UNIT-VI

Recurrence Relation

Generating Functions:

In mathematics, a generating function is a formal power series in one indeterminate, whose

coefficients encode information about a sequence of numbers an that is indexed by the natural
numbers. Generating functions were first introduced by Abraham de Moivre in 1730, in order
to solve the general linear recurrence problem. One can generalize to formal power series in
more than one indeterminate, to encode information about arrays of numbers indexed by
several natural numbers.

Generating functions are not functions in the formal sense of a mapping from a domain to a
codomain; the name is merely traditional, and they are sometimes more correctly called
generating series.

Ordinary generating function

The ordinary generating function of a sequence an is

When the term generating function is used without qualification, it is usually taken to mean
an ordinary generating function.

If an is the probability mass function of a discrete random variable, then its ordinary
generating function is called a probability-generating function.

The ordinary generating function can be generalized to arrays with multiple indices. For
example, the ordinary generating function of a two-dimensional array am, n (where n and m are
natural numbers) is

Example:
Exponential generating function
The exponential generating function of a sequence an is

Example:

Function of Sequences:
Generating functions giving the first few powers of the nonnegative integers are given in the
following table.
series
1

There are many beautiful generating functions for special functions in number theory. A few
particularly nice examples are
(
2
)

(
3
)

(
4
)

for the partition function P, where is a q-Pochhammer symbol, and

(
5
 )

(
6
)

(
7
)

for the Fibonacci numbers .

Generating functions are very useful in combinatorial enumeration problems. For example,
the subset sum problem, which asks the number of ways to select out of given integers
such that their sum equals , can be solved using generating functions.

Calculating Coefficient of generating function:

By using the following polynomial expansions, we can calculate the coefficient of a

generating function.

Polynomial Expansions:
1  x m 1
1) 1  x  x 2  ...  x m
1 x
1
2) 1  x  x 2  ...
1 x
3) (1  x) n 1  C ( n,1) x  C ( n, 2) x 2  ...  C ( n, r ) x r  ...  C ( n, n) x n

4) (1  x m ) n 1  C (n,1) x m  C (n, 2) x 2 m  ...  (  1) k C ( n, k ) x km  ...  (  1) n C ( n, n) x nm

1
5) n
1  C (1  n  1,1) x  C (2  n  1, 2) x 2  ...  C (r  n  1, r ) x r  ...
(1  x)
6) If h(x)=f(x)g(x), where f(x) a0  a1 x  a2 x 2  ...and g(x) b0  b1 x  b2 x 2  ..., then
h(x) a0b0  (a1b0  a0b1 ) x  (a2b0  a1b1  a0b2 ) x 2  ...  (ar b0  ar  1b1  ar  2 b2  ...  a0 br ) x r  ...
 Example
x16 in ( x 2  x 3  x 4 )5
Find the 16
x in x10 (1  x )  5 [i.e, the x 6 term in (1  x )  5 is
coefficient of
multiplied by x10 to become the x16 term in x10 (1  x )  5 ]
2
To simplify the expression, we extract x from each polynomial factor and the apply
identity (2).
( x 2  x 3  x 4  )5 [ x 2 (1  x  x 2  )]5
 x10 (1  x  x 2  )5
1
 x10
(1  x)5

Thus the coefficient of x16 in ( x 2  x3  x 4 is)5the coefficient of

x16 in x10 (1-x) –5 [i.e., the x6 term in (1-x) –5 is multiplied by to become the x16 term
in x10 (1-x) –5 ]

1
n
1  C (1  n  1,1) x  C (2  n  1, 2) x 2  ...  C ( r  n  1, r ) x r  ...
(1  x)
From expansion (5) we see that the coefficient of x 6in (1  x )  5 is C (6  5  1, 6)
More generally, the coefficient of xr in

x r in x10 (1  x)  5 equals the coefficient of x r  10 in

(1  x)  5 , namely, C (( r  10)  5  1, ( r  10)).

