Algorithms Section (6)

Section Content:
Recurrence relation

Recurrence relation:
Ex: Fibonacci
0 1 2 3 4 5 6

0 1 1 2 3 5 8

int fib(n){
if(n==0 || n==1){
return n;
}
Else{
Returm fib(n-1)+fib(n-2)
}
}

Fib (4)=1+0+1+1+0=3

Eng. Rania Ahmed Eng. Sohaila Ahmed Eng. Nadeen Qadry Eng. Mostafa Sayed
Notes:
Assume that time taken to calculate fib of n is T(n)
We make assumption that each single operation take 1 unite
T(n)= T(n-1) + T(n-2) + C
where C is the constant that represent the unit of time 4
T(n-1) ≅ T(n-2)
T(n-1) & T(n+1) is very close in value, which are approximately equal so, we
have two choices:
 T(n-1)  Upper bound
 T(n-2)  Lower bound

Example 1:
Assume that time taken to calculation fib(n) is T(n), T(n)= T(n-1) + T(n-2) + C,
where C is constant find the big O for the algorithm.
Answer:
Lower case: T(n-2)
T(n)= 2T(n-2) +C K=1  equ (1)
T(n-2) =2T((n-2)-2) +C
T(n) =2(2T(n-4) +C) +C
T(n) =4T(n-4) +2C+C
T(n)=4T(n-4) +3C K=2 (substituted T(n-4) in equ (1))
T(n-4) =2T((n-4) -2) +C
T(n) =4(2T(n-6) +C) +3C
T(n) =8T(n-6) +4C+3C
T(n)=8T(n-6) +7C K=3 (substituted T(n-4) in equ (1))
T(n-6) =2T((n-6)-2) +C
T(n) =8(2T(n-8) +C) +7C
T(n) =16T(n-8) +8C+7C
T(n) =16T(n-8) +15C K=4
General form:
T(n)= 𝟐𝒌 𝑻(𝒏 − 𝟐𝒌) + (𝟐𝒌 − 𝟏)𝑪
n-2k=0  n=2k  k=n/2 substitute in T(n)
T(n)=2(n/2) T(n-(2(n/2))) +( 2(n/2) -1) C
T(n)=2(n/2) T(n-n) +( 2(n/2) -1) C
=2(n/2) T (0) +( 2(n/2) -1) C
= (2(n/2) -1) C
Big O = O ( 2(n/2) )

Eng. Rania Ahmed Eng. Sohaila Ahmed Eng. Nadeen Qadry Eng. Mostafa Sayed
Example2:
𝟎 𝒊𝒇 𝒏 = 𝟏

T(n)=
𝑛
𝑇( )+1 𝒊𝒇 𝒏 > 𝟏
2
Answer:
T (1) =0
𝒏
T(n)= T ( )+1 K=1  equ (1)
𝟐
𝑛
𝑛 2
T ( )=T ( )+1
2 2
𝑛
T(n)= T ( )+1+1
4
𝒏
T(n)= T ( )+2 K=2 (substituted T(n/4) in equ (1))
𝟒
𝑛
𝑛 4
T ( )= T ( )+1
4 2
𝑛
T(n)= T ( )+2+1
8
𝒏
T(n)= T ( )+3 K=3 (substituted T(n/8) in equ (1))
𝟖
𝑛
𝑛 8
T ( )= T ( )+1
8 2
𝑛
T(n)= T ( )+1+3
16
𝐧
T(n)= T ( )+4 K=4
𝟏𝟔
General form:
n
T(n)= T ( K)+K
2
n
=1  n=2k  log 2 2k =log 2 n  k=log 2 n
2k
n
T(n)=T ( )+log 2 n
2log2 n
n
= T ( ) +log 2 n
n
= T (1) + log 2 n (where T(1) = 0)
= 0+log 2 n=log 2 n
Big O= O (𝐥𝐨𝐠 𝟐 𝒏)
Remember:
log 𝑎 𝑎 = 1
log 𝑎 𝑎𝑏 = 𝑏 log 𝑎 𝑎 = 𝑏
𝑎log𝑎 𝑏 = 𝑏

Eng. Rania Ahmed Eng. Sohaila Ahmed Eng. Nadeen Qadry Eng. Mostafa Sayed
Example 3:

𝟏 𝐢𝐟 𝐧 = 𝟏
T(n)=

𝐧
𝟐𝐓 ( ) + 𝐧 𝐢𝐟 𝐧 > 𝟏
𝟐
Answer:
T (1) =1
𝐧
T(n)= 2T ( )+n K=1  equ (1)
𝟐
n
n 2 n
T ( )=2T ( )+
2 2 2
n n
T(n)= 2(2T ( )+ )+n
4 2
𝐧
T(n)= 4T ( )+2n K=2 (substituted T(n/4) in equ (1))
𝟒
n
n 4 n
T ( )= 2T ( )+
4 2 4
n n
T(n)= 4(2T ( )+ )+2n
8 4
𝐧
T(n)= 8T ( )+3n K=3 (substituted T(n/8) in equ (1))
𝟖
n
n 8 n
T ( )=2 T ( )+
8 2 8
n n
T(n)= 8(2T ( )+ )+3n
16 28
𝐧
T(n)= 16T ( )+4n K=4
𝟏𝟔
General form:
𝑛
T(n)= 2𝑘 T ( 𝐾)+Kn
2
𝑛
=1  n=2𝐾  log 2 2𝑘 = log 2 𝑛  k = log 2 𝑛
2𝐾
𝑛
T(n)= 2log2 𝑛 T ( )+nlog 2 𝑛
2log2 𝑛
𝑛
=n T ( )+ nlog 2 𝑛
𝑛
=n(1)+ nlog 2 𝑛
= n+𝑛 log 2 𝑛
Big O= O (𝐧𝐥𝐨𝐠 𝟐 𝒏)

Eng. Rania Ahmed Eng. Sohaila Ahmed Eng. Nadeen Qadry Eng. Mostafa Sayed
 Assignment
Example:
Assume that time taken to calculation fib(n) is T(n), T(n)= T(n-1) +
T(n-2) + C, where C is constant find the big O for the algorithm use
upper case.

Eng. Rania Ahmed Eng. Sohaila Ahmed Eng. Nadeen Qadry Eng. Mostafa Sayed