Recurrence Relations

Generating Functions of Sequences

Sequences
A = {ar}, r = 0 .. .
Examples:

1. A = {ar}, r = 0 .. , where ar = 2r.

= {1, 2, 4, 8, 16, …, 2r, …}
2. B = {br}, r = 0 .. , where
br = 0, if 0  r  4
= 2, if 5  r  9
= 3, if r = 10
= 4, if 11  r
= {0, 0, 0, 0, 0, 2, 2, 2, 2, 2, 3, 4, 4, …}
3. C = {cr}, r = 0 .. , where cr = r + 1.
= {1, 2, 3, 4, 5, …}

4. D = {dr}, r = 0 .. , where dr = r2.

= {0, 1, 4, 9, 16, 25, …}
Generating function for the sequence A = {ar}, r = 0 .. .

A(X) = a0 + a1X + a2 X2 + … + anXn + …

=  ar Xr, r = 0 .. .

Examples:

1. Generating function for the sequence A = {a r}, r = 0 .. ,

where ar = 2r.
A(X) = 1 + 2X + 4X2 + … + 2n Xn + …
=  2r Xr, r = 0 .. .
2. Generating function for the sequence B = {b r}, r = 0 .. ,
where
br = 0, if 0  r  4
= 2, if 5  r  9
= 3, if r = 10
= 4, if 11  r
B(X) = 2X5 + 2X6 + 2X7 + 2X8 + 2X9 + 3X10 + 4X11 +
4X12 + … + 4Xn + …

3. Generating function for the sequence C = {c r}, r = 0 .. ,

where cr = r + 1.
C(X) = 1 + 2X + 3X2 + … + (n+1)Xn + …
=  (r+1)Xr, r = 0 .. .

4. Generating function for the sequence D = {d r}, r = 0 .. ,

where dr = r2.
D(X) = X + 4X2 + 9X3+ 16X4 + 25X5 + … + n2 Xn + …
=  r2 Xr, r = 0 .. .
Definitions

Let the Generating Functions / Formal Power Series be

A(X) =  ar Xr, r = 0 .. .
and B(X) =  bs Xs, s = 0 .. .

1. Equality
A(X) = B(X), iff an = bn for each n  0.

2. Multiplication by a scalar number C

C A(X) =  (C ar) Xr, r = 0 .. .

3. Sum
A(X) + B(X) =  (an + bn) Xr, r = 0 .. .

4. Product

A(X) B(X) =  Pn Xn, n = 0 .. ,

where Pn = j+k-n aj bk.
Exercises:
1. Find a Generating function for the sequence
A = {ar}, r = 0 .. , where
ar = 1, if 0  r  2
= 3, if 3  r  5
= 0, if r 
A(X) = 1 + X + X2 + 3X3 + 3X4 + 3X5

2. Build a generating function for ar = no. of integral solutions to the

equation e1 + e2 + e3 = r, if 0  ei  3 for each i.
A(X) = (1 + X + X2 + X3)3
3. Write a generating function for ar = no. of ways of selecting r balls from 3
red balls, 5 blue balls, and 7 white balls.
A(X) = (1 + X + X2 + X3) (1 + X + X2 + X3 + X4+ X5)
(1 + X + X2 + X3 + X4 + X5 + X6 + X7)
4. Find the coefficient of X23 in (1 + X5 + X9)10.
e1 + e2 + … + e10 = 23 where ei = 0, 5, 9.
1 x 5 + 2 x 9 + 7 x 0 = 23
Coefficient of X23 = 10! / (1! 2! 7!)
= 10 . 9 . 8 / (2)
= 10. 9 . 4
= 360
5. Find the coefficient of X32 in (1 + X5 + X9)10.
e1 + e2 + … + e10 = 32 where ei = 0, 5, 9.
1 x 5 + 3 x 9 + 6 x 0 = 32
Coefficient of X32 = 10! / (1! 3! 6!)
= 10 . 9 . 8 . 7/ (3 . 2)
= 10. 3 . 4 . 7
= 840
6. Find a Generating function for the no. of r-combinations of {3.a, 5.b, 2.c}.
A(X) = (1 + X + X2 + X3) (1 + X + X2 + X3 + X4+ X5 )
(1 + X + X2)
Calculating Coefficient of generating function

If A(X) =  ar Xr, r = 0 .. , then A(X) is said to have a multiplicative inverse

if there is B(X) =  bk Xk, k = 0 .. 
such that A(X) B(X) = 1.

