T(n)=4T(n/2)+cn

=4[4T(n/4)+cn/2]+cn

=16T(n/4)+3cn

=16[4T(n/8)+cn/4]+3cn

=64T(n/8)+7cn

=64[4T(n/16)+cn/8]+7cn

=256T(n/16)+15cn

......................

......................

After k substitutions:

=4k+1T(n/2k+1)+(2k+1−1)cn

Putting k=log(n)

as after log(n)

steps the T(n/2k)

converges to T(1)

=4n2T(1/2)+(2n−1)cn

=n2.c′+n(2n−1)c
= O(n2)

4.3 The substitution method for solving

recurrences
4.3-1
Show that the solution of T(n)=T(n−1)+n is O(n2).

We guess (n)≤cn , 2

2,T(n)≤c(n−1) +n=cn2−2cn+c+n=cn +n(1−2c)+c≤cn ,

2 2 2

where the last step holds for c>1/2.

4.3-2
Show that the solution of T(n)=T(⌈n/2⌉)+1 O(lgn).

We guess �(�)≤�lg⁡(�−�)T(n)≤clg(n−a),

�(�)≤�lg⁡(⌈�/2⌉−�)+1≤�lg⁡((�+1)/2−�)+1=�lg⁡((�+1−2�)/
2)+1=�lg⁡(�+1−2�)−�lg⁡2+1(�≥1)≤�lg⁡(�+1−2�)
(�≥1)≤�lg⁡(�−�),T(n)≤clg(⌈n/2⌉−a)+1≤clg((n+1)/2−a)
+1=clg((n+1−2a)/2)+1=clg(n+1−2a)−clg2+1≤clg(n+1−2a)≤clg(n−a),
(c≥1)(a≥1)

4.3-3
We saw that the solution of �(�)=2�(⌊�/2⌋)+�T(n)=2T(⌊n/2⌋)
+n is �(�lg⁡�)O(nlgn). Show that the solution of this recurrence is
also Ω(�lg⁡�)Ω(nlgn). Conclude that the solution is Θ(�lg⁡�)Θ(nlgn).

First, we guess �(�)≤��lg⁡�T(n)≤cnlgn,

�(�)≤2�⌊�/2⌋lg⁡⌊�/2⌋+�≤��lg⁡(�/
2)+�=��lg⁡�−��lg⁡2+�=��lg⁡�+(1−�)�≤��lg⁡�,T(n)
≤2c⌊n/2⌋lg⌊n/2⌋+n≤cnlg(n/
2)+n=cnlgn−cnlg2+n=cnlgn+(1−c)n≤cnlgn,
where the last step holds for �≥1c≥1.

Next, we guess �(�)≥�(�+�)lg⁡(�+�)T(n)≥c(n+a)lg(n+a),

�(�)≥2�(⌊�/2⌋+�)(lg⁡(⌊�/2⌋+�)+�≥2�((�−1)/2+�)(lg⁡((�−1)/
2+�))+�=2��−1+2�2lg⁡�−1+2�2+�=�(�−1+2�)lg⁡(�−1+2�)
−�(�−1+2�)lg⁡2+�=�(�−1+2�)lg⁡(�−1+2�)
+(1−�)�−(2�−1)�(0≤�<1,�≥(2�−1)�1−�)≥�(�−1+2�)lg⁡(�−1
+2�)(�≥1)≥�(�+�)lg⁡(�+�),T(n)≥2c(⌊n/2⌋+a)(lg(⌊n/2⌋+a)
+n≥2c((n−1)/2+a)(lg((n−1)/2+a))+n=2c2n−1+2alg2n−1+2a
+n=c(n−1+2a)lg(n−1+2a)−c(n−1+2a)lg2+n=c(n−1+2a)lg(n−1+2a)
+(1−c)n−(2a−1)c≥c(n−1+2a)lg(n−1+2a)≥c(n+a)lg(n+a),
(0≤c<1,n≥1−c(2a−1)c)(a≥1)

4.3-4
Show that by making a different inductive hyptohesis, we can overcome the difficulty with
the boundary condition �(1)=1T(1)=1 for recurrence (4.19)(4.19) without adjusting the
boundary conditions for the inductive proof.