Recurrence relations:

Introduction :A recurrence relation is a formula that relates for any integer n ≥ 1, the n-th
term of a sequence A = {ar}∞r=0 to one or more of the terms a0,a1,….,an-1.
Example. If Sn denotes the sum of the first n positive integers, then

(1) Sn = n + Sn-1. Similarly if d is a real number, then the nth term of an arithmetic
progression with common difference d satisfies the relation

(2) an = an -1 + d. Likewise if pn denotes the nth term of a geometric progression with

common ratio r, then

(3) pn = r pn – 1. We list other examples as:

(4) an – 3an-1 + 2an-2 = 0.
(5) an – 3 an-1+ 2 an-2 = n2 + 1.
(6) an – (n - 1) an-1 - (n - 1) an-2 = 0.
(7) an – 9 an-1+ 26 an-2 – 24 an-3 = 5n.
(8) an – 3(an-1)2 + 2 an-2 = n.
(9) an = a0 an-1+ a1 an-2+ … + an-1a0.
(10) a2n + (an-1)2 = -1.

Definition. Suppose n and k are nonnegative integers. A recurrence relation of the form
c0(n)an + c1(n)an-1 + …. + ck(n)an-k = f(n) for n ≥ k, where c0(n), c1(n),…., ck(n), and f(n)
are functions of n is said to be a linear recurrence relation. If c0(n) and ck(n) are not
identically zero, then it is said to be a linear recurrence relation degree k. If c0(n), c1(n),
…., ck(n) are constants, then the recurrence relation is known as a linear relation with
constant coefficients. If f(n) is identically zero, then the recurrence relation is said to be
homogeneous; otherwise, it is inhomogeneous.

Thus, all the examples above are linear recurrence relations except (8), (9), and (10); the
relation (8), for instance, is not linear because of the squared term.
The relations in (3), (4) , (5), and (7) are linear with constant coefficients.
Relations (1), (2), and (3) have degree 1; (4), (5), and (6) have degree 2; (7) has degree 3.
Relations (3) , (4), and (6) are homogeneous.
There are no general techniques that will enable one to solve all recurrence relations.
There are, nevertheless, techniques that will enable us to solve linear recurrence relations
with constant coefficients.

SOLVING RECURRENCE RELATIONS BY SUSTITUTION AND GENERATING

FUNCTIONS
We shall consider four methods of solving recurrence relations in this and the next two
sections:
1. Substitution (also called iteration),
2. Generating functions,
3. Characteristics roots, and
4. Undetermined coefficients.
In the substitution method the recurrence relation for an is used repeatedly to solve for
a general expression for an in terms of n. We desire that this expression involve no other
terms of the sequence except those given by boundary conditions.
The mechanics of this method are best described in terms of examples. We used this
method in Example5.3.4. Let us also illustrate the method in the following examples.

Example

Solve the recurrence relation an = a n-1 + f(n) for n ³1 by substitution

 a1= a0 + f(1)
a2 = a1 + f(2) = a0 + f(1) + f(2))

a3 = a2 + f(3)= a0 + f(1) + f(2) + f(3)

.
.

.
an = a0 + f(1) + f(2) +….+ f(n)
n
= a0 + ∑ f(k)
K=1
Thus, an is just the sum of the f(k) ‘s plus a0.

More generally, if c is a constant then we can solve an = c a n-1 + f(n) for n ³1 in the same
way:

a1 = c a0 + f(1)
a2 = c a1 + f(2) = c (c a0 + f(1)) + f(2)
= c2 a0 + c f(1) + f(2)
a3= c a2 + f(3) = c(c 2 a0 + c f(1) + f(2)) + f(3)
=c3 a0 + c2 f(1) + c f(2) + f(3)
.
.
.
an = c a n-1 + f(n) = c(c n-1 a0 + c n-2 f(1) +. . . + c n-2 + f(n-1)) + f(n)
=c n a0 + c n-1 f(1) + c n-2 f(2) +. . .+ c f(n-1) + f(n)
Or
an = c n a0 + ∑c n-k f(k)

