mirror of
https://github.com/heqin-zhu/algorithm.git
synced 2024-03-22 13:30:46 +08:00
Finished red black tree, but the script has some bugs and is to be polished
This commit is contained in:
mbinary
parent
3e84fa4a0b
commit
64c62a0ff8
18
README.md
18
README.md
|
|
@ -1,13 +1,19 @@
|
|||
# 算法笔记
|
||||
记录学习算法的一些笔记, 想法, 以及代码实现 :smiley:
|
||||
# Algorithm and data structures
|
||||
>Notes and codes for learning algorithm and data structures :smiley:
|
||||
|
||||
目前正在看`<<算法导论>>`
|
||||
Some pictures and idead are from `<<Introduction to Algotithm>>
|
||||
|
||||
I use python 3.6+ and c++ to implements them.
|
||||
Since I used f-Strings in python, you may use python 3.6+ to run the following python scripts.
|
||||
|
||||
> 目前 github 上的文档不支持 latex 数学公式渲染
|
||||
所以如果要读下面的一些笔记(md文件), 可以移步到[我的博客](https://mbinary.coding.me)
|
||||
>>I am still learning new things and this repo is always updating.
|
||||
Some scripts may have bugs or not be finished yet.
|
||||
|
||||
# 索引
|
||||
# Notice
|
||||
Currently, Github can't render latex math formulas.
|
||||
So,if you wannt to view the notes which contains latex math formulas and are in markdown format, you can visit [my blog](https://mbinary.coding.me)
|
||||
|
||||
# index
|
||||
* [.](.)
|
||||
* [notes](./notes)
|
||||
* [alg-general.md](./notes/alg-general.md)
|
||||
|
|
|
|||
313
dataStructure/redBlackTree.py
Normal file
313
dataStructure/redBlackTree.py
Normal file
|
|
@ -0,0 +1,313 @@
|
|||
'''
|
||||
#########################################################################
|
||||
# File : redBlackTree.py
|
||||
# Author: mbinary
|
||||
# Mail: zhuheqin1@gmail.com
|
||||
# Blog: https://mbinary.coding.me
|
||||
# Github: https://github.com/mbinary
|
||||
# Created Time: 2018-07-12 20:34
|
||||
# Description:
|
||||
#########################################################################
|
||||
'''
|
||||
from functools import total_ordering
|
||||
from random import randint, shuffle
|
||||
|
||||
@total_ordering
|
||||
class node:
|
||||
def __init__(self,val,left=None,right=None,isBlack=False):
|
||||
self.val =val
|
||||
self.left = left
|
||||
self.right = right
|
||||
self.isBlack = isBlack
|
||||
def __lt__(self,nd):
|
||||
return self.val < nd.val
|
||||
def __eq__(self,nd):
|
||||
return nd is not None and self.val==nd.val
|
||||
def setChild(self,nd,isLeft = True):
|
||||
if isLeft: self.left = nd
|
||||
else: self.right = nd
|
||||
def getChild(self,isLeft):
|
||||
if isLeft: return self.left
|
||||
else: return self.right
|
||||
def __bool__(self):
|
||||
return self.val is not None
|
||||
def __str__(self):
|
||||
color = 'B' if self.isBlack else 'R'
|
||||
return f'{color}-{self.val:}'
|
||||
def __repr__(self):
|
||||
return f'node({self.val},isBlack={self.isBlack})'
|
||||
class redBlackTree:
|
||||
def __init__(self):
|
||||
self.root = None
|
||||
def getParent(self,val):
|
||||
if isinstance(val,node):val = val.val
|
||||
if self.root.val == val:return None
|
||||
nd = self.root
|
||||
while nd:
|
||||
if nd.val>val and nd.left is not None:
|
||||
if nd.left.val == val: return nd
|
||||
else: nd = nd.left
|
||||
elif nd.val<val and nd.right is not None:
|
||||
if nd.right.val == val: return nd