We guess �(�)≤�lg⁡�+�T(n)≤nlgn+n,

�(�)≤2(�⌊�/2⌋lg⁡⌊�/2⌋+⌊�/2⌋)+�≤2�(�/2)lg⁡(�/2)+2(�/
2)+�=��lg⁡(�/
2)+2�=��lg⁡�−��lg⁡2+2�=��lg⁡�+(2−�)�≤��lg⁡�+�,T(n)
≤2(c⌊n/2⌋lg⌊n/2⌋+⌊n/2⌋)+n≤2c(n/2)lg(n/2)+2(n/2)+n=cnlg(n/
2)+2n=cnlgn−cnlg2+2n=cnlgn+(2−c)n≤cnlgn+n,

where the last step holds for �≥1c≥1.

This time, the boundary condition is

�(1)=1≤��lg⁡�+�=0+1=1.T(1)=1≤cnlgn+n=0+1=1.

4.3-5
Show that Θ(�lg⁡�)Θ(nlgn) is the solution to the "exact" recurrence (4.3)(4.3) for
merge sort.

The recurrence is
�(�)=�(⌈�/2⌉)+�(⌊�/2⌋)+Θ(�)(4.3)T(n)=T(⌈n/2⌉)+T(⌊n/2⌋)+Θ(n)
(4.3)

To show ΘΘ bound, separately show �O and ΩΩ bounds.

 For �(�lg⁡�)O(nlgn), we
guess �(�)≤�(�−2)lg⁡(�−2)−2�T(n)≤c(n−2)lg(n−2)−2c,

2⌉−2)lg⁡(⌈�/2⌉−2)+�(⌊�/2⌋−2)lg⁡(⌊�/2⌋−2)+��≤�(�/2+1−2)lg⁡(�/
�(�)≤�(⌈�/

2+1−2)−2�+�(�/2−2)lg⁡(�/2−2)−2�+��≤�(�/2−1)lg⁡(�/
2−1)+�(�/2−1)lg⁡(�/
2−1)+��=��−22lg⁡�−22+��−22lg⁡�−22−4�+��=�(�−2)lg⁡�−
22−4�+��=�(�−2)lg⁡(�−2)−�(�−2)−4�+��=�(�−2)lg⁡(�−2)+
(�−�)�+2�−4�≤�(�−2)lg⁡(�−2)−2�,T(n)
≤c(⌈n/2⌉−2)lg(⌈n/2⌉−2)+c(⌊n/2⌋−2)lg(⌊n/2⌋−2)+dn≤c(n/2+1−2)lg(n/
2+1−2)−2c+c(n/2−2)lg(n/2−2)−2c+dn≤c(n/2−1)lg(n/2−1)+c(n/
2−1)lg(n/2−1)+dn=c2n−2lg2n−2+c2n−2lg2n−2−4c+dn=c(n−2)lg2n−2
−4c+dn=c(n−2)lg(n−2)−c(n−2)−4c+dn=c(n−2)lg(n−2)+
(d−c)n+2c−4c≤c(n−2)lg(n−2)−2c,

where the last step holds for �>�c>d.

 For Ω(�lg⁡�)Ω(nlgn), we
guess �(�)≥�(�+2)lg⁡(�+2)+2�T(n)≥c(n+2)lg(n+2)+2c,

2⌉+2)lg⁡(⌈�/2⌉+2)+�(⌊�/2⌋+2)lg⁡(⌊�/2⌋+2)+��≥�(�/2+2)lg⁡(�/
�(�)≥�(⌈�/

2+2)+2�+�(�/2−1+2)lg⁡(�/2−1+2)+2�+��≥�(�/2+1)lg⁡(�/
2+1)+�(�/2+1)lg⁡(�/
2+1)+4�+��≥��+22lg⁡�+22+��+22lg⁡�+22+4�+��=�(�+2)l
g⁡�+22+4�+��=�(�+2)lg⁡(�+2)−�(�+2)+4�+��=�(�+2)lg⁡(
�+2)+(�−�)�−2�+4�≥�(�+2)lg⁡(�+2)+2�,T(n)
≥c(⌈n/2⌉+2)lg(⌈n/2⌉+2)+c(⌊n/2⌋+2)lg(⌊n/2⌋+2)+dn≥c(n/2+2)lg(n/
2+2)+2c+c(n/2−1+2)lg(n/2−1+2)+2c+dn≥c(n/2+1)lg(n/2+1)+c(n/
2+1)lg(n/2+1)+4c+dn≥c2n+2lg2n+2+c2n+2lg2n+2
+4c+dn=c(n+2)lg2n+2
+4c+dn=c(n+2)lg(n+2)−c(n+2)+4c+dn=c(n+2)lg(n+2)+
(d−c)n−2c+4c≥c(n+2)lg(n+2)+2c,

where the last step holds for �>�d>c.