a0 b 0 = 1
a1 b 0 + a0 b1 = 0
a0 b 2 + a1 b1 + a2 b0 = 0
a0 b 3 + a1 b2 + a 2 b1 + a3 b0 = 0
…
a0 b 2 + a0 b2 + … + a 0 b2 = 0
…
b0 = 1 / a0
b1 = –a1 b0 / a0
b2 = –a1 b1 – a2 b0 / a0
…
Geometric Series

A(X) = 1 – X
a0 = 1, a1 = –1.
b0 = 1 / a 0 = 1
b1 = –a1 b0 / a0 = –(–1) (1) / (1) = 1
b2 = –a1 b1 – a2 b0 / a0 = –(–1) (1) – (0) (1) / (1) = 1
…
bi = 1
1 / (1 – X) =  Xr, r = 0 .. .
Replace X by aX, where a is a real no.

1 / (1 – aX) =  ar Xr, r = 0 .. .

Let a = –1
1 / (1 + X) =  (–1)r Xr, r = 0 .. .
1 / (1 + aX) =  (–1)r ar Xr, r = 0 .. .
1 / (1 – X)n = ( Xk) n, k = 0 .. .
=  C(n – 1 + r, r) Xr, r = 0 .. .
1 / (1 + X)n = ( (–1)r Xk) n, k = 0 .. .

=  C(n – 1 + r, r) (–1)r Xr, r = 0 .. .

1 / (1 – aX)n = ( ar Xk ) n, k = 0 .. .

=  C(n – 1 + r, r) ar Xr, r = 0 .. .

1 / (1 – Xk) =  Xkr, k = 0 .. .

1 / (1 + Xk) =  (–1)r Xkr, k = 0 .. .

1 / (a – X) = (1 / a)  Xr / ar, r = 0 .. .

1 / (X – a) = (–1 / a)  Xr / ar, r = 0 .. .

1 / (X + a) = (1 / a)  Xr / ((–1)r ar), r = 0 .. .

1 + X + X2 + … + Xn = (1 – Xn+1 ) / (1 – x)
Special Cases of Binomial Theorem

(1 + X)n = 1 + C(n, 1) X + C(n, 2) X2 + … + C(n, n) Xn

(1 + Xk) n = 1 + C(n, 1) Xk + C(n, 2) X2k + … + C(n, n) Xnk

(1 – X) = 1 – C(n, 1) X + C(n, 2) X2 + … + (–1)n C(n, n) Xn

(1 – Xk) n = 1 – C(n, 1) Xk + C(n, 2) X2k + … + (–1)n C(n, n) Xnk

Examples:

1. Calculate A(X) =  ar Xr, r = 0 ..  = 1 / (X2 – 5X + 6).

(X2 – 5X + 6) = (X – 3) (X – 2)
1 / (X2 – 5X + 6) = A / (X – 3) + B / (X – 2)
 A(X – 2) + B(X – 3) = 1
Let X = 2, Then B = –1
Let X = 3, Then A = 1
1 / (X2 – 5X + 6)
= 1 / (X – 3) – 1 / (X – 2)
= (–1 / 3)  Xr / 3r – (–1 / 2)  Xr / 2r, r = 0 .. .
= (–1 / 3)  Xr / 3r + (1 / 2)  Xr / 2r, r = 0 .. .
2. Compute the coefficients of A(X) =  ar Xr, r = 0 .. 
= (X2 – 5X + 3) / (X4 – 5X2 + 4).

(X4 – 5X2 + 4) = (X2 – 1) (X2 – 4)

