Algorithmics

CT065-3.5-3

Advanced Data Structures -2

Level 3 – Computing (Software Engineering)
 Topic & Structure of Lesson

• Heaps
– Binary Heaps
– Binomial Heaps
– Fibonacci Heaps

Module Code and Module Title Title of Slides

 Learning Outcomes

By the end of this lesson you should

be able to:
• Understand the characteristics and
operations of various heaps

Module Code and Module Title Title of Slides

 Binary Heap

• Data structure created using a binary tree,

with two additional properties:
– Shape: tree is either complete (all leaves at
same level), or almost complete (leave
nodes filled from left to right and leaves are at
different levels)
– Heap: each parent node is greater than or
equal to it’s children according to some
comparison function that defines the heap

Module Code and Module Title Title of Slides

 Binary Heap

• Heap: “Greater than” is not meant in the

numerical sense, but according to the
comparison function that sorts the heap
– Does not affect ordering of sibling nodes.
– Child nodes can be freely interchanged (only
if heap property is not violated)

Module Code and Module Title Title of Slides

 Examples

This is legal This is also legal

Min-Heap Max-Heap

Module Code and Module Title Title of Slides

 Insertion

• Up-Heap, Bubble/Percolate/Sift-Up
• Worst -> O(log n)
• Average -> O(1)

1. Add value as a leaf (at bottom level)

2. Compare value with parent, stop if order is
correct.
3. If not, swap with parent.
4. Repeat steps 2 – 3 until order is correct.

Module Code and Module Title Title of Slides

 Insertion

Module Code and Module Title Title of Slides

 Insertion

Module Code and Module Title Title of Slides

 Root Deletion

Deletion can only be done on the root.

1. Swap Root with lowest + rightmost leaf node

2. Delete that leaf
3. Compare root value with child nodes, stop if
order is correct
4. If not, swap with child with value that is “greater
than” root
5. Repeat 3 – 4 until order is correct

Module Code and Module Title Title of Slides

 Root Deletion

1) Swap 89 with 1
2) Check 89 with 23 and 12. This is a Min-
Heap, so swap with lowest value
Module Code and Module Title Title of Slides
 Root Deletion

3) Swap 89 with 19. The Heap is now

properly sorted.

Module Code and Module Title Title of Slides

 Binary Heap: Implementation

• Heaps are either complete, or almost complete binary

trees
– Can be stored compactly in arrays
– Pointers not required
• For each node in the ith index
– Parent (i) = floor (i / 2) //the largest integer not greater than x
– Left (i) = 2i + 1
– Right (i) = 2i + 2

Module Code and Module Title Title of Slides

 Array Implementation

Parent (1) = floor (1 / 2) = 0th index

Left (1) = 2 * 1 + 1 = 3rd index
Right (1) = 2 * 1 + 2 = 4th index

Module Code and Module Title Title of Slides

 Binomial Tree

• Recursive definition:

Module Code and Module Title Title of Slides

 Binomial Tree

• Useful properties of order k binomial tree

Bk
– Number of nodes = 2k.
– Height = k.
– Degree of root = k.
– Deleting root yields binomial trees Bk-1, Bk-2,…
B0.
• Proof – by induction on k.

Module Code and Module Title Title of Slides

 Binomial Tree

The resulting binomial trees created once the root is deleted

are also binomial trees of order k-1,…0.

Module Code and Module Title Title of Slides

 Binomial Heap

• Ordered sequence of binomial trees that satisfy

binomial heap property.
– Each tree is min-heap ordered.
– 0 or 1 binomial trees of order k.

B0 B1 B4
Module Code and Module Title Title of Slides
 Binomial Heap
• Properties of N-node binomial heap
– Min key contained in root of B0, B1, …, Bk.
– Contains binomial tree Bi iff bi = 1 where bn b2b1b0 is binary representation of
N.
– At most log2 N + 1 binomial trees.
– Height  log2 N.

N = 19
binary = 10011
# trees = 3
height = 4

B0 B1 B4
Module Code and Module Title Title of Slides
 Binomial Heap: Implementation

• Represent trees using left-child, right

sibling pointers.
– 3 links per node (parent, left, right)
– Coding for variable no. of links per node
possible, but painful
• Root of trees connected with singly linked
list.
– Degree of trees strictly decreasing/increasing
from left to right.

Module Code and Module Title Title of Slides

 Binomial Heap: Implementation

Module Code and Module Title Title of Slides

 Merging

• Create heap H that is union of heaps H'

and H''
– Trivial if H' and H'' are binomial trees of the
same order.
• Connect roots of H' and H''
• Choose smaller key to be root of H
• If H' and H'' are of order k, then the merged heap
H is of order k+1.