4.3-6
Show that the solution
to �(�)=2�(⌊�/2⌋+17)+�T(n)=2T(⌊n/2⌋+17)+n is �(�lg⁡�)O(nlgn).

We guess �(�)≤�(�−�)lg⁡(�−�)T(n)≤c(n−a)lg(n−a),

�(�)≤2�(⌊�/2⌋+17−�)lg⁡(⌊�/2⌋+17−�)+�≤2�(�/2+17−�)lg⁡(�/
2+17−�)
+�=�(�+34−2�)lg⁡�+34−2�2+�=�(�+34−2�)lg⁡(�+34−2�)
−�(�+34−2�)+�(�>1,�>�0=�(�))≤�(�+34−2�)lg⁡(�+34−2�)
(�≥34)≤�(�−�)lg⁡(�−�).T(n)≤2c(⌊n/2⌋+17−a)lg(⌊n/2⌋+17−a)
+n≤2c(n/2+17−a)lg(n/2+17−a)+n=c(n+34−2a)lg2n+34−2a
+n=c(n+34−2a)lg(n+34−2a)−c(n+34−2a)
+n≤c(n+34−2a)lg(n+34−2a)≤c(n−a)lg(n−a).(c>1,n>n0=f(a))(a≥34)

4.3-7
Using the master method in Section 4.5, you can show that the solution to the
recurrence �(�)=4�(�/3)+�T(n)=4T(n/3)+n is �(�)=Θ(�log⁡34)T(n)=
Θ(nlog34). Show that a substitution proof with the
assumption �(�)≤��log⁡34T(n)≤cnlog34 fails. Then show how to subtract off a
lower-order term to make the substitution proof work.

We guess �(�)≤��log⁡34T(n)≤cnlog34 first,

�(�)≤4�(�/3)log⁡34+�=��log⁡34+�.T(n)≤4c(n/3)log34+n=cnlog34+n.
We stuck here.

We guess �(�)≤��log⁡34−��T(n)≤cnlog34−dn again,

�(�)≤4(�(�/3)log⁡34−��/3)+�=4(��log⁡34/4−��/
3)+�=��log⁡34−43��+�≤��log⁡34−��,T(n)≤4(c(n/3)log34−dn/
3)+n=4(cnlog34/4−dn/3)+n=cnlog34−34dn+n≤cnlog34−dn,

where the last step holds for �≥3d≥3.

4.3-8
Using the master method in Section 4.5, you can show that the solution to the
recurrence �(�)=4�(�/2)+�T(n)=4T(n/2)+n is �(�)=Θ(�2)T(n)=Θ(n2
). Show that a substitution proof with the assumption �(�)≤��2T(n)≤cn2 fails.
Then show how to subtract off a lower-order term to make the substitution proof work.

First, let's try the guess �(�)≤��2T(n)≤cn2. Then, we have

�(�)=4�(�/2)+�≤4�(�/2)2+�=��2+�.T(n)=4T(n/2)+n≤4c(n/
2)2+n=cn2+n.
We can't proceed any further from the inequality above to
conclude �(�)≤��2T(n)≤cn2.

Alternatively, let us try the guess

�(�)≤��2−��,T(n)≤cn2−cn,
which subtracts off a lower-order term. Now we have

�(�)=4�(�/2)+�=4(�(�/2)2−�(�/2))+�=4�(�/2)2−4�(�/
2)+�=��2+(1−2�)�≤��2,T(n)=4T(n/2)+n=4(c(n/2)2−c(n/2))
+n=4c(n/2)2−4c(n/2)+n=cn2+(1−2c)n≤cn2,

where the last step holds for �≥1/2c≥1/2.

4.3-9
Solve the recurrence �(�)=3�(�)+log⁡�T(n)=3T(n)+logn by making a change
of variables. Your solution should be asymptotically tight. Do not worry about whether
values are integral.