Solution of Linear Inhomogeneous Recurrence Relations:

The equation 𝑎𝑛+𝑐1𝑎𝑛−1+𝑐2𝑎𝑛−2=(𝑛), where 𝑐1and 𝑐2 are constant, and 𝑓(𝑛) is not
identically 0, is called a second-order linear inhomogeneous recurrence relation (or
difference equation) with constant coefficients. The homogeneous case, which we’ve looked
at already, occurs when (𝑛)≡0. The inhomogeneous case occurs more frequently. The
homogeneous case is so important largely because it gives us the key to solving the
inhomogeneous equation. If you’ve studied linear differential equations with constant
coefficients, you’ll see the parallel. We will call the difference obtained by setting the right-

Say that 𝑉 is a solution. Now suppose that (𝑛) is any particular solution of the
hand side equal to 0, the “associated homogeneous equation.” We know how to solve this.

initial data.) Then 𝑈=𝑉+(𝑛) is a solution to the inhomogeneous equation, which you can see
inhomogeneous equation. (That is, it solves the equation, but does not necessarily match the

simply by substituting 𝑈 into the equation. On the other hand, every solution 𝑈 of the
inhomogeneous equation is of the form 𝑈=𝑉+(𝑛) where 𝑉 is a solution of the homogeneous
equation, and 𝑔(𝑛) is a particular solution of the inhomogeneous equation. The proof of this
is straightforward. If we have two solutions to the inhomogeneous equation, say 𝑈1 and 𝑈2,
then their difference 𝑈1−𝑈2=𝑉 is a solution to the homogeneous equation, which you can
check by substitution. But then 𝑈1=𝑉+𝑈2, and we can set 𝑈2=(𝑛), since by assumption,
𝑈2 is a particular solution. This leads to the following theorem: the general solution to the
inhomogeneous equation is the general solution to the associated homogeneous
equation, plus any particular solution to the inhomogeneous equation. This gives the
following procedure for solving the inhomogeneous equation:
1) Solve the associated homogeneous equation by the method we’ve learned. This will
involve variable (or undetermined) coefficients.
2) Guess a particular solution to the inhomogeneous equation. It is because of the guess that
I’ve called this a procedure, not an algorithm. For simple right-hand sides , we can say how to
compute a particular solution, and in these cases, the procedure merits the name “algorithm.”
3) The general solution to the inhomogeneous equation is the sum of the answers from the
two steps above.
4) Use the initial data to solve for the undetermined coefficients from step 1.
To solve the equation 𝑎𝑛 − 6𝑎𝑛−1 + 8𝑎𝑛−2 = 3. Let’s suppose that we are also given the
initial data 𝑎0 = 3, 𝑎1 = 3. The associated homogeneous equation is 𝑎𝑛 − 6𝑎𝑛−1 + 8𝑎𝑛−2 =
0, so the characteristic equation is 𝑟2 − 6𝑟 + 8 = 0, which has roots 𝑟1 = 2 and 𝑟2 = 4. Thus,
the general solution to the associated homogeneous equation is 𝑐12𝑛 + 𝑐24 . When the right-
hand side is a polynomial, as in this case, there will always be a particular solution that is a

that there is a constant 𝐶 that solves the homogeneous equation. If that is so, then 𝑎𝑛 = 𝑎𝑛
polynomial. Usually, a polynomial of the same degree will work, so we’ll guess in this case

−1 = 𝑎𝑛 −2 = 𝐶, and substituting into the equation gives 𝐶 − 6𝐶 + 8𝐶 = 3, and we find that

𝐶 = 1. Now, the general solution to the inhomogeneous equations is 𝑐12𝑛 + 𝑐24𝑛 + 1.

equations 𝑐1 + 𝑐2 + 1 = 3 and 2𝑐1 + 4𝑐2 + 1 = 3, whose solution is 𝑐1 = 3, 𝑐2 = −1.

