Hashing:

Introduction
Michael Levin
Department of Computer Science and Engineering
University of California, San Diego

Data Structures Fundamentals

Algorithms and Data Structures
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
Blockchain
Programming Languages
Programming Languages

dict
 Programming Languages

HashMap

dict
Programming Languages

Keywords: for, if, while, int, …

File Systems
Digital Signature
Digital Signature
Digital Signature
Digital Signature
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
Who’s Calling?
Who’s Calling?
Who’s Calling?
 Phone Book

Phone number Name

01707773331 Maria
239-17-17 Sasha
575-75-75 Helen
 Phone to Name

We are going to focus on retrieving name by

phone number for now
 Local Phone Numbers

Like 123-23-23
 Local Phone Numbers

Like 123-23-23
Typically up to 7 digits
 Local Phone Numbers

Like 123-23-23
Typically up to 7 digits
Sufficient for 107 = 10 000 000 phone
numbers
Convert Phone Number to Integer

Examples
123-23-23 → 1 232 323
049 12 12 → 491 212
5757575 → 5 757 575
 Direct Addressing


 Phone number Name





 0000000



 …



 2391717
7
Sasha
10 rows

 …





 5757575 Helen



 …



 9999999
 Direct Addressing

Store phone book as array of size 107

Names are values of the array
To retrieve name by phone number,
convert phone number to integer first
Use the resulting integer as index in the
array of names
GetName(phoneNumber)
index ← ConvertToInt(phoneNumber)
return phoneBookArray[index]

SetName(phoneNumber, name)
index ← ConvertToInt(phoneNumber)
phoneBookArray[index] ← name
GetName(phoneNumber)
index ← ConvertToInt(phoneNumber)
return phoneBookArray[index]

SetName(phoneNumber, name)
index ← ConvertToInt(phoneNumber)
phoneBookArray[index] ← name
GetName(phoneNumber)
index ← ConvertToInt(phoneNumber)
return phoneBookArray[index]

SetName(phoneNumber, name)
index ← ConvertToInt(phoneNumber)
phoneBookArray[index] ← name
GetName(phoneNumber)
index ← ConvertToInt(phoneNumber)
return phoneBookArray[index]

SetName(phoneNumber, name)
index ← ConvertToInt(phoneNumber)
phoneBookArray[index] ← name
GetName(phoneNumber)
index ← ConvertToInt(phoneNumber)
return phoneBookArray[index]

SetName(phoneNumber, name)
index ← ConvertToInt(phoneNumber)
phoneBookArray[index] ← name
GetName(phoneNumber)
index ← ConvertToInt(phoneNumber)
return phoneBookArray[index]

SetName(phoneNumber, name)
index ← ConvertToInt(phoneNumber)
phoneBookArray[index] ← name
 Asymptotics

For a phone book with n contacts,

Retrieve name by phone number in O(1)
 Asymptotics

For a phone book with n contacts,

Retrieve name by phone number in O(1)
Set name for a phone number in O(1)
 Asymptotics

For a phone book with n contacts,

Retrieve name by phone number in O(1)
Set name for a phone number in O(1)
Memory consumption is O(|U|), where
U is the set of all possible phone
numbers
 Conclusion

Local phone numbers are up to 7 digits

long
Can store them in an array of size 107
This scheme is called direct addressing
It is the simplest form of hashing
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
International Phone Numbers
Like +1-800-700-00-00
International Phone Numbers
Like +1-800-700-00-00
Can be up to 15 digits:
+594 700 123 233 455
International Phone Numbers
Like +1-800-700-00-00
Can be up to 15 digits:
+594 700 123 233 455
Using direct addressing requires array of
size 1015, which would take 7PB (7
petabytes) to store one phone book
(1PB = 1024TB, 1TB = 1024GB)
International Phone Numbers
Like +1-800-700-00-00
Can be up to 15 digits:
+594 700 123 233 455
Using direct addressing requires array of
size 1015, which would take 7PB (7
petabytes) to store one phone book
(1PB = 1024TB, 1TB = 1024GB)
Your phone memory is probably at most
256GB, so you would need 28762
phones to store your phone book :)
 Idea
Direct addressing requires too much
memory
 Idea
Direct addressing requires too much
memory
Array is huge because it has a cell for
every possible phone number
 Idea
Direct addressing requires too much
memory
Array is huge because it has a cell for
every possible phone number
Let’s store only the known phone
numbers
 Idea
Direct addressing requires too much
memory
Array is huge because it has a cell for
every possible phone number
Let’s store only the known phone
numbers
Put pairs (Phone number, Name) into a
doubly-linked list
 Idea