First,

�(�)=3�(�)+lg⁡� let �=lg⁡��(2�)=3�(2�/2)+��(�)=3�(�/

2)+�.T(n)T(2m)S(m)=3T(n)+lgn=3T(2m/2)+m=3S(m/2)+m. let m=lgn

Now we guess �(�)≤��lg⁡3+��S(m)≤cmlg3+dm,

�(�)≤3(�(�/2)lg⁡3+�(�/2))
+�≤��lg⁡3+(32�+1)�(�≤−2)≤��lg⁡3+��.S(m)≤3(c(m/
2)lg3+d(m/2))+m≤cmlg3+(23d+1)m≤cmlg3+dm.(d≤−2)

Then we guess �(�)≥��lg⁡3+��S(m)≥cmlg3+dm,

�(�)≥3(�(�/2)lg⁡3+�(�/2))
+�≥��lg⁡3+(32�+1)�(�≥−2)≥��lg⁡3+��.S(m)≥3(c(m/
2)lg3+d(m/2))+m≥cmlg3+(23d+1)m≥cmlg3+dm.(d≥−2)
Thus,

�(�)=Θ(�lg⁡3)�(�)=Θ(lg⁡lg⁡3�).S(m)T(n)=Θ(mlg3)=Θ(lglg3n).

Solve the following recurrence relation using recursion tree method-

T(n) = 2T(n/2) + n

The given recurrence relation shows-

 The cost of dividing a problem of size n into its 2 sub-problems and then
combining its solution is n.
 The cost of dividing a problem of size n/2 into its 2 sub-problems and
then combining its solution is n/2 and so on.

This is illustrated through following recursion tree where each node represents
the cost of the corresponding sub-problem-
Step-02:

Determine cost of each level-

 Cost of level-0 = n
 Cost of level-1 = n/2 + n/2 = n
 Cost of level-2 = n/4 + n/4 + n/4 + n/4 = n and so on.

Step-03:
Determine total number of levels in the recursion tree-
 Size of sub-problem at level-0 = n/20
1
 Size of sub-problem at level-1 = n/2
2
 Size of sub-problem at level-2 = n/2

Continuing in similar manner, we have-

Size of sub-problem at level-i = n/2i

Suppose at level-x (last level), size of sub-problem becomes 1. Then-

n / 2x = 1
2x = n
Taking log on both sides, we get-
xlog2 = logn
x = log2n

∴ Total number of levels in the recursion tree = log2n + 1

Step-04:

Determine number of nodes in the last level-

 Level-0 has 20 nodes i.e. 1 node
1
 Level-1 has 2 nodes i.e. 2 nodes
2
 Level-2 has 2 nodes i.e. 4 nodes

Continuing in similar manner, we have-

Level-log2n has 2log2n nodes i.e. n nodes

Step-05:

Determine cost of last level-

Cost of last level = n x T(1) = θ(n)
Step-06:

Add costs of all the levels of the recursion tree and simplify the expression so
obtained in terms of asymptotic notation-
= n x log2n + θ (n)
= nlog2n + θ (n)
= θ (nlog2n)

Problem-02:

Solve the following recurrence relation using recursion tree method-

T(n) = T(n/5) + T(4n/5) + n

Solution-

Step-01:

Draw a recursion tree based on the given recurrence relation.

The given recurrence relation shows-
 A problem of size n will get divided into 2 sub-problems- one of size n/5
and another of size 4n/5.
 Then, sub-problem of size n/5 will get divided into 2 sub-problems- one
of size n/52 and another of size 4n/52.
 On the other side, sub-problem of size 4n/5 will get divided into 2 sub-
problems- one of size 4n/52 and another of size 42n/52 and so on.
 At the bottom most layer, the size of sub-problems will reduce to 1.

This is illustrated through following recursion tree-

The given recurrence relation shows-

 The cost of dividing a problem of size n into its 2 sub-problems and then
combining its solution is n.
 The cost of dividing a problem of size n/5 into its 2 sub-problems and
then combining its solution is n/5.
 The cost of dividing a problem of size 4n/5 into its 2 sub-problems and
then combining its solution is 4n/5 and so on.

