Scribd
Upload
95 views

Design and Analysis of Algorithms: Recursion

1) Recursion is a process of defining objects or calculations in terms of simpler instances of the same object or calculation. 2) A recurrence relation expresses a problem in terms of smaller instances of the same problem plus additional work. It is used to analyze recursive problems. 3) The master theorem provides a way to solve recurrence relations of the form T(n) = aT(n/b) + f(n) where a >= 1, b > 1, and determining the runtime of T(n).

Uploaded by

Talha Khalid
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
95 views

Design and Analysis of Algorithms: Recursion

1) Recursion is a process of defining objects or calculations in terms of simpler instances of the same object or calculation. 2) A recurrence relation expresses a problem in terms of smaller instances of the same problem plus additional work. It is used to analyze recursive problems. 3) The master theorem provides a way to solve recurrence relations of the form T(n) = aT(n/b) + f(n) where a >= 1, b > 1, and determining the runtime of T(n).

Uploaded by

Talha Khalid
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 20

DESIGN AND ANALYSIS OF

ALGORITHMS

Lecture 7

Recursion
Recursive Definition
 To define objects/processes in terms of simpler
objects/processes
In simple words converting a problem into similar

problem ,but with reduced size.

2
Recurrence Relation
 To analyse a recursive problem we need to write
recurrence relation.
 A formula which represent a bigger problem into
smaller problem.

But..How to write a recurrence relation?


Analogy!
 Problem: To eat an apple.
 Problem size: n grams
 P(n) could be expressed as:
 P(n)=P(n/2)+P(n/2)
n
 Stopping/terminating condition: remaining
size1gram

Note: Without terminating condition a problem


n/2 n/2
remain unsolved.
Problem 1:
How many binary sequences possible of size n?

 Goal: Use recursive relation to express the problem.

1 2 3 …… n
 Total work= Fill all places with general sequence.
 How to fill first position? Either: 0|1

0
n-1
1 2 3 …… n

1
n-1
1 2 3 …… n
Problem 1:
How many binary sequences possible of size n?

 T(n)=Binary sequence of length n.


 T(2)= Binary sequence of length 2.

 T(3)= Binary sequence of length 3.

 T(4)= Binary sequence of length 4.

 T(k)= Binary sequence of length k.

So far,
 T(n)=T(n-1)+T(n-1)

 T(n)=2T(n-1)

What about the terminating condition?


Problem 1:
How many binary sequences possible of size n?

 Base Case/Terminating condition:


 T(1)=2 ,we could fill either 0 or 1
 This factor should be considered in the solution:

T(n)=2T(n-1) , n>1
T(1)=2 recurrence relation

Provided the base condition ,we could now solve the


rest of the sub problems till size n.
Example – Factorial program
Using Iteration Using Recursion
int Fact(int n)
int Fact(int n)
{
int i=1; int fac=1; {
for(i=1;i=n;i++) if(n==1 || n==0)
{ return 1;
fac=fac*i;
} Else
return fac; return n*fact(n-1);
} }
More Examples of Recursion
Solving Recurrence Relations

 Back Substitution Method


 The Recursion Tree

 The Master Theorem


Recurrence (Back Substitution Method)

A(){ T (n)  1  T (n  1) 1
if(n>1)
return(A(n-1)) T (n  1)  1  T (n  2) 2
} T (n  2)  1  T (n  3) 3

Substitute eq(2) in eq(1) for kth value


 k  T (n  k )
T (n)  1  1  T (n  2) Stopping condition
 2  T (n  2) T (n)  1  T (n  1); n  1
So k=n-1, substitute the value of k
now substitute eq(3)
 (n  1)  T (n  (n  1))
 2  1  T (n  3)  (n  1)  T (1)  n  1  1
 3  T (n  3) n O(n)
Recurrence (Back Substitution Method)
Let’s consider a recursive equation
T (n)  n  T (n  1) 1
T (n)  n  T (n  1); n  1 T (n  1)  (n  1)  T (n  2) 2
T (1)  1 ;n 1 T (n  2)  (n  2)  T (n  3) 3

Substitute eq(2) in eq(1) Stopping condition


T (n)  n  (n  1)  T (n  2) n  (k  1)  1
now substitute eq(3) in above equation n  k 1  1
T (n)  n  (n  1)  (n  2)  T (n  3) k  n2
for kth value
T (n)  n  (n  1)  (n  2)  T (n  3) .. .. ... T (n  k )  T (n  (k  1))
Recurrence (cont)
T (n)  n  (n  1)  (n  2)  T (n  3) .. .. ... T (n  k )  T (n  (k  1))
So k=n-2, substitute the value of k in above equation

T (n)  n  (n  1)  (n  2)  T (n  3) .. .. ... T (n  (n  2))  T (n  (n  2  1))


T (n)  n  (n  1)  (n  2)  T (n  3) .. .. ... 2  1
n(n  1)
T (n) 
2

O(n2)
Recurrence (Master Theorem)
Recurrence (Master Theorem)
T (n)  aT ( )   (n log n)
n k p
b
where a  1, b  1, k  0 and p is real number
1) if a  b , then T (n)   (n
k logba
)

2) if a  b
k

a) if p  1, then T (n)   (n log p 1 n)


logba

b) if p  1, then T (n)   (n logba


log log n)
c) if p  1, then T (n)   (n logba
)
3) if a  bk
a) if p  0, then T (n)   (n log n)
k p

b) if p  0, then T (n)   (nk )


Recurrence (Master Theorem)
Example 1

T (n)  3T ( n )  n2
2
a  3, b  2, k  2 and p  0
Compare a with bk
a bk
3 22
So
3 4
a < bk
Because a<bk , condition 3 of Master theorem is true

As we have p=0 Therefore it falls in the condition 3(a)


So, T (n)   (nk log p n)   (n 2 )
Recurrence (Master Theorem)
Example 2

T (n)  4T ( n )  n2
2
a  4, b  2, k  2 and p  0
Compare a with bk
a bk
4 22
So
4 4
a = bk
Because a=bk , condition 2 of Master theorem is true

As we have p=0 Therefore it falls in the condition 2(a)


So,T (n)   (nlog2 log01 n)   (n log n)
4
2
Recurrence (Master Theorem)
Example 3

T ( n)  T ( n )  n 2
2
a  1, b  2, k  2 and p  0
Compare a with bk
a bk
1 22
So
1 4
a < bk
Because a<bk , condition 3 of Master theorem is true

As we have p=0 Therefore it falls in the condition 3(a)


So, T (n)   (n2 log 0 n)   (n )2
Recurrence (Master Theorem)
Example 4

T (n)  2T ( n )  n log n
2
a  2, b  2, k  1 and p  1
Compare a with bk
a bk
2 21
So
2 2
a = bk
Because a=bk , condition 2 of Master theorem is true

As we have p=1 Therefore it falls in the condition 2(a)


So, T (n)   (n log 2 n)   (n log n) 2
Recurrence (Master Theorem)
Example 5
T (n)  2T ( n )  n
2 log n
 2T ( n )  n log 1 n
2
a  2, b  2, k  1 and p  1
Compare a with bk
2 2
So
a = bk
Because a=bk , condition 2 of Master theorem is true

As we have p=-1 Therefore it falls in the condition 2(b)


So, T (n)   (n log log n)   (n log log n)

You might also like