Design and Analysis of Algorithms

CS345

Lecture 34
Fibonacci Heap - I

1
REPRESENTING ANY TREE

as a binary tree

2
 Representing any tree as a binary tree

Any arbitrary rooted tree A binary tree

a
a

b
b c d e
c
d
f g h f e
g
k Left link : child
h
Right link: sibling

k
This representation has many applications 
3
 Representing any tree as a binary tree

Any arbitrary rooted tree A binary tree

a
a

b
b c d e
c
d
f g h f e
g
Left link : child
h
Right link: sibling

k
4
 EVOLUTION OF HEAPS

Necessity is the mother of invention…

5
 Two fundamental graph problems

Minimum Spanning Tree Single Source Shortest Paths

Prim’s Algorithm Dijkstra’s Algorithm

Heap

• O() time
Decrease-Label operations
O()
• Extract-smallest-label operation O() time

O() Optimal algorithm for

6
 Operations on a Heap
Binary heap Binomial heap Binomial heap Fibonacci Heap

Find-min() O() O() O() O()

Insert(, ) O() O() O() O()

Extract-min() O() O() O() O()

amortized amortized

Decrease-key(,,) O() O() O() O()

amortized

Merge-heaps(,) O() O() O() O()

Inspiration :
Binary numbers

7
 BINOMIAL TREE

Binomial tree of degree

8
 What is ?

𝑩 𝑘 −1

𝑩 𝑘 −1
𝑩𝑘

𝑩0 𝑩1 𝑩2 𝑩3
9
 What is ?

Height of = ? 𝑘
𝒉 ( 𝑘−1 ) + 1 𝑩 𝑘 −1

No. of nodes in = ? 2𝑘 𝑩 𝑘 −1
𝟐 𝑵 ( 𝑘−1 ) 𝑩𝑘

No. of nodes at depth in = ? ( )

𝑘
𝑖
+

𝑩4 10
 BINOMIAL HEAP

A collection of
Heap ordered Binomial trees

11
 Binomial heap of size
Least Most
significant significant
bit bit

𝑛=19

Head

𝑩0 𝑩1 𝑩4

Note that there is at most one Binomial tree of degree in a Binomial heap. 12
 Binomial heap of size

𝑛=19
How to implement
this structure ?

Head
3 5 1

9 2 7 4 9

8 7 6 19 23 67
𝑩0 𝑩1 𝑩4
26 91 12 63

34

13
 Binomial heap of size

𝑛=19

Head
3 5 1

9 2 7 4 9

8 7 6 19 23 67
𝑩0 𝑩1 𝑩4
26 91 12 63

34

14
 Binomial heap of size

𝑛=19
How to access parent?

Head
3 5 1

9 2 7 4 9

8 7 6 19 23 67
𝑩0 𝑩1 𝑩4
26 91 12 63 Degree
Key
Parent
34
Left Right
15
Structure of a node
 MERGING
TWO BINOMIAL HEAPS

Spend some time on your own to find out an efficient way for this operation

16
 Merging two Binomial Heaps
𝑛=11 𝑛=5

3 6 1 8 5

7 4 9
11 9 17

19 23 67
𝑩0 𝑩1 𝑩3 𝑩0
21
𝑩2
63

17
 Merging two Binomial Heaps
𝑛=11 𝑛=5 𝑛=16
4

Inspiration: How to fix this problem ?

Addition of binary numbers

3 3
8 63 5 1

7 4 9
8 6 8
11 9 17

19 23 67
𝑩0 𝑩 0111 𝑩 21 21 𝑩2 𝑩3
63

The only problem is that there are multiple binomial

trees of same degree in the root list.
18
 Merging two Binomial Heaps
𝑛=11 𝑛=5 𝑛=16
4

3 35 1

69 8 3 4 9
6 85 17 7

9 17 5 11 6 8 19 23 67
11 𝑩2 𝑩 21
3 𝑩2 𝑩3
21 9 17 11 63

21

19
 OTHER OPERATIONS
ON A BINOMIAL HEAP

20
 Find-min on a Binomial Heap

O() time

Head
3 6 1

7 4 9
11

19 23 67
𝑩0 𝑩1 𝑩3
63

21
 Decrease-key on a Binomial Heap

Head
3 6 1

7 4 9
11

19 23 67
𝑩0 𝑩1 𝑩3
63
2

22
 Decrease-key on a Binomial Heap

Head
3 6 1

7 4 9
11

2 23 67
𝑩0 𝑩1 𝑩3
19

23
 Decrease-key on a Binomial Heap

O() time

Head
3 6 1

2 4 9
11

7 23 67
𝑩0 𝑩1 𝑩3
19

24
 Extract-min on a Binomial Heap

Can you exploit this similarity to

perform Extract-min efficiently ?
Do you notice any similarity between
the root list of a binomial heap and
children list of any node?
Head
3 6 1

2 4 9
11

7 23 67
𝑩0 𝑩1 𝑩3
19

The children list, after reversal, can be seen as a

binomial heap.
25
 Extract-min on a Binomial Heap

O() time

Merge the two

heaps

Head Head
3 6 9 4 2

67 7 23
11

19
𝑩0 𝑩1 𝑩0 𝑩1 𝑩2

26
 Insert() on a Binomial Heap

O() time

Merge the two

heaps

Head Head
3 6 1 𝑥

7 4 9
11

19 23 67
𝑩0 𝑩1 𝑩3
63

27
 Operations on a Heap
Binary heap Binomial heap Binomial heap

Maintain
Find-min() O() O() O() it explicitly

Insert(, ) O() O() O()

Extract-min() O() O() O()

amortized

Decrease-key(,,) O() O() O() Do these

Operations
Merge-heaps(,) O() O() O() lazily

Keep root list as a doubly linked circular list instead of single link list
Keep siblings in a doubly linked circular list instead of single link list
Wake up only when Extract-min is invoked
28
 Insert() on a Binomial Heap

Insert()

Head Head
8 6 5 4

7 15 9
11

19 23 67

63

𝑯
29
 Insert() on a Binomial Heap

Insert()

Merge the two

heaps

Head Head
8 6 5 4

7 15 9
11

19 23 67

63

𝑯
30
 Insert() on a Binomial Heap

Insert()
Insert() O() time

Head

Head
8 6 5 4 19

7 15 9
11

19 23 67

63

𝑯
31
 Insert() on a Binomial Heap

Insert()
Insert() O() time
Insert()

Head

8 6 5 4 19 2

7 15 9
11

19 23 67

63

𝑯
32
 Insert() on a Binomial Heap

Insert()
Insert() O() time
Insert()
Extract-min()

Head Head

8 6 5 4 19 2

7 15 9
11

19 23 67

63

𝑯
Homework: Make sincere attempt to design ways to implement Extract-min in 33
amortized O() time. We shall discuss it in the next class 