14052391717 15025757575 01707773331

Sasha Helen Maria
 Operations
To add a contact, just insert new pair
(Phone number, Name) into the list in
O(1)
 Operations
To add a contact, just insert new pair
(Phone number, Name) into the list in
O(1)
To retrieve name by phone number,
search through the list...
 Operations
To add a contact, just insert new pair
(Phone number, Name) into the list in
O(1)
To retrieve name by phone number,
search through the list...
...in O(n), where n is the total number
of contacts
 Operations
To add a contact, just insert new pair
(Phone number, Name) into the list in
O(1)
To retrieve name by phone number,
search through the list...
...in O(n), where n is the total number
of contacts
Too slow
 Idea 2

Retrieving a name by phone number is

slow, because we need to look through
the whole list
 Idea 2

Retrieving a name by phone number is

slow, because we need to look through
the whole list
Let’s put the pairs (Phone number,
Name) in a dynamic array sorted by
phone number!
 Idea 2

01707773331 Maria
14052391717 Sasha
15025757575 Helen
 Operations
Retrieve name by phone number using
binary search in O(log n)
 Operations
Retrieve name by phone number using
binary search in O(log n)
To insert a new contact, find
appropriate position in O(log n), then
insert in...
 Operations
Retrieve name by phone number using
binary search in O(log n)
To insert a new contact, find
appropriate position in O(log n), then
insert in...
...O(n), because we need to first move
part of the array 1 position to the right
 Operations
Retrieve name by phone number using
binary search in O(log n)
To insert a new contact, find
appropriate position in O(log n), then
insert in...
...O(n), because we need to first move
part of the array 1 position to the right
Too slow again
 Conclusion

International numbers can be up to 15

digits long
Direct addressing requires 7 petabytes
of memory
Simple list-based and array-based
approaches are too slow
Next videos — solution using hashing
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
Encoding Phone Numbers

Encode international phone numbers

with small numbers
Encoding Phone Numbers

Encode international phone numbers

with small numbers
E.g. numbers from 0 to 999
Encoding Phone Numbers

Encode international phone numbers

with small numbers
E.g. numbers from 0 to 999
Different codes for the phone numbers
in the phone book
 Hash Function

Definition
For any set of objects S and any integer
m > 0, a function h : S → {0, 1, . . . , m − 1}
is called a hash function.
 Hash Function

Definition
For any set of objects S and any integer
m > 0, a function h : S → {0, 1, . . . , m − 1}
is called a hash function.

Definition
m is called the cardinality of hash function h.
 Desirable Properties
Hash function should be fast to
compute
 Desirable Properties
Hash function should be fast to
compute
Different values for different objects
 Desirable Properties
Hash function should be fast to
compute
Different values for different objects
Direct addressing with O(m) memory
 Desirable Properties
Hash function should be fast to
compute
Different values for different objects
Direct addressing with O(m) memory
Want small cardinality m
 Desirable Properties
Hash function should be fast to
compute
Different values for different objects
Direct addressing with O(m) memory
Want small cardinality m
Impossible to have all different values if
number of objects |S| is more than m
(by pigeonhole principle)
 Collisions

Definition
When h(o1) = h(o2) and o1 ̸= o2, this is a
collision.
 Desirable Properties