Module Code and Module Title Title of Slides

 Merging

Binomial Heap

Leftist-power-of-2 Heap
Module Code and Module Title Title of Slides
 Merging

• If H' and H'' are binomial heaps consisting of

binomial trees with various orders…

Module Code and Module Title Title of Slides

 Merging

• …merge two binomial trees of the same order

one by one. Similar to binary addition.
– Running time = O (log N)

11 + 7 = 18

1111
1011
+ 0111
10010

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

Module Code and Module Title Title of Slides

 Merging

1111
1011
+ 0111
10010
H=
B4 B1

Module Code and Module Title Title of Slides

 Delete Min

• Delete node with minimum key in binomial

heap H.
– Find root x with min key in root list of H, and
delete
– H'  broken binomial trees
– H  Merge(H', H)
– Running time = O (log N)

Module Code and Module Title Title of Slides

 Delete Min

Module Code and Module Title Title of Slides

 Decrease Key

• Decrease key of node x in binomial heap H.

– Assume x is in binomial tree Bk.
– Bubble node x up the tree if x is too small.
– Running time = O (log n).

Module Code and Module Title Title of Slides

 Delete Node

• Delete node x in binomial heap H.

– Decrease key of x to - .
– Delete min.
– Running time = O (log N).

Module Code and Module Title Title of Slides

 Insert Node

• Insert a new node x into binomial heap H.

– H'  MakeHeap(x)
– H  Merge(H', H)

Module Code and Module Title Title of Slides

 Fibonacci Heap

• Collection of min-heap ordered trees.

• Structure of heap more flexible than binomial
trees.
– Every element can be in it’s own separate trees, or in
a single tree of depth n.
minimum

Module Code and Module Title Title of Slides

 Fibonacci Heap: Implementation

• Represent trees using left-child, right sibling

pointers and circular, doubly linked list.
– Easy to splice off sub-trees
• Roots of all trees linked with circular, doubly
linked list
– Easy to perform node deletion + insertion
• Each node maintains no. of children +
marked/unmarked state.
• Maintain pointer to root with minimum key value

Module Code and Module Title Title of Slides

 Find Minimum

• Pointer to minimum value -> finding it is

trivial
– Actual + amortized cost = constant

Module Code and Module Title Title of Slides

 Merging

• Concatenate both lists of roots to each

other.
– Actual + amortized cost = constant

Module Code and Module Title Title of Slides

 Insert Node

• Create new heap consisting of one node,

then merge.
– Actual + amortized cost = constant

Module Code and Module Title Title of Slides

 Delete Minimum

• Remove node with minimum key, then convert

children into roots of new trees.
– Amortized running time -> O(d) = O(log n).

Module Code and Module Title Title of Slides

 Delete Minimum

• Need to update pointer to root with minimum key…

– Possibly many roots that also act as single-node trees.
• …therefore, decrease number of roots.
– Link together roots of the same degree.
– Make one of them a child of the other, such that the one with the
smaller key remains the root.
– Repeat until every root has a different degree.
– Use array of length (log N) to keep pointers to each node of
each degree.
– Update array when linking roots.
• At most (log N) roots at the end.
• Amortized running time = O (log N).

Module Code and Module Title Title of Slides

 Delete Minimum

N = 8, log n = 3

a[ ] = 6 2 -

1st occurrences of roots with 0 No root of

and 1 degrees respectively degree 2 exists
yet

N = 8, log n = 3

a[ ] = 4 2 -

Only 1 root with 0

degrees exists now

Module Code and Module Title Title of Slides

 Delete Minimum

N = 8, log n = 3

a[ ] = 4 7 2

Module Code and Module Title Title of Slides

 Delete Minimum

• Check all roots and find minimum value

– Amortized running time = O (log N)
minimum

Overall amortized running time for deleting minimum

key = O (log N).

Module Code and Module Title Title of Slides

 Decrease Key

• Decrease key of node x in fibonacci heap H.

• If heap property is violated
– Cut node from parent
– If parent is not root and not marked, mark it.
– If parent is not root and marked, cut parent.
– Repeat process until root is reached or a parent is
marked instead of cut.
– Finally, unmark all roots, and point to minimum.

Module Code and Module Title Title of Slides

 Decrease Key

Assuming 4, 7 and 8 are already marked…

Original value of 9 decreased to 0.

Module Code and Module Title Title of Slides

 Decrease Key

Cut 0… Then cut 8…

Module Code and Module Title Title of Slides

 Decrease Key

Then cut 7…

Root has been reached, stop cutting.

Module Code and Module Title Title of Slides

 Decrease Key

Finally, unmark all roots and point to minimum.

minimum

Module Code and Module Title Title of Slides

 Delete Node

• Delete node x in binomial heap H.

– Decrease key of x to - .
– Delete min.
– Amortized running time = O (log N).

Module Code and Module Title Title of Slides

 Summary

• Heaps
– Binary Heaps
– Binomial Heaps
– Fibonacci Heaps

Module Code and Module Title Title of Slides

 Next Lesson

• Shortest Path algorithms

– Prims algorithm
– Kruskals algorithm
– Dijkshtras algorithm

Module Code and Module Title Title of Slides