|
||||
else: nd = nd.right
|
||||
def find(self,val):
|
||||
if isinstance(val,node):val = val.val
|
||||
nd = self.root
|
||||
while nd:
|
||||
if nd.val ==val:
|
||||
return nd
|
||||
elif nd.val>val:
|
||||
nd = nd.left
|
||||
else:
|
||||
nd = nd.right
|
||||
@staticmethod
|
||||
def checkBlack(nd):
|
||||
return nd is None or nd.isBlack
|
||||
@staticmethod
|
||||
def setBlack(nd,isBlack):
|
||||
if nd is not None:
|
||||
if isBlack is None or isBlack:
|
||||
nd.isBlack = True
|
||||
else:nd.isBlack = False
|
||||
def insert(self,val):
|
||||
if isinstance(val,node):val = val.val
|
||||
def _insert(root,nd):
|
||||
'''return parent'''
|
||||
while root:
|
||||
if root == nd:return None
|
||||
elif root>nd:
|
||||
if root.left :
|
||||
root=root.left
|
||||
else:
|
||||
root.left = nd
|
||||
return root
|
||||
else:
|
||||
if root.right:
|
||||
root = root.right
|
||||
else:
|
||||
root.right = nd
|
||||
return root
|
||||
# insert part
|
||||
nd = node(val)
|
||||
if self.root is None:
|
||||
self.root = nd
|
||||
self.setBlack(self.root,True)
|
||||
return
|
||||
parent = _insert(self.root,nd)
|
||||
if parent is None: return
|
||||
if not parent.isBlack: self.fixUpInsert(parent,nd)
|
||||
|
||||
def fixUpInsert(self,parent,nd):
|
||||
''' adjust color and level, there are two red nodes: the new one and its parent'''
|
||||
while not self.checkBlack(parent):
|
||||
grand = self.getParent(parent)
|
||||
isLeftPrt = grand.left == parent
|
||||
uncle = grand.getChild(not isLeftPrt)
|
||||
if not self.checkBlack(uncle):
|
||||
# case 1: new node's uncle is red
|
||||
self.setBlack(grand, False)
|
||||
self.setBlack(grand.left, True)
|
||||
self.setBlack(grand.right, True)
|
||||
nd = grand
|
||||
parent = self.getParent(nd)
|
||||
else:
|
||||
# case 2: new node's uncle is black(including nil leaf)
|
||||
isLeftNode = parent.left==nd
|
||||
if isLeftNode ^ isLeftPrt:
|
||||
# case 2.1 the new node is inserted in left-right or right-left form
|
||||
# grand grand
|
||||
# parent or parent
|
||||
# nd nd
|
||||
parent.setChild(nd.getChild(isLeftPrt),not isLeftPrt)
|
||||
nd.setChild(parent,isLeftPrt)
|
||||
grand.setChild(nd,isLeftPrt)
|
||||
nd,parent = parent,nd
|
||||
# case 2.2 the new node is inserted in left-left or right-right form
|
||||
# grand grand
|
||||
# parent or parent
|
||||
# nd nd
|
||||
grand.setChild(parent.getChild(not isLeftPrt),isLeftPrt)
|
||||
parent.setChild(grand,not isLeftPrt)
|
||||
self.setBlack(grand, False)
|
||||
self.setBlack(parent, True)
|
||||
self.setBlack(self.root,True)
|
||||
def sort(self,reverse = False):
|
||||
''' return a generator of sorted data'''
|
||||
def inOrder(root):
|
||||
if root is None:return
|
||||
if reverse:
|
||||
yield from inOrder(root.right)
|
||||
else:
|
||||
yield from inOrder(root.left)
|
||||
yield root
|
||||
if reverse:
|
||||
yield from inOrder(root.left)
|
||||
else:
|
||||
yield from inOrder(root.right)
|
||||
yield from inOrder(self.root)
|
||||
def getSuccessor(self,val):
|
||||
if isinstance(val,node):val = val.val
|
||||
def _inOrder(root):
|
||||
if root is None:return
|
||||
if root.val>= val:yield from _inOrder(root.left)
|
||||
yield root
|
||||