= (X – 1) (X + 1) (X – 2) (X + 2)
(X2 – 5X + 3) / (X4 – 5 X2 + 4)
= A / (X – 1) + B / (X + 1) + C / (X – 2) + D / (X + 2)
(X2 – 5X + 3) = A (X + 1) (X – 2) (X + 2)
+ B (X – 1) (X – 2) (X + 2)
+ C (X – 1) (X + 1) (X + 2)
+ D (X – 1) (X + 1) (X – 2)
For X = 1, A = 1 / 6
For X = –1, B = 3 / 2
For X = 2, C = –1 / 4
For X = –2, D = –17 / 12
(X2 – 5X + 3) / (X4 – 5 X2 + 4)
= 1/(6(X – 1)) + 3/(2(X + 1)) – 1/(4(X – 2)) – 17/(12(X + 2))
= (–1/6)Xr + 3/2(–1) r Xr – 1/4(–1/2) Xr /2r
17/12(1/2) Xr /((–1)r 2r), r = 0 .. 
=  [(–1/6) + 3/2(–1)r + 1/8(1/ 2r) – 17/24(–1)r / 2r)] Xr, r = 0 .. 
3. Find the coefficient of X20 in (X3 + X4 + X5 + …)5.
(X3 + X4 + X5 + …)5
= [X3 (1 + X + X2 + …)]5
= X15 (Xr)5, r = 0 .. 
= X15 C(5 – 1 + r, r) Xr, r = 0 .. 
= X15 C(4 + r, r) Xr, r = 0 .. 
Coefficient of X20 in (X3 + X4 + X5 + …)5
= Coefficient of X5 in C(4 + r, r) Xr, r = 0 .. 
r=5
C(4 + r, r)
= C(9, 5)
= 9! / (5! 4!)
= 9 . 8 . 7 . 6 / (4 . 3 . 2)
=9.7.2
= 126
Recurrence relations

Recurrence relation
Formula that relates for any integer n ≥ 1, the nth term of a sequence
A = {ar}, r = 0 ..  to one or more of the terms a0, a1, …, an-1.

Examples
an + 5an-1 + 6an-2 = 0.
an + 5an-1 + 6an-2 = 3n2 – 2n + 1.
Linear recurrence relation
A recurrence relation of the form
c0(n) + c1(n)an + … + ck(n) an-k = f(n) for n ≥ k,
where c0(n), c1(n), …, ck(n), and f(n) are functions of n.
Example
an – (n – 1)an-1 – (n – 1)an-2 = 5n.

Linear recurrence relation of degree k

c0(n) and ck(n) are not identically zero.
Example
an + 5an-1 + 6an-2 = 0.
Linear recurrence relation with constant coefficients
c0(n), c1(n), …, ck(n) are constants.
Example
an + 5an-1 + 6an-2 = 0.

Homogeneous recurrence relation

f(n) is identically zero.
Example
an + 5an-1 + 6an-2 = 0.

Inhomogeneous recurrence relation

f(n) is not identically zero.
Example
an + 5an-1 + 6an-2 = 5n.
Solving recurrence relation by substitution and Generating
functions
Solving recurrence relation by substitution / Backtracking
Technique for finding an explicit formula for the sequence defined
by a recurrence relation.
Backtrack the value of an by substituting the definition of an-1, an-2,
… until a pattern is clear.
Examples
1. Use the technique of backtracking, to find an explicit formula for the
sequence defined by the recurrence relation and initial condition for
an = an-1+ 3, a1=2.
an = an-1 + 3 or an = an-1+ 1.3
= (an-2 + 3) + 3 = an-2+ 2.3
= ((an-3 + 3) + 3) + 3 = an-3+ 3.3
… …
= an-(n-1) + (n–1).3
= a1+ (n–1).3
= 2 + (n–1).3
 The explicit formula for the sequence is
an = 2 + (n-1).3
2. Use the technique of backtracking, to find an explicit formula for the
sequence defined by the recurrence relation and initial condition for
an = 2.5an-1, a1 = 4.

an = 2.5an-1
= 2.5(2.5an-2)
= (2.5)2an-2
= (2.5)3an-3
…
= (2.5)n-1an-(n-1)
= (2.5)n-1a1
= 4(2.5)n-1
 Explicit formula is an = 4(2.5)n-1
3. Use the technique of backtracking, to find an explicit formula for the
sequence defined by the recurrence relation and initial condition for
an= 5an-1+ 3, a1= 3.

an = 5an-1 + 3
= 5(5an-2+ 3) + 3
= 52an-2 + (5 + 1)3
= 52 (5an-3+ 3) + (5 + 1)3
= 53an-3 + (52 + 5 + 1)3
…
= 5n-1an-(n-1)+ (5n-2+ … + 52 + 5 + 1)3
=5n-1 a1 + (5n-2+ … + 52+ 5 + 1)3
= 5n-13 + (5n-2+ … + 52 + 5 + 1)3
= (5n-1+ 5n-2+ … + 52 + 5 + 1)3
= 3(5n– 1) / 4
 Explicit formula is an = 3(5n– 1) / 4
4. Use the technique of backtracking, to find an explicit formula for the
sequence defined by the recurrence relation and initial condition for
an = a n-1+ n, a1 = 4.

