Generating Functions in Combinatorics

Aneesha Manne, Lara Zeng

Mentor: Uma Roy

Bedford High School, Belmont High School

April 19-20, 2018

MIT PRIMES Conference

Aneesha Manne, Lara Zeng Generating Functions 1 / 20

A Motivating Problem

Problem: A fish population starts out at 50 fish and grows 4-fold each
year with 100 fish dying each year

Mathematical Formalism
Population at time t is pt
Recurrence: pt = 4 · pt−1 − 100
Base case: p0 = 50

Natural question: What is pt for any t?

Aneesha Manne, Lara Zeng Generating Functions 2 / 20

A Fish population

Recurrence and Base Case: pt = 4 · pt−1 − 100, with p0 = 50

Iterative Calculations
p0 = 50
p1 = 100
p2 = 300
p3 = 1100
p4 = 4300

We want a closed form!

Aneesha Manne, Lara Zeng Generating Functions 3 / 20

Generating functions

A generating function takes a sequence of real numbers and makes it the

coefficients of a formal power series.

Generating Function
Let {fn }n≥0 be a sequence of real numbers. Then the formal power series
X
F (x) = fn x n
n≥0

is called the ordinary generating function of the sequence {fn }n≥0 .

Aneesha Manne, Lara Zeng Generating Functions 4 / 20

Formal Power Series

When using generating functions we will look at power series formally,

meaning we ignore convergence.
Convergence
Consider the power series expansion
1
= 1 + x + x2 + . . . .
1−x
When |x| < 1, you can plug in x and the RHS = LHS. For example, when
x = 12 :
1 1 1 1
= 2 = 1 + + + + ....
1 − 1/2 2 4 8

Aneesha Manne, Lara Zeng Generating Functions 5 / 20

Formal Power Series Cont.

Example Cont.
1
= 1 + x + x2 + . . . .
1−x
When |x| > 1, plugging in x does not yield meaningful equalites. Consider
x = 2:
1 1
= − 6= 1 + 2 + 4 + 8 + . . . = ∞.
1−2 2

Formal power series: Do not plug in values for x, because it is

meaningless! We only care about the coefficients of the series.

Aneesha Manne, Lara Zeng Generating Functions 6 / 20

Generating Functions for Solving Fish Population Problem

Define the generating function:

X
G (x) = pn x n .
n≥0

First few terms: G (x) = 50 + 100x + 300x 2 + . . .

Express Recurrence: pt+1 = 4 · pt − 100

X X
pn+1 · x n+1 = (4 · pn − 100) · x n+1
n≥0 n≥0
X X
= 4 · pn · x n+1 − 100 · x n+1
n≥0 n≥0

Aneesha Manne, Lara Zeng Generating Functions 7 / 20

Solving Fish Population Problem Cont.

Generating Function equality:

X X X
pn+1 · x n+1 = 4 · pn · x n+1 − 100 · x n+1
n≥0 n≥0 n≥0

Left hand side: G (x) − p0 , since it’s missing the first term of the
sequence {pn }n≥0
Right hand side term 1: 4x · G (x)
Right hand side term 2: − 100x
1−x , since
1
1−x = 1 + x + x2 + . . .
Recurrence in terms of G (x):
100x
G (x) − p0 = 4x · G (x) −
1−x

Aneesha Manne, Lara Zeng Generating Functions 8 / 20

Solving Fish Population Problem Cont.

Want to solve following equation for closed form for pt :

100x
G (x) − p0 = 4x · G (x) −
1−x
After rearranging,
p0 100x
G (x) = − .
1 − 4x (1 − x)(1 − 4x)

We have obtained an explicit formula for the G (x), the generating

function of the sequence {pn }.

Aneesha Manne, Lara Zeng Generating Functions 9 / 20

Finding formula for coefficients

Want closed form for coefficient of x n in G (x) because this is pn .

p0 100x
G (x) = − .
1 − 4x (1 − x)(1 − 4x)

First term’s contribution is easy to calculate:

p0 X X
= 50 (4x)n = 50 4n x n
1 − 4x
n≥0 n≥0

Aneesha Manne, Lara Zeng Generating Functions 10 / 20

Finding formula for coefficients cont.