This is illustrated through following recursion tree where each node represents
the cost of the corresponding sub-problem-

Step-02:

Determine cost of each level-

 Cost of level-0 = n
 Cost of level-1 = n/5 + 4n/5 = n
2 2 2 2 2
 Cost of level-2 = n/5 + 4n/5 + 4n/5 + 4 n/5 = n

Step-03:
Determine total number of levels in the recursion tree. We will consider the
rightmost sub tree as it goes down to the deepest level-
  Size of sub-problem at level-0 = (4/5)0n
1
 Size of sub-problem at level-1 =(4/5) n
2
 Size of sub-problem at level-2 =(4/5) n

Continuing in similar manner, we have-

Size of sub-problem at level-i = (4/5)in
Suppose at level-x (last level), size of sub-problem becomes 1. Then-
(4/5)xn = 1
(4/5)x = 1/n
Taking log on both sides, we get-
xlog(4/5) = log(1/n)
x = log5/4n

∴ Total number of levels in the recursion tree = log5/4n + 1

Step-04:

Determine number of nodes in the last level-

 Level-0 has 20 nodes i.e. 1 node
1
 Level-1 has 2 nodes i.e. 2 nodes
2
 Level-2 has 2 nodes i.e. 4 nodes

Continuing in similar manner, we have-

Level-log5/4n has 2log5/4n nodes

Step-05:

Determine cost of last level-

Cost of last level = 2log5/4n x T(1) = θ(2log5/4n) = θ(nlog5/42)
Step-06:

Add costs of all the levels of the recursion tree and simplify the expression so
obtained in terms of asymptotic notation-

= nlog5/4n + θ(nlog5/42)
= θ(nlog5/4n)

Another example
T(n)=2T(sqrt(n))+1
T(1)=1

You are looking at this recurrence relation: lg lg n

Suppose T is a solution. Then we have T(1)=2T(1)+1=2+1, so =3. That is,

this recurrence relation has no solution.
x x
To solve 2x √ n≈1, note that this is equivalent to exp((logn)/2 )≈1, thus log n ≈ 2 ≈
2
and x≈log log n

exp((logn)/2x)=1, which has no finite solution x if n>1. Hence the idea of the method
is to estimate the number x of steps needed to make 2x√ n smaller than any finite fixed
level >1, say, the smallest integer x such that n2x√ n <2. Rigorously speaking, this is
the smallest x such that 2x>log2n, thus x is indeed of order loglogn
x

Problem-03:

Solve the following recurrence relation using recursion tree method-

T(n) = 3T(n/4) + cn2

Solution-

Step-01:

Draw a recursion tree based on the given recurrence relation-

 (Here, we have directly drawn a recursion tree representing the cost of sub
problems)

Step-02:

Determine cost of each level-

 Cost of level-0 = cn2
2 2 2 2
 Cost of level-1 = c(n/4) + c(n/4) + c(n/4) = (3/16)cn
2 2 2
 Cost of level-2 = c(n/16) x 9 = (9/16 )cn

Step-03:

Determine total number of levels in the recursion tree-

 Size of sub-problem at level-0 = n/40
1
 Size of sub-problem at level-1 = n/4
2
 Size of sub-problem at level-2 = n/4

Continuing in similar manner, we have-

Size of sub-problem at level-i = n/4i
Suppose at level-x (last level), size of sub-problem becomes 1. Then-
n/4x = 1
4x = n
Taking log on both sides, we get-
xlog4 = logn
x = log4n

∴ Total number of levels in the recursion tree = log4n + 1

Step-04:

Determine number of nodes in the last level-

  Level-0 has 30 nodes i.e. 1 node
1
 Level-1 has 3 nodes i.e. 3 nodes
2
 Level-2 has 3 nodes i.e. 9 nodes

Continuing in similar manner, we have-

Level-log4n has 3log4n nodes i.e. nlog43 nodes

Step-05:

Determine cost of last level-

Cost of last level = nlog43 x T(1) = θ(nlog43)

Step-06:

Add costs of all the levels of the recursion tree and simplify the expression so
obtained in terms of asymptotic notation-

= cn2 { 1 + (3/16) + (3/16)2 + ……… } + θ(nlog43)

Now, { 1 + (3/16) + (3/16)2 + ……… } forms an infinite Geometric progression.

On solving, we get-
= (16/13)cn2 { 1 – (3/16)log4n } + θ(nlog43)
= (16/13)cn2 – (16/13)cn2 (3/16)log4n + θ(nlog43)
= O(n2)
Iteration method

Merge sort
Quick sort
Binary search