L9 Hashing

Posted on 2021-08-02 Edited on 2024-12-04 In Computer Science Views:

Outline:

Ref:

The searching problem

也称为Chaining
Each address is a linked list
Insert to the head of the list( 链表不是常数时间寻址 )
Assumption - simple uniform hashing
- For $j=0,1,2,\dots, m-1$, the average length of the list at $E[j]$ is n/m

The average cost for an unsuccessful search
- Any key that is not in the table is equally likely to hash to any of the m address
- Total cost $\Theta(1+n/m)$
  - The average cost to determine that the key is not in the $E[h(k)]$ is the cost to search to the end of the list, which is n/m
  - 1是算哈希， n/m是链表平均长度

For successful search( assuming that $x_i$ is the $i^{th}$ element inserted into the table, $i=1,2,\dots,n$ )
- For each i, the probability of that $x_i$ is searched is 1/n
- For a specific $x_i$, the number of elements examined in a successful search is t+1, where t is the number of elements inserted into the same list as $x_i$, after $x_i$ has been inserted(即链表里在该元素之前的元素的个数). \[ 1 + \frac 1 n \sum\limits_{i=1}^{n}{(1+t)} \]
- And for any j, the probability of that $x_j$ is inserted into the same list of $x_i$ is 1/m. So, the cost is:
  
  \[ 1 + \frac 1 n \sum\limits_{j=i+1}^{n}{\frac 1 m} \]
The average cost of a successful search:
- Define $\alpha = n/m$ as load factor, The average cost of a successful search is: \[ \frac 1 n { ( 1 + \sum\limits_{j = i + 1}^n {\frac 1 m} )} = 1 + \frac 1 {nm} {\sum\limits_{i = 1}^n {(n-i)}} = 1 + \frac 1 {nm} \sum_{i=1}^{n-1}i = 1 + \frac {n-1} {2m} = 1 + \frac \alpha 2 - \frac \alpha {2n} = \Theta( 1 + \alpha ) \]

All elements are stored in the hash table
- No linked list is used
- The load factor $\alpha$ can't be larger than 1
Collision is settled by "rehashing"
- A function is used to get a new hashing address for each collided address
  - The hash table slots are probed successively, until a valid location is found.
Th probe sequence can be seen as a permutation of $(0,1,2,\dots,m-1)$

Linear probing:
- Given an ordinary hash function h', which is called an auxiliary hash function, the hash function is:( clustering may occur ) \[ h(k,i) = (h'(k)+i)\quad mod \quad m \quad ( i = 0,1, \dots, m-1 ) \]
Quadratic Probing:
- Given auxiliary function h' and nonzero auxiliary constant $c_1$ and $c_2$, the hash function is: (secondary clustering may occur) \[ h(k,i) = (h'(k)+c_1 i + c_2 i^2)\quad mod \quad m \quad ( i = 0,1, \dots, m-1 \]
Double hashing:
- Given auxiliary functions h_1 and h_2, the hash function is: \[ h(k,i) = (h_1(k)+i h_2(k))\quad mod \quad m \quad ( i = 0,1, \dots, m-1 \]

Assumption
- Each key is equally likely to have any of the m! permutations of $(1,2,\dots,m)$ as its probe sequence
Note
- Both linear and quadratic probing have only m distinct probe sequence, as determined by the first probe.(m个格子的合法排列)

The average number of probes in an successful search is at most $\frac 1 \alpha ln \frac 1 {1-\alpha}$ ($\alpha = n /m <1$)
- Assuming uniform hashing
  
  To search for the $(i+1)^{th}$ inserted element in the table, the cost is the same as that for inserting it when there are just i elements in the table. At that time, $\alpha = \frac i m$. So the cost is( 用 unsuccessful的结论 ) $\frac 1 {1- \frac i m}$ = $\frac m {m-i}$, \[ \frac 1 n \sum\limits_{i=0}^{n-1} \frac m {m-i} = \frac m n \sum\limits _{i=0}^{n-1} \frac 1 {m-i} = \frac 1 \alpha \sum\limits _{i=m-n+1}^{m} \frac 1 i \le \frac 1 \alpha \int _{m-n} ^m {\frac {dx} x} = \frac 1 \alpha ln \frac m {m-n} = \frac 1 \alpha ln \frac 1 {1-\alpha} \]

For n execution of insertion operations
- Note that the expansion is required during the $i^{th}$ operation only if $i = 2^{th}$, and the cost of the $i^{th}$ operation $ c_i = i $( if i - 1 is exactly the power of 2 ) or $c_i = 1 $ (otherwise)
So the total cost is: $\sum\limits _{i=1} ^n c_{i} \le n + \sum\limits _{j=0} ^{\lfloor logn \rfloor} 2^j < n + 2n = 3n$

Unusually expensive operations
Relation between expensive and usual operations
- Each piece of the doubling cost corresponds to some previous insert

Amortized equation:

amortized cost = actual cost + accounting cost
Design goals for accounting cost
- In any legal sequence of operations, the sum of the accounting costs is nonnegative
- The amortized cost of each operation is fairly regular, in spite of the wide fluctuate possible for the actual cost of individual operations
Examples:
1. Multi-pop Stacks
Amortized Actual Accounting

Push 2 1 1

Multi-pop 0 k -k
- 相当于每个元素在出生的时候就要记下它死亡的时候的代价
1. Binary Counter
Amortized Actual Account

Set 1 2 1 1

Set 0 0 1 -1

	Amortized	Actual	Accounting
Push	2	1	1
Multi-pop	0	k	-k

	Amortized	Actual	Account
Set 1	2	1	1
Set 0	0	1	-1

	Amortized	Actual	Account
Insert ( normal )	3	1	2
Insert ( doubling )	3	k+1	-k + 2