an = a n-1 + n
= an-2 + (n–1) + n
= an-3 + (n–2) + (n–1) + n
…
= an-(n-1)+ [n– (n–1) + 1] + … + (n–1) + n
= a1 + 2 + … + (n–1) + n
= a1 – 1 + [1 + 2 + … + (n-1) + n]
= 4 – 1 + n(n+1)/2
= 3 + n(n+1)/2

 Explicit formula is an = 3 + n(n+1)/2

Solving recurrence relations by Generating functions
Shifting properties of generating functions
Xk A(X) = Xk  anXn, n = 0 .. 
=  an Xn+k, n = 0 .. 
Replacing n+k by r, we get
 ar-k Xr, r = k .. 
Equivalent expressions for generating functions
If A(X) =  anXn, n = 0 .. , then
 anXn, n = k ..  = A(X) – a0 – a1X – … – ak-1Xk-1.
 an-1Xn, n = k ..  = X(A(X) – a0– a1X – … – ak-2Xk-2).
 an-2 Xn, n = k ..  = X2(A(X) – a0 – a1X – … – ak-3Xk-3).
 an-3 Xn, n = k ..  = X3(A(X) – a0 – a1X – … – ak-4Xk-4).
…
 an-k Xn, n = k ..  = Xk(A(X)).
Examples
 1. Solve the recurrence relation an – 7 an-1 + 10 an-2 = 0, n ≥ 0,
a0 = 10, a1 = 41, using generating functions.

1. Let A(X) =  an Xn, n = 0 .. .

2. Multiply each term in the recurrence relation by X n and sum from
2 to ∞.
 an Xn – 7 an-1 Xn + 10 an-2 Xn = 0, n = 2 .. .
3. Replace each infinite sum by an equivalent expression.
[A(X) – a0 – a1X] – 7X[A(X) – a0] + 10X2[A(X)] = 0.
4. Simplify.
A(X)(1 – 7X +10X2) = a0 + a1X – 7 a0 X.
A(X) = [a0 + (a1 – 7 a0)X] / (1 – 7X +10X2)
= [a0 + (a1 – 7 a0)X] / [(1 – 2X) (1 – 5X)]
5. Decompose A(X) as a sum of partial fractions.
A(X) = C1 / (1 – 2X) + C2 / (1 – 5X)
6. Express A(X) as a sum of familiar series.
A(X) = C1  2n Xn + C2  5n Xn , n = 0 .. .
= (C1 2n + C2 5n) Xn , n = 0 .. .
7. Express an as the coefficient of Xn in A(X) and in the sum of the
other series.
an = C1 2n + C2 5n.
8. Determine the values of C1 and C2 .
For n = 0, a0 = C1 + C2 = 10 … (1)
For n = 1, a1 = 2 C1 + 5 C2 = 41 … (2)
Solving (1) and (2), we get
C1 = 3
C2 = 7
 an = (3) 2n + (7) 5n.
 2. Solve the recurrence relation an – 9 an-1 + 26 an-2 – 24 an-3 = 0,
n ≥ 3, a0 = 0, a1 = 1, and a2 = 10 using generating functions.

1. Let A(X) =  an Xn, n = 0 .. .

2. Multiply each term in the recurrence relation by X n and sum from 3
to ∞.
 an Xn – 9  an-1 Xn + 26  an-2 Xn – 24  an-3 Xn = 0,
n = 3 .. .
3. Replace each infinite sum by an equivalent expression.
[A(X) – a0 – a1 X – a2 X2] – 9X [A(X) –a0 – a1X] –
26X2 [A(X) – a0] –24X3[A(X)] = 0.
4. Simplify.
A(X)(1 – 9X + 26X2 – 24X3)
= a0 + a1 X + a2 X2 – 9 a0 X – 9 a1 X2 + 26 a0 X2.
A(X) = [a0 + ( a1 – 9 a0 ) X + ( a2 – 9 a1+ 26 a0 ) X2] /
(1– 9X + 26X2 – 24X3)
= [a0 + (a1 – 9 a0) X + (a2 – 9 a1 + 26 a0) X2] /
[(1 – 2X) (1 – 3X) (1 – 4X)]
5. Decompose A(X) as a sum of partial fractions.
A(X) = C1 / (1 – 2X) + C2/ (1 – 3X) + C3/ (1 – 4X)
6. Express A(X) as a sum of familiar series.
A(X) = C1  2n Xn + C2  3n Xn + C3  4n Xn, n = 0 .. .
= (C12n + C2 3n + C2 3n + C3 4n) Xn, n = 0 .. .
7. Express an as the coefficient of Xn in A(X) and in the sum of the
other series.
an = C1 2n + C2 3n + C3 4n.
8. Determine the values of C1, C2 and C3.
Substituting a0 = 0, a1 = 1, and a2 = 10 in step 4, we get
A(X) = [X + X2] / [(1 – 2X) (1 – 3X) (1 – 4X)]
= C1 / (1 – 2X) + C2 / (1 – 3X) + C3 / (1 – 4X)
i.e., C1(1 – 3X) (1 – 4X) + C2 (1 – 2X) (1 – 4X)
+ C3(1 – 2X) (1 – 3X) = X + X2
for X = 1/2, C1 = 3/2
for X = 1/3, C2 = –4
for X = 1/4, C3 = 5/2

