Discrete Math For Computing II

Discrete Math For Computing II

• Main Text:
• Topics in enumeration; principle of inclusion and
exclusion, Partial orders and lattices. Algorithmic
complexity; recurrence relations, Graph theory.
• Prerequisite: CS 2305
• "Discrete Mathematics and its Applications"
Kenneth Rosen, 5th Edition, McGraw Hill.
Contact Information

B. Prabhakaran
Department of Computer Science
University of Texas at Dallas
Mail Station EC 31, PO Box 830688
Richardson, TX 75083
Email: [email protected]
Phone: 972 883 4680; Fax: 972 883 2349
URL: http://www.utdallas.edu/~praba/cs3305.html
Office: ES 3.706
Office Hours: 1-2pm Tuesdays, Thursdays
Other times by appointments through email
Announcements: Made in class and on course web page.
TA: TBA.
Course Outline

• Selected topics in chapters 6 through 9.

• Chapter 6: Advanced Counting Techniques: recurrence
relations, principle of inclusion and exclusion
• Chapter 7: Relations: properties of binary relations,
representing relations, equivalence relations, partial
orders
• Chapter 8: Graphs: graph representation,
isomorphism, Euler paths, shortest path algorithms,
planar graphs, graph coloring
• Chapter 9: Trees: tree applications, tree traversal,
trees and sorting, spanning trees
Course ABET Objectives

• Ability to construct and solve recurrence relations

• Ability to use the principle of inclusion and exclusion to solve
problems
• Ability to understand binary relations and their applications
• Ability to recognize and use equivalence relations and partial
orderings
• Ability to use and construct graphs and graph terminology
• Ability to apply the graph theory concepts of Euler and
Hamilton circuits
• Ability to identify and use planar graphs and shortest path
problems
• Ability to use and construct trees and tree terminology
• Ability to use and construct binary search trees
 Evaluation

• 1 Mid-terms: in class. 75 minutes. Mix of MCQs

(Multiple Choice Questions) & Short Questions.
• 1 Final Exam: 75 minutes or 2 hours (depending on
class room availability). Mix of MCQs and Short
Questions.
• 2 - 3 Quizzes: 5-6 MCQs or very short questions. 15-20
minutes each.
• Homeworks/assignments: 3 or 4 spread over the
semester.
 Homeworks

• Each homework will be for 10 marks.

• Homeworks Submission:
• Submit on paper to TA/Instructor.
 Grading

• Home works: 5%
• Quizzes: 15%
• Mid-term : 40%
• Final: 40%
 Likely Letter Grades

• Mostly Relative grading

• A-, A, A+: 1.2 – 1.4 times class average
• B-, B, B+: 1 – 1.2 class average
• C-, C, C+: 0.8 – 1.0 class average
• D-, D, D+: 0.7 – 0.8 class average
• F: Below 0.6 times class average
• Absolute scores will also influence above the ranges.
 Schedule

• Quizzes: Dates announced in class & web, a week

ahead. Mostly before midterm and final.
• Mid-term: February 23, 2006
• Final Exam: Last day of class (April 20th) OR 11am, May
1, 2006 (As per UTD schedule)
• Subject to minor changes
• Quiz and homework schedules will be announced in
class and course web page, giving sufficient time for
preparation.
Cheating

• Academic dishonesty will be taken

seriously.
• Cheating students will be handed over to
Head/Dean for further action.
• Remember: home works (and exams too!)
are to be done individually.
• Any kind of cheating in home works/exams
will be dealt with as per UTD guidelines.
 Chapter 6
Advanced Counting Techniques
§ 6.1: Recurrence Relations

Definition: A recurrence relation for the

sequence {an} is an equation expressing an
in terms of one or more of the previous
terms of the sequence:
a1,a2,a3,…,an-1, with n>=n0, (n0 being a
nonnegative integer).
A sequence is called a solution of a
recurrence relation if its terms satisfy the
recurrence relation.
Recurrence Example

• Consider the recurrence relation

an = 2an−1 − an−2 (n≥2).
• Which of the following are solutions?
an = 3n Yes -> 2 [3(n-1)] – 3(n-2)
= 3n => an
a n = 2n No -> a0 = 1, a1 = 2, a2 = 4;
a2 = 2a1 – a0 = 2.2 – 1 = 3 ≠ a2
an = 5 Yes -> an = 2.5 – 5 = 5 = an
Initial conditions

The initial conditions (base conditions):

Specify the terms that precede the first term where
the recurrence relation takes effect.
i.e., specify a0
Example Applications

• Growth of bank account

Initial Amount P0=$10,000
After n Years= Pn
Compound Interest I = 11%
Soln: Pn =Pn-1+(I/100)Pn-1=(1.11)Pn-1
Using Iteration we get, Pn =(1.11)n P0
i.e P30 =(1.11)3010,000=$228,922.97
Example Application

• Rabbits and Fibonacci Numbers

Growth of rabbit population in which each rabbit yields
1 new one every period starting 2 periods after its
birth.
Pn = Pn−1 + Pn−2 (Fibonacci relation)
 Classic Tower of Hanoi Example
• Problem: Get all disks from peg 1 to peg 2.
• Only move 1 disk at a time.
• Never set a larger disk on a smaller one.
Hanoi Recurrence Relation

• Let Hn = # moves for a stack of n disks.

• Optimal strategy:
• Move top n−1 disks to spare peg. (Hn−1
moves)
• Move bottom disk. (1 move)
• Move top n−1 to bottom disk. (Hn−1 moves)
• Note: Hn = 2Hn−1 + 1
Why is Hn = 2Hn-1 + 1

• Only move 1 disk at a time.