yield from _inOrder(root.right)
|
||||
gen = _inOrder(self.root)
|
||||
for i in gen:
|
||||
if i.val==val:
|
||||
try: return gen.__next__()
|
||||
except:return None
|
||||
|
||||
def delete(self,val):
|
||||
# delete node in a binary search tree
|
||||
if isinstance(val,node):val = val.val
|
||||
nd = self.find(val)
|
||||
if nd is None: return
|
||||
y = None
|
||||
if nd.left and nd.right:
|
||||
y= self.getSuccessor(val)
|
||||
else:
|
||||
y = nd
|
||||
py = self.getParent(y.val)
|
||||
x = y.left if y.left else y.right
|
||||
if py is None:
|
||||
self.root = x
|
||||
elif y==py.left:
|
||||
py.left = x
|
||||
else:
|
||||
py.right = x
|
||||
if y != nd:
|
||||
nd.val = y.val
|
||||
|
||||
# adjust colors and rotate
|
||||
if self.checkBlack(y): self.fixUpDel(py,x)
|
||||
|
||||
def fixUpDel(self,prt,chd):
|
||||
if self.root == chd or not self.checkBlack(chd):
|
||||
self.setBlack(chd, True)
|
||||
return
|
||||
isLeft = prt.left == chd
|
||||
brother = prt.getChild(not isLeft)
|
||||
if self.checkBlack(brother):
|
||||
# case 1: brother is black
|
||||
lb = self.checkBlack(brother.left)
|
||||
rb = self.checkBlack(brother.right)
|
||||
if lb and rb:
|
||||
# case 1.1: brother is black and two kids are black
|
||||
self.setBlack(brother,False)
|
||||
chd = prt
|
||||
elif lb or rb:
|
||||
# case 1.2: brother is black and two kids's colors differ
|
||||
if self.checkBlack(brother.getChild(not isLeft)):
|
||||
# uncle's son is nephew, and niece for uncle's daughter
|
||||
nephew = brother.getChild(isLeft),
|
||||
print(nephew)
|
||||
self.setBlack(nephew,True)
|
||||
self.setBlack(brother,False)
|
||||
|
||||
# brother right rotate
|
||||
prt.setChild(nephew,not isLeft)
|
||||
nephew.setChild(brother, not isLeft)
|
||||
brother = prt.right
|
||||
|
||||
# case 1.3: brother is black and two kids are red
|
||||
brother.isBlack = prt.isBlack
|
||||
self.setBlack(prt,True)
|
||||
self.setBlack(brother.right,True)
|
||||
|
||||
# prt left rotate
|
||||
prt.setChild(brother.getChild(isLeft),not isLeft)
|
||||
brother.setChild(prt,isLeft)
|
||||
chd = self.root
|
||||
|
||||
|
||||
else:
|
||||
# case 2: brother is red
|
||||
prt.setChild(brother.getChild(isLeft), not isLeft)
|
||||
brother.setChild(prt, isLeft)
|
||||
self.setBlack(prt,False)
|
||||
self.setBlack(brother,True)
|
||||
self.setBlack(chd,True)
|
||||
|
||||
def display(self):
|
||||
def getHeight(nd):
|
||||
if nd is None:return 0
|
||||
return max(getHeight(nd.left),getHeight(nd.right)) +1
|
||||
def levelVisit(root):
|
||||
from collections import deque
|
||||
lst = deque([root])
|
||||
level = []
|
||||
h = getHeight(root)
|
||||
lv = 0
|
||||
ct = 0
|
||||
while lv<=h:
|
||||
ct+=1
|
||||
nd = lst.popleft()
|
||||
if ct >= 2**lv:
|
||||
lv+=1
|
||||
level.append([])
|
||||
level[-1].append(str(nd))
|
||||
if nd is not None:
|
||||
lst.append(nd.left)
|
||||
lst.append(nd.right)
|
||||
else:
|
||||
lst.append(None)
|
||||
lst.append(None)
|
||||
return level
|
||||
lines = levelVisit(self.root)
|
||||
print('-'*5+ 'level visit' + '-'*5)
|
||||
return '\n'.join([' '.join(line) for line in lines])
|
||||
|
||||
def __str__(self):
|
||||
return self.display()
|
||||
|
||||
|
||||