Reassuringly, this is the answer given in the back of the book. Our initial data lead to the

Finally, the solution to the inhomogeneous equation, with the initial condition given, is 𝑎𝑛 =
3 ∙ 2𝑛 − 4𝑛 + 1. Sometimes, a polynomial of the same degree as the right-hand side doesn’t

𝑎𝑛 − 6𝑎𝑛 −1 + 5𝑎𝑛−2 = 3, when we guessed that the particular solution was a constant 𝐶,
work. This happens when the characteristic equation has 1 as a root. If our equation had been

we’d have arrived at the equation 𝐶 − 6𝐶 + 5𝐶 = 3, or 0 = 3. The way to deal with this is to
increase the degree of the polynomial. Instead of assuming that the solution is constant, we’ll

𝑔 𝑛 =𝑛𝐶. Then we have 𝑛𝐶−6 𝑛−1 𝐶+5 𝑛−2 𝐶=3, which simplifies to 6𝐶−10𝐶=3 so that
assume that it’s linear. In fact, we’ll guess that it is of the form

𝐶=−34 . Thus, 𝑔 𝑛 = −3𝑛4 . This won’t be enough if 1 is a root of multiplicity 2, that is, if
𝑟−1 2 is a factor of the characteristic polynomial. Then there is a particular solution of the
form 𝑔 𝑛 =𝐶𝑛2. For second-order equations, you never have to go past this. If the right-hand
side is a polynomial of degree greater than 0, then the process works juts the same, except

and then once more, if need be. For example, if the right-hand side were 𝑓 𝑛 =2𝑛−1, we
that you start with a polynomial of the same degree, increase the degree by 1, if necessary,

would start by guessing a particular solution 𝑔 𝑛 =𝐶1𝑛+𝐶2. If it turned out that 1 was a
characteristic root, we would amend our guess to 𝑔 𝑛 =𝐶1𝑛2+𝐶2𝑛+𝐶3. If 1 is a double
root, this will fail also, but 𝑔 𝑛 =𝐶1𝑛3+𝐶2𝑛2+𝐶3𝑛+𝐶4 will work in this case.

hand side is an exponential, say 𝑓 𝑛 =𝐶𝑛. In that case, we guess that a particular solution is
Another case where there is a simple way of guessing a particular solution is when the right-

just a constant multiple of 𝑓, say (𝑛)=𝑘𝐶𝑛. Again, we gave trouble when 1 is a

characteristic root. We then guess that 𝑔 𝑛 =𝑘𝑛𝐶𝑛, which will fail only if 1 is a double root.
In that case we must use 𝑔 𝑛 =𝑘𝑛2𝐶𝑛, which is as far as we ever have to go in the second-
order case. These same ideas extend to higher-order recurrence relations, but we usually
solve them numerically, rather than exactly. A third-order linear difference equation with
constant coefficients leads to a cubic characteristic polynomial. There is a formula for the
roots of a cubic, but it’s very complicated. For fourth-degree polynomials, there’s also a
formula, but it’s even worse. For fifth and higher degrees, no such formula exists. Even for
the third-order case, the exact solution of a simple-looking inhomogeneous linear recurrence
relation with constant coefficients can take pages to write down. The coefficients will be
complicated expressions involving square roots and cube roots. For most, if not all, purposes,
a simpler answer with numerical coefficients is better, even though they must in the nature of
things, be approximate.
The procedure I’ve suggested may strike you as silly. After all, we’ve already solved the
characteristic equation, so we know whether 1 is a characteristic root, and what it’s
multiplicity is. Why not start with a polynomial of the correct degree? This is all well and
good, while you’re taking the course, and remember the procedure in detail. However, if you
have to use this procedure some years from now, you probably won’t remember all the
details. Then the method I’ve suggested will be valuable. Alternatively, you can start with a
general polynomial of the maximum possible degree This leads to a lot of extra work if
you’re solving by hand, but it’s the approach I prefer for computer solution.