 an = (3/2) 2n – (4) 3n + (5/2) 4n .

Exercises
1. Solve the recurrence relation an – an-1 – 9 an-2+ 9 an-3= 0, n ≥ 3,
a0 = 0, a1 = 1, and a2 = 2 using generating functions.
2. Solve the recurrence relation an– 3 an-2 + 2 an-3= 0, n ≥ 3, a0 = 1,
a1 = 0, and a2 = 0 using generating functions
Method of Characteristics roots
Characteristic equation for a linear homogeneous recurrence relation
of degree k, an = r1an-1+ … + rkan-k is
xk = r1xk-1+ r2xk-2+ … + rk.
1. Characteristic equation x2 – r1x – r2 = 0 of the recurrence relation
an = r1an-1+ r2an-2, having two distinct roots s1 and s2.
Explicit formula for the sequence is an = us1n + vs2n and u and v
depend on the initial conditions.
2. Characteristic equation x2 – r1x – r2 = 0 of the recurrence relation
an = r1an-1+ r2an-2 having a single root s.
Explicit formula for the sequence is an = usn + vnsn and u and v
depend on the initial conditions.
Examples
1. Solve the recurrence relation an = 4an-1 + 5an-2, a1 = 2, a2 = 6.
The associated equation is x2 – 4x – 5 = 0
i.e. (x – 5)(x + 1) = 0
 The different roots are s1 = 5 and s2 = –1.
Explicit formula is an = us1n + vs2n
a1 = u(5) + v(–1) = 5u – v
Given a1 = 2
 5u – v = 2 (1)
a2 = u(5)2 + v(–1)2 = 25u + v
Given a2 = 6
 25u + v = 6 (2)
Solving the equations (1) and (2), we get
u = 4/15 and v = –2/3
 Explicit formula is an = us1n + vs2n
= 4/15(5)n –2/3(–1)n
2. Solve the recurrence relation an = -6an-1 – 9an-2,
a1 = 2.5, a2 = 4.7.
The associated equation is x2 + 6x + 9 = 0
i.e. (x + 3)2 = 0
 The multiple root is s = –3.

Explicit formula is an = usn + vsn

a1 = u(–3) + vn(–3) = –3u + 3v

Given a1 = 2.5
 –3u + 3v = 2.5 (1)

a2 = u(–3)2 + vn(–3)2 = 9u + 18v

Given a2 = 4.7
 9u + 18v = 4.7 (2)

Solving the equations (1) and (2), we get

u = –19.7/9 and v = 12.2/9

 Explicit formula is an = usn + vnsn

= (–19.7/9)(–3)n + (12.2/9)n(–3)n
3. Solve the recurrence relation an = 2an-2 , a1 = 2, a2 = 6.

The associated equation is x2 – 2 = 0

i.e. (x – 2)(x + 2) = 0
 The different roots are s1 = 2 and s2 = –2.

Explicit formula is an = us1n + vs2n

a1 = u(2) + v(–2) = 2u – 2v

Given a1 = 2
 2u – 2v = 2
u–v=1 (1)

a2 = u(2)2 + v(–2)2 = 2u + 2v

Given a2 = 6
 2u + 2v = 6
u+v=3 (2)

Solving the equations (1) and (2), we get

u = 2 and v = 1

 Explicit formula is an = us1n + vs2n

= 2(2)n + (–2)n