def genNum(n =10):
|
||||
nums =[]
|
||||
for i in range(n):
|
||||
while 1:
|
||||
d = randint(0,100)
|
||||
if d not in nums:
|
||||
nums.append(d)
|
||||
break
|
||||
#nums = [3,4,2,0,1,6]
|
||||
return nums
|
||||
|
||||
def buildTree(n=10,nums=None,visitor=None):
|
||||
if nums is None: nums = genNum(n)
|
||||
rbtree = redBlackTree()
|
||||
print(f'build a red-black tree using {nums}')
|
||||
for i in nums:
|
||||
if visitor:
|
||||
visitor(rbtree)
|
||||
rbtree.insert(i)
|
||||
return rbtree
|
||||
def testInsert():
|
||||
def visitor(t):
|
||||
print(t)
|
||||
nums = [66, 14, 7, 2, 52, 96, 63, 51, 16, 53]
|
||||
rbtree = buildTree(nums = nums,visitor = visitor)
|
||||
print('-'*5+ 'in-order visit' + '-'*5)
|
||||
for i,j in enumerate(rbtree.sort()):
|
||||
print(f'{i+1}: {j}')
|
||||
|
||||
def testSuc():
|
||||
rbtree = buildTree()
|
||||
for i in rbtree.sort():
|
||||
print(f'{i}\'s suc is {rbtree.getSuccessor(i)}')
|
||||
|
||||
def testDelete():
|
||||
#nums = [56, 89, 31, 29, 24, 8, 62, 96, 20, 75] #tuple
|
||||
nums = [66, 14, 7, 2, 52, 96, 63, 51, 16, 53]
|
||||
rbtree = buildTree(nums = nums)
|
||||
print(rbtree)
|
||||
shuffle(nums)
|
||||
for i in nums:
|
||||
print(f'deleting {i}')
|
||||
rbtree.delete(i)
|
||||
print(rbtree)
|
||||
|
||||
if __name__=='__main__':
|
||||
testInsert()
|
||||
#testDelete()
|
||||
#testSuc()
|
||||
|
|
@ -1,7 +1,7 @@
|
|||
---
|
||||
title: 『算法』general
|
||||
date: 2018-07-04
|
||||
categories: 算法与数据结构
|
||||
categories: 数据结构与算法
|
||||
tags: [算法]
|
||||
keywords:
|
||||
mathjax: true
|
||||
|
|
@ -61,13 +61,19 @@ top:
|
|||
|
||||
|
||||
<a id="markdown-4-算法设计" name="4-算法设计"></a>
|
||||
# 4. 算法设计
|
||||
# 4. 算法设计与分析
|
||||
<a id="markdown-41-分治divide-and-conquer" name="41-分治divide-and-conquer"></a>
|
||||
## 4.1. 分治(divide and conquer)
|
||||
结构上是递归的,
|
||||
步骤: 分解,解决, 合并
|
||||
eg 快排,归并排序
|
||||
|
||||
## 随机化
|
||||
## 递归
|
||||
## 动态规划
|
||||
## 贪心算法
|
||||
## 平摊分析
|
||||
|
||||
<a id="markdown-5-递归式" name="5-递归式"></a>
|
||||
# 5. 递归式
|
||||
$T(n) = aT(\frac{n} {b})+f(n)$
|
||||
|
|
@ -78,6 +84,7 @@ eg 快排,归并排序
|
|||
### 5.1.1. 步骤
|
||||
* 猜测解的形式
|
||||
* 用数学归纳法找出常数
|
||||
|
||||
<a id="markdown-512-例子" name="512-例子"></a>
|
||||
### 5.1.2. 例子
|
||||
$T(n) = 2T(\frac{n} {2})+n$
|
||||
|
|
@ -91,12 +98,14 @@ $
|
|||
T(n) &\leqslant 2c\frac{n}{2}log(\frac{n}{2}) + n \leqslant cnlog(\frac{n}{2}) \\
|
||||
\end{aligned}
|
||||
$
|
||||
|
||||
<a id="markdown-513-放缩" name="513-放缩"></a>
|
||||
### 5.1.3. 放缩
|
||||
对于 $T(n) = 2T(\frac{cn}{2}) + 1$
|
||||
如果 直接猜测 $T(n) = O (n)$ 不能证明,
|
||||
而且不要猜测更高的界 $O (n^2)$
|
||||
可以放缩为 n-b
|
||||
|
||||
<a id="markdown-514-改变变量" name="514-改变变量"></a>
|
||||
### 5.1.4. 改变变量
|
||||
对于 $ T(n) = 2T(\sqrt{n})+logn $
|
||||
|
|
@ -176,6 +185,7 @@ for i in range(n):
|
|||
初始化: i=1 成立
|
||||
保持 : 假设 在第 i-1 次迭代之前,成立, 证明在第 i 次迭代之后, 仍然成立,
|
||||
终止: 在 结束后, i=n+1, 得到 概率为 $\frac{1}{n!}$
|
||||
|
||||
<a id="markdown-7-组合方程的近似算法" name="7-组合方程的近似算法"></a>
|
||||
# 7. 组合方程的近似算法
|
||||
* Stiring's approximation: $ n! \approx \sqrt{2\pi n}\left(\frac{n}{e}\right)^n$
|
||||
|
|
@ -188,7 +198,7 @@ for i in range(n):
|
|||
## 8.1. 球与盒子
|
||||
把相同的秋随机投到 b 个盒子里,问在每个盒子里至少有一个球之前,平均至少要投多少个球?