Hash function should be fast to

compute
 Desirable Properties

Hash function should be fast to

compute
Different values for different objects
Small probability of collision
 Desirable Properties

Hash function should be fast to

compute
Different values for different objects
Small probability of collision
Small enough cardinality m
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
 Map

Store mapping from objects to other objects:

Filename → location of the file
Phone number → name
Name → phone number
 Map

Definition
Map from set S of objects to set V of
values is a data structure with methods
HasKey(object), Get(object),
Set(object, value), where
object ∈ S, value ∈ V.
 Map

Definition
In a Map from S to V, objects from S are
usually called keys of the Map. Objects from
V are called values of the Map.
 Chaining for Phone Book

0
1
2
3
4
5
6
7
 Chaining for Phone Book
h(01707773331) = 4
0
1
2
3
4
5
6
7
 Chaining for Phone Book
h(01707773331) = 4
0
1
2
3
4 Maria01707773331
5
6
7
 Chaining for Phone Book
h(01707773331) = 4
0 h(14052391717) = 1
1
2
3
4 Maria01707773331
5
6
7
 Chaining for Phone Book
h(01707773331) = 4
0 h(14052391717) = 1
1 Sasha14052391717
2
3
4 Maria01707773331
5
6
7
 Chaining for Phone Book
h(01707773331) = 4
0 h(14052391717) = 1
1 Sasha14052391717
h(15025757575) = 4
2
3
4 Maria01707773331
5
6
7
 Chaining for Phone Book
h(01707773331) = 4
0 h(14052391717) = 1
1 Sasha14052391717
h(15025757575) = 4
2
3
4 Maria01707773331 Helen15025757575
5
6
7
Chaining for Phone Book
Select hash function h of cardinality m
Chaining for Phone Book
Select hash function h of cardinality m
Create array Chains of size m
Chaining for Phone Book
Select hash function h of cardinality m
Create array Chains of size m
Each element of Chains is a
doubly-linked list of pairs
(name, phoneNumber), called chain
Chaining for Phone Book
Select hash function h of cardinality m
Create array Chains of size m
Each element of Chains is a
doubly-linked list of pairs
(name, phoneNumber), called chain
Pair (name, phoneNumber) goes into
chain at position
h(ConvertToInt(phoneNumber)) in
the array Chains
Chaining for Phone Book
To look up name by phone number, go
to the chain corresponding to phone
number and look through all pairs
Chaining for Phone Book
To look up name by phone number, go
to the chain corresponding to phone
number and look through all pairs
To add a contact, create a pair
(name, phoneNumber) and insert it into
the corresponding chain
Chaining for Phone Book
To look up name by phone number, go
to the chain corresponding to phone
number and look through all pairs
To add a contact, create a pair
(name, phoneNumber) and insert it into
the corresponding chain
To remove a contact, go to the
corresponding chain, find the pair
(name, phoneNumber) and remove it
from the chain
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
 Implementation
Chains — array of chains
Each chain is a list of pairs (object, value)
HasKey(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return true
return false
 Implementation
Chains — array of chains
Each chain is a list of pairs (object, value)
HasKey(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return true
return false
 Implementation
Chains — array of chains
Each chain is a list of pairs (object, value)
HasKey(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return true
return false
 Implementation
Chains — array of chains
Each chain is a list of pairs (object, value)
HasKey(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return true
return false
 Implementation
Chains — array of chains
Each chain is a list of pairs (object, value)
HasKey(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return true
return false
 Implementation
Chains — array of chains
Each chain is a list of pairs (object, value)
HasKey(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return true
return false
 Implementation

Get(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return value
return N/A
 Implementation

Get(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return value
return N/A
 Implementation

Get(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return value
return N/A
 Implementation

Get(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return value
return N/A
 Implementation

Get(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return value
return N/A
 Implementation

Get(object)
chain ← Chains[hash(object)]
for (key, value) in chain:
if key == object:
return value
return N/A
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Implementation

