loglog∣U∣) time, O(∣U∣)O(|U|)O(∣U∣) space

How to improve the time:

Idea: Exclude min(SSS) and/or max(SSS) from the set of keys stored in summary and clusters.

line 10 to line 13: enter the next depth

6. RS-vEB: expected O(loglog|U|) time, O(nloglog|U|)

Use a hash table to store clusters.
Time: Since the depth of the veb tree is O(log⁡log⁡∣U∣)O(\log\log|U|)O(loglog∣U∣), and in hash table the time for updates is O(1)O(1)O(1);
Space: each time we insert a new element, the space cost would be O(log⁡log⁡∣U∣)O(\log\log|U|)O(loglog∣U∣) in worst case, thus the total space would be the number of elements times the depth.

7. R2SR^2SR2S-vEB: expected O(loglog|U|) time, O(n) space

van Emde Boas Trees(vEB树)(Introduction to Algorithms, 算法导论，CLRS)学习笔记相关推荐

【adoo】Van Emde Boas trees
各种姿势补充中--转自http://www.roading.org/algorithm/introductiontoalgorithm/Van_Emde_Boas_trees.html van Emd ...
Van Emde Boas Trees
## 介绍 ## van Emde Boas trees 支持所有优先级优先级队列的操作,并且巧妙的是它对于SEARCH, INSERT,DELETE,MINIMUM,MAXMUN,SUCCESSOR ...
Hashing哈希函数(Introduction to Algorithms, 算法导论，CLRS)学习笔记
Hashing Hashing fundamentals Application: Unordered sets Hash table with chaining Practical universa ...
Linear Programming线性规划(Introduction to Algorithms, 算法导论，CLRS)学习笔记
Linear Programming 1. Fundamentals objective function and constraints: min/max3x1+24x2+13x3+9x4...s. ...
《算法导论3rd第二十章》van Emde Boas树
前言前面介绍的二叉堆,红黑树以及斐波那契堆,其重要的操作都要O(lgn).当特定条件下,能否够规避Ω(lglgn)下界的限制?在本章中,我们将看到:van Emde Boas树支持优先队列操作及一些 ...
Van emde boas树Java实现
简介 VEB树的基础理论在算法导论第三版第20章有详细介绍,这里不再赘述. VEB树的优势 VEB树支持插入.删除.搜索.最大值.最小值.前驱.后继等操作. 这些操作的时间复杂度都是O(lglgu), ...
算法导论读书笔记（20）van Emde Boas树
第五部分高级数据结构第20章 van Emde Boas树 van Emde Boas树支持优先队列操作以及一些其他操作,每个操作最坏情况运行时间为O(lglgn).而这种数据结构限制关键字必须为 ...
van Emde Boas 树数据结构说解
van Emde Boas 树的定义直观上看,vEB 树保存了一个有序的集合,并支持以 O(lglgn) 的时间复杂度在 vEB 树上进行最小最大值查询.单值存在性查询.单值前驱后继查询.单值插入维 ...
算法导论-van Emde Boas树
van Emde Boas树 van Emde Boas树中文名不知道,所以暂且叫它v树吧.v树是一种数据结构,和二叉树.红黑树类似.一种数据结构被创建出来,肯定有其特别的优点,v树的优点就是实现数据 ...

van Emde Boas Trees(vEB树)(Introduction to Algorithms, 算法导论，CLRS)学习笔记

van Emde Boas Trees

1. Predecessor search/ordered sets

2.Naive: O(∣U∣)O(|U|)O(∣U∣) space

3. Twolevel: O(∣U∣)O(|U|)O(∣U∣) space

4. Recursive: O(∣U∣)O(|U|)O(∣U∣) space

Time in operations

5. vEB: worst case O(log⁡log⁡∣U∣)O(\log\log|U|)O(loglog∣U∣) time, O(∣U∣)O(|U|)O(∣U∣) space

6. RS-vEB: expected O(loglog|U|) time, O(nloglog|U|)

7. R2SR^2SR2S-vEB: expected O(loglog|U|) time, O(n) space

van Emde Boas Trees(vEB树)(Introduction to Algorithms, 算法导论，CLRS)学习笔记相关推荐

最新文章

热门文章