|
||||
称投入一个空盒为击中, 即求取得 b 次击中的概率
|
||||
设投 n 次, 称第 i 个阶段包括第 i-1 次击中到 第 i 次击中的球, 则 $p_i=\frac{b-i+1}{b}$
|
||||
设投 n 次, 称第 i 个阶段包括第 i-1 次击中到 第 i 次击中的球, 则第 i 次击中的概率为 $p_i=\frac{b-i+1}{b}$
|
||||
用 $n_i$表示第 i 阶段的投球数,则 $n=\sum_{i=1}^b n_i$
|
||||
且 $n_i$服从几何分布, $E(n_i)=\frac{b}{b-i+1}$,
|
||||
则由期望的线性性,
|
||||
|
|
|
|||
|
|
@ -1,7 +1,7 @@
|
|||
---
|
||||
title: 『数据结构』散列表
|
||||
date: 2018-07-08 23:25
|
||||
categories: 算法与数据结构
|
||||
categories: 数据结构与算法
|
||||
tags: [数据结构,散列表]
|
||||
keywords:
|
||||
mathjax: true
|
||||
|
|
@ -28,7 +28,6 @@ description:
|
|||
|
||||
<!-- /TOC -->
|
||||
|
||||
**[github地址](https://github.com/mbinary/algorithm-and-data-structure.git)**
|
||||
|
||||
哈希表 (hash table) , 可以实现 $O(1)$ 的 read, write, update
|
||||
相对应 python 中的 dict, c语言中的 map
|
||||
|
|
@ -115,9 +114,9 @@ $$
|
|||
对于 m 个槽位的表, 只需 $\Theta(n)$的期望时间来处理 n 个元素的 insert, search, delete,其中 有$O(m)$个insert 操作
|
||||
<a id="markdown-2313-实现" name="2313-实现"></a>
|
||||
#### 2.3.1.3. 实现
|
||||
选择足够大的 prime p, 记$Z_p=\{0,1,\ldots,p-1\}, Z_p^*=\{1,\ldots,p-1\},$
|
||||
选择足够大的 prime p, 记$Z_p=\{0,1,\ldots,p-1\}, Z_p^{*}=\{1,\ldots,p-1\},$
|
||||
令$h_{a,b}(k) = ((ak+b)mod\ p) mod\ m$
|
||||
则 $H_{p,m}=\{h_{a,b}|a\in Z_p^*,b\in Z_p\}$
|
||||
则 $H_{p,m}=\{h_{a,b}|a\in Z_p^{*},b\in Z_p\}$
|
||||
<a id="markdown-24-开放寻址法" name="24-开放寻址法"></a>
|
||||
## 2.4. 开放寻址法
|
||||
所有表项都在散列表中, 没有链表.
|
||||
|
|
@ -141,9 +140,16 @@ $h(k,i) = (h_1(k)+i*h_2(k))mod\ m$
|
|||
|
||||
<a id="markdown-241-不成功查找的探查数的期望" name="241-不成功查找的探查数的期望"></a>
|
||||
### 2.4.1. 不成功查找的探查数的期望
|
||||
对于开放寻址散列表,且 $\alpha<1$,在一次不成功的查找中,有 $E(\text{探查数})\leqslant \frac{1}{1-\alpha}$
|
||||
不成功探查是这样的: 前面都是检查到槽被占用但是关键字不同, 最后一次应该是空槽.
|
||||
对于开放寻址散列表,且 $\alpha<1$,一次不成功的查找,是这样的: 已经装填了 n 个, 总共有m 个,则空槽有 m-n 个.