Set(object, value)
chain ← Chains[hash(object)]
for pair in chain:
if pair.key == object:
pair.value ← value
return
chain.Append((object, value))
 Asymptotics

Lemma
Let c be the length of the longest chain in
Chains. Then the running time of HasKey,
Get, Set is Θ(c + 1).
 Asymptotics
Proof
If the chain corresponding to the
object is non-empty, but the object
is not found in the chain, we will scan
all c items — Θ(c) = Θ(c + 1)
 Asymptotics
Proof
If the chain corresponding to the
object is non-empty, but the object
is not found in the chain, we will scan
all c items — Θ(c) = Θ(c + 1)
If c = 0, we still need O(1) time, thus
the need for “+1”
 Asymptotics

Lemma
Let n be the number of different objects
currently in the map and m be the cardinality
of the hash function. Then the memory
consumption for chaining is Θ(n + m).
 Asymptotics

Proof
Θ(n) to store n pairs (object, value)
 Asymptotics

Proof
Θ(n) to store n pairs (object, value)
Θ(m) for array Chains of size m
 Outline
1 Applications

2 Phone Book

3 International Phone Numbers

4 Hash Functions

5 Chaining

6 Chaining Implementation and Analysis

7 Hash Tables
 Set

Definition
Set is a data structure with methods
Add(object), Remove(object),
Find(object).
 Set

Examples
Students on campus
 Set

Examples
Students on campus
Phone numbers of contacts
 Set

Examples
Students on campus
Phone numbers of contacts
Keywords in a programming language
 Implementing Set

Two ways to implement a set using chaining:

Set is equivalent to map from S to

V = {true}
 Implementing Set

Two ways to implement a set using chaining:

Set is equivalent to map from S to

V = {true}
Store just objects instead of pairs
(object, value) in the chains
 Implementation

Find(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return true
return false
 Implementation

Find(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return true
return false
 Implementation

Find(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return true
return false
 Implementation

Find(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return true
return false
 Implementation

Find(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return true
return false
 Implementation

Find(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return true
return false
 Implementation

Add(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return
chain.Append(object)
 Implementation

Add(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return
chain.Append(object)
 Implementation

Add(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return
chain.Append(object)
 Implementation

Add(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return
chain.Append(object)
 Implementation

Add(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return
chain.Append(object)
 Implementation

Add(object)
chain ← Chains[hash(object)]
for key in chain:
if key == object:
return
chain.Append(object)
 Implementation

Remove(object)
if not Find(object):
return
chain ← Chains[hash(object)]
chain.Erase(object)
 Implementation

Remove(object)
if not Find(object):
return
chain ← Chains[hash(object)]
chain.Erase(object)
 Implementation

Remove(object)
if not Find(object):
return
chain ← Chains[hash(object)]
chain.Erase(object)
 Implementation

Remove(object)
if not Find(object):
return
chain ← Chains[hash(object)]
chain.Erase(object)
 Implementation

Remove(object)
if not Find(object):
return
chain ← Chains[hash(object)]
chain.Erase(object)
 Hash Table

Definition
An implementation of a Set or a Map using
hashing is called a hash table.
 Programming Languages
Set:
unordered_set in C++
HashSet in Java
set in Python
Map:
unordered_map in C++
HashMap in Java
dict in Python
 Conclusion

Chaining is a technique to implement a

hash table
 Conclusion

Chaining is a technique to implement a

hash table
Number of objects n, hash function
cardinality m, longest chain length c
 Conclusion

Chaining is a technique to implement a

hash table
Number of objects n, hash function
cardinality m, longest chain length c
Memory consumption is Θ(n + m)
 Conclusion

Chaining is a technique to implement a

hash table
Number of objects n, hash function
cardinality m, longest chain length c
Memory consumption is Θ(n + m)
Operations work in time Θ(c + 1)
 Conclusion

Chaining is a technique to implement a

hash table
Number of objects n, hash function
cardinality m, longest chain length c
Memory consumption is Θ(n + m)
Operations work in time Θ(c + 1)
How to make both m and c small?