Expanding 2nd term yields confusion:

100x X X
= 100x · xn · 4n x n .
(1 − x)(1 − 4x)
n≥0 n≥0

Another approach: partial fraction decomposition

We want to find constants A and B such that

100x A B
= + .
(1 − x)(1 − 4x) 1−x 1 − 4x
100 −100
With A = 3 and B = 3 ,

100x 100 1 100 1

= · − · .
(1 − x)(1 − 4x) 3 1 − 4x 3 1−x

Aneesha Manne, Lara Zeng Generating Functions 11 / 20

Using Partial Fractions

100x 100 1 100 1

= · − · .
(1 − x)(1 − 4x) 3 1 − 4x 3 1−x
Expanding using power series yields:
 
100 1 100 1 100 X n n X n 
· − · = 4 x − x .
3 1 − 4x 3 1−x 3
n≥0 n≥0

Thus 2nd term’s contribution to coefficient of x n is:

100 n
(4 − 1).
3

Aneesha Manne, Lara Zeng Generating Functions 12 / 20

An explicit formula for pn

Recall
p0 100x
G (x) = − .
1 − 4x (1 − x)(1 − 4x)
First term’s contribution:

50 · 4n .
Second term’s contribution:
100 n
(4 − 1).
3
Combining contributions, closed-form formula for pn is:

4n − 1
pn = 50 · 4n − 100 · .
3

Aneesha Manne, Lara Zeng Generating Functions 13 / 20

Exponential Generating Functions

Exponential generating functions are every similar to ordinary generating

functions.

Exponential Generating Function

Let {fn }n≥0 be a sequence of real numbers. Then the formal power series
X xn
F (x) = fn ,
n!
n≥0

is called the exponential generating function of the sequence {fn }n≥0 .

Intuition: Dividing by n! allows for fn to grow faster.

Aneesha Manne, Lara Zeng Generating Functions 14 / 20

Motivating Example
Recurrence Relation: Solve for an if a0 = 1, and an satisfies the
following recurrence

an+1 = (n + 1)(an − n + 1).

First few terms

a0 = 1
a1 = 2
a2 = 4
a3 = 9
a4 = 28
a5 = 125

This series grows too fast for an ordinary generating function. Therefore
an exponential generating function is used.
Aneesha Manne, Lara Zeng Generating Functions 15 / 20
Solving recurrence with exponential generating functions

Defining generating function:

∞
X xn
A(x) = an ,
n!
n=0

is the exponential generating function of the sequence {an }n≥0 .

Expressing recurrence an+1 = (n + 1)(an − n + 1):

∞ ∞ ∞
X x n+1 X x n+1 X x n+1
an+1 = an − (n − 1) .
(n + 1)! n! n!
n=0 n=0 n=0

Aneesha Manne, Lara Zeng Generating Functions 16 / 20

Solving recurrence cont.

∞ ∞ ∞
X x n+1 X x n+1 X x n+1
an+1 = an − (n − 1) .
(n + 1)! n! n!
n=0 n=0 n=0

LHS = A(x) − 1
RHS first term: xA(x)
RHS second term: −x 2 e x + xe x = (x − x 2 )e x
Plugging in above:

A(x) − 1 = xA(x) − x 2 e x + xe x .
Rearranging yields,
1
A(x) = + xe x .
1−x
xn
Thus coefficient an for n! is an = n! + n.

Aneesha Manne, Lara Zeng Generating Functions 17 / 20

Acknowledgements

Mentor: Uma Roy

Parents for driving us to Alewife
Dr. Slava Gerovitch, Prof. Pavel Etingof, Prof. Tanya Khovanova
Isabel Vogt for organizing and coordinating
MIT math department and MIT PRIMES program

Aneesha Manne, Lara Zeng Generating Functions 18 / 20

References

Miklos Bona (2012)

A Walk Through Combinatorics: An Introduction to Enumeration and Graph
Theory

Aneesha Manne, Lara Zeng Generating Functions 19 / 20

 The End

Aneesha Manne, Lara Zeng Generating Functions 20 / 20