|
||||
不成功的探查是这样的: 一直探查到已经装填的元素(但是不是要找的元素), 直到遇到没有装填的空槽. 所以这服从几何分布, 即
|
||||
$$
|
||||
p(\text{不成功探查})=p(\text{第一次找到空槽})=\frac{m-n}{m}
|
||||
$$
|
||||
有
|
||||
$$ E(\text{探查数})=\frac{1}{p}\leqslant \frac{1}{1-\alpha}$$
|
||||
|
||||
|
||||
|
||||
<a id="markdown-2411-插入探查数的期望" name="2411-插入探查数的期望"></a>
|
||||
#### 2.4.1.1. 插入探查数的期望
|
||||
所以, 插入一个关键字, 也最多需要 $\frac{1}{1-\alpha}$次, 因为插入过程就是前面都是被占用了的槽, 最后遇到一个空槽.与探查不成功是一样的过程
|
||||
|
|
@ -162,6 +168,8 @@ $$
|
|||
$$
|
||||
|
||||
代码
|
||||
|
||||
**[github地址](https://github.com/mbinary/algorithm-and-data-structure.git)**
|
||||
```python
|
||||
class item:
|
||||
def __init__(self,key,val,nextItem=None):
|
||||
|
|
|
|||
|
|
@ -1,7 +1,7 @@
|
|||
---
|
||||
title: 『算法』排序
|
||||
date: 2018-7-6
|
||||
categories: 算法与数据结构
|
||||
categories: 数据结构与算法
|
||||
tags: [算法,排序]
|
||||
keywords:
|
||||
mathjax: true
|
||||
|
|
@ -36,7 +36,6 @@ description:
|
|||
|
||||
<!-- /TOC -->
|
||||
|
||||
**[github地址](https://github.com/mbinary/algorithm-and-data-structure.git)**
|
||||
|
||||
排序的本质就是减少逆序数, 根据是否进行比较,可以分为如下两类.
|
||||
* 比较排序
|
||||
|
|
@ -411,7 +410,7 @@ if __name__ == '__main__':
|
|||
设有 n 个元素, 则设立 n 个桶
|
||||
将各元素通过数值线性映射到桶地址,
|
||||
类似 hash 链表.
|
||||
然后在每个桶内, 进行插入排序,
|
||||
然后在每个桶内, 进行插入排序($O(n_i^2)$)
|
||||
最后合并所有桶.
|
||||
这里的特点是 n 个桶实现了 $\Theta(n)$的时间复杂度, 但是耗费的空间 为 $\Theta(n)$
|
||||
|
||||
|
|
@ -474,3 +473,4 @@ def select(lst,i):
|
|||
return _select(0,len(lst)-1)
|
||||
```
|
||||
|
||||
**[github地址](https://github.com/mbinary/algorithm-and-data-structure.git)**
|
||||
|
|
|
|||
|
|
@ -1,7 +1,7 @@
|
|||
---
|
||||
title: 『数据结构』树
|
||||
date: 2018-7-11 18:56
|
||||
categories: 算法与数据结构
|
||||
categories: 数据结构与算法
|
||||
tags: [数据结构,树]
|
||||
keywords:
|
||||
mathjax: true
|
||||
|
|
@ -26,6 +26,7 @@ description:
|
|||
- [5.2.2. Zig-zig step](#522-zig-zig-step)
|
||||
- [5.2.3. Zig-zag step](#523-zig-zag-step)
|
||||
- [5.3. read-black Tree](#53-read-black-tree)
|
||||
- [5.4. treap](#54-treap)
|
||||
- [6. 总结](#6-总结)
|
||||
- [7. 附代码](#7-附代码)
|
||||
- [7.1. 二叉树(binaryTree)](#71-二叉树binarytree)
|
||||
|
|
@ -35,7 +36,6 @@ description:
|
|||
|
||||
<!-- /TOC -->
|
||||
|
||||
**[github地址](https://github.com/mbinary/algorithm-and-data-structure.git)**
|
||||
<a id="markdown-1-概念" name="1-概念"></a>
|
||||
# 1. 概念
|
||||
* 双亲
|
||||
|
|
@ -87,13 +87,13 @@ $$
|
|||
然后我们来看`<<算法导论>>`(p162,思考题12-4)上怎么求的吧( •̀ ω •́ )y
|
||||
设生成函数
|
||||
$$B(x)=\sum_{n=0}^{\infty}b_n x^n$$
|
||||
下面证明$B(x)=xB(x)^2+1\quad(\#)$
|
||||
下面证明$B(x)=xB(x)^2+1$
|
||||
易得$$xB(x)^2=\sum_{i=1}^{\infty}\sum_{n=i}^{\infty}b_{i-1}b_{n-i}x^n$$
|
||||
对比$B(x), xB(x)^2+1$的 x 的各次系数,分别是 $b_k,a_{k}$
|
||||
当 k=0, $a_k=1=b_k$
|
||||
当 k>0
|
||||
$$a_{k} = \sum_{i=1}^{k}b_{i-1}b_{k-i} = b_k$$
|
||||
所以$B(x)=xB(x)^2+1\quad(\#)$
|
||||
所以$B(x)=xB(x)^2+1$
|
||||
由此解得
|
||||
$$B(x)=\frac{1-\sqrt{1-4x} }{2x}$$
|
||||
在点 x=0 处,
|
||||
|
|
@ -121,7 +121,9 @@ $$
|
|||
* 若A,B都是好括号列, 则串联后 AB是好括号列
|
||||
* 若A是好括号列, 则 (A)是好括号列
|
||||
|
||||
>充要条件: 好括号列 <=> 左右括号数相等, 且从左向右看, 看到的右括号数不超过左括号数
|
||||
>充要条件: 好括号列 $\Longleftrightarrow$ 左右括号数相等, 且从左向右看, 看到的右括号数不超过左括号数
|
||||
|
||||
|
||||
|
||||
>定理: 由 n个左括号,n个右括号组成的好括号列个数为$c(n)=\frac{C_{2n}^{n}}{n+1}$
|
||||
|
||||
|
|
@ -209,6 +211,27 @@ Aho-Corasick automation,是在字典树上添加匹配失败边(失配指针),
|
|||
## 5.3. read-black Tree
|
||||
同样是平衡的二叉树, 以后单独写一篇关于红黑树的.
|
||||
|
||||
<a id="markdown-54-treap" name="54-treap"></a>
|
||||
## 5.4. treap
|
||||
[前面提到](#21-随机构造的二叉查找树), 随机构造的二叉查找树高度为 $h=O(logn)$,以及在[算法 general](https://mbinary.coding.me/alg-genral.html) 中说明了怎样 随机化(shuffle)一个给定的序列.
|
||||
|
||||
所以,为了得到一个平衡的二叉排序树,我们可以将给定的序列随机化, 然后再进行构造二叉排序树.
|
||||
|
||||
但是如果不能一次得到全部的数据,也就是可能插入新的数据的时候,该怎么办呢? 可以证明,满足下面的条件构造的结构相当于同时得到全部数据, 也就是随机化的二叉查找树.
|
||||
|
||||
|
||||
|
||||
这种结构叫 `treap`, 不仅有要排序的关键字 key, 还有随机生成的,各不相等的关键字`priority`,代表插入的顺序.
|
||||
|
||||
* 二叉查找树的排序性质: 双亲结点的 key 大于左孩子,小于右孩子
|
||||
* 最小(大)堆的堆序性质: 双亲的 prority小于(大于) 孩子的 prority
|
||||
|
||||
|
||||
插入的实现: 先进行二叉查找树的插入,成为叶子结点, 再通过旋转 实现 `上浮`(堆中术语).
|
||||
将先排序 key, 再排序 prority(排序prority 时通过旋转保持 key 的排序)
|
||||
|
||||
|
||||
|
||||
<a id="markdown-6-总结" name="6-总结"></a>
|
||||
# 6. 总结
|
||||
还有很多有趣的树结构,
|
||||
|
|
@ -218,6 +241,8 @@ Aho-Corasick automation,是在字典树上添加匹配失败边(失配指针),
|
|||
|
||||
<a id="markdown-7-附代码" name="7-附代码"></a>
|
||||
# 7. 附代码
|
||||
**[github地址](https://github.com/mbinary/algorithm-and-data-structure.git)**
|
||||
|
||||
<a id="markdown-71-二叉树binarytree" name="71-二叉树binarytree"></a>
|
||||
## 7.1. 二叉树(binaryTree)
|
||||
```python
|
||||
|
|
|
|||
Loading…
Reference in New Issue
Block a user