• Never set a larger disk on a smaller one.
Solving Tower of Hanoi RR

Hn = 2 Hn−1 + 1
= 2 (2 Hn−2 + 1) + 1 = 22 Hn−2 + 2 + 1
= 22(2 Hn−3 + 1) + 2 + 1 = 23 Hn−3 + 22
+2+1
…
= 2n−1 H1 + 2n−2 + … + 2 + 1
= 2n−1 + 2n−2 + … + 2 + 1 (since H1=1)
= 2n − 1
§6.2: Solving Recurrences

• Definition: A linear homogeneous recurrence relation of

degree k with constant coefficient is a recurrence
relation of the form
an = c1an−1 + … + ckan−k,
where the ci are all real, and ck ≠ 0.
• The solution is uniquely determined if k initial conditions
a0…ak−1 are provided
§6.2: Solving Recurrences..

• Linear?: Right hand side is sum of

multiples of previous terms.
• Homogenous?: No terms that are NOT
multiples of the ajs.
• Degree?: k-degree since previous k terms
are used.
Solving with const. Coefficients

• Basic idea: Look for solutions of the form

an = rn, where r is a constant.
• This requires the characteristic
equation:
rn = c1rn−1 + … + ckrn−k, i.e.,
rk − c1rk−1 − … − ck = 0 (Dividing both sides
by rn-k and subtracting right hand side from left).
• The solutions (characteristic roots) can
yield an explicit formula for the sequence.
Solving ……

• Theorem1: Let c1 and c2 be real numbers.

Suppose that r2 − c1r − c2 = 0 has two distinct roots
r1and r2. Then the sequence {an} is a solution of the
recurrence relation an = c1an−1 + c2an−2 if and only if an
= α1r1n + α2r2n for n≥0, where α1, α2 are constants.
Theorem 1: Proof

• 2 things to prove
• Case 1: Roots are r1 & r2, i.e., {an = α1r1n +
α2r2n}  an is a solution.
• r1 & r2 are roots of r2 − c1r − c2 = 0  r12 = c1r1 +
c2; r22 = c1r2 + c2.
•  c1an−1 + c2an−2 = c1(α1r1n-1 + α2r2n-1) + c2(α1r1n-2
+ α2r2n-2)
•  α1r1n-2 (c1r1 + c2) + α2r2n-2 (c1r2 + c2)
•  α1r1n-2 r12 + α2r2n-2 r22
•  an
Theorem 1: Proof …

• 2 things to prove
• Case 2: an is a solution  an = α1r1n + α2r2n for some α1 &
α2.
• an is a solution  a0 = C0 = α1 + α2.
•  C1 = α1r1 + α2r2
•  α1 = (C1 – C0r2)/(r1- r2)
•  α2 = C0 – α1 = (C0r1 - C1)/(r1- r2)
•  Works only if r1≠ r2
• an = α1r1n + α2r2n works for 2 initial conditions
•  Since the initial conditions uniquely determine
the sequence, an = α1r1n + α2r2n
Example

• Solve the recurrence an = an−1 + 2an−2 given the

initial conditions a0 = 2, a1 = 7.
• An = rn  rn = rn-1 + 2rn-2  r2 = r +2
• Solution: Use theorem 1
• c1 = 1, c2 = 2
• Characteristic equation:
r2 − r − 2 = 0
• Solutions: r = [−(−1)±((−1)2 − 4·1·(−2))1/2] /
2·1
= (1±91/2)/2 = (1±3)/2, so r = 2 or r =
−1.
• So an = α1 2n + α2 (−1)n.
Example Continued…

• To find α1 and α2, solve the equations for the

initial conditions a0 and a1:
a0 = 2 = α120 + α2 (−1)0
a1 = 7 = α121 + α2 (−1)1
Simplifying, we have the pair of equations:
2 = α1 + α2
7 = 2α1 − α2
which we can solve easily by substitution:
α2 = 2−α1; 7 = 2α1 − (2−α1) = 3α1 − 2;
9 = 3α1; α1 = 3; α2 = -1.
• Final answer: an = 3·2n − (−1)n
The Case of Degenerate Roots

• Theorem2: Let c1 and c2 be real

numbers with c2 ≠ 0. Suppose that r2
− c1r − c2 = 0 has only one root r0.
A sequence {an} is a solution of the
recurrence relation an=c1an-1 + c2an-2 if
and only if an = α1r0n + α2nr0n, for all n≥0,
for some constants α1, α2.
k-Linear Homogeneous Recurrence
Relations with Constant Coefficients
Theorem3: Let c1,c2,….ck be real numbers.
Suppose the C.E. r k 
k k

an  ci an  i
k i
c r 0 i
i 1
If this has k distinct roots ri, then the solutions
to the recurrence are of the form: i 1
k
an   i ri n if and only if
i 1

for all n≥0, where the αi are constants.

Theorem 3: Example

Let an = 6an-1 – 11an-2 +6an-3;

a0=2,a1=5, & a2=15.
C.E.: r3 – 6r2 + 11r – 6; Roots = 1,2, & 3.
Solution: an = α1.1n + α2.2n + α3.3n
a0 = 2 = α1 + α2 + α3
a1 = 5 = α1 + α2.2 + α3.3
a2 = 15 = α1 + α2.4 + α3.9
α1 = 1; α2 = 1; α3 = 2.
an = 1 – 2n + 2.3n
Degenerate t-roots

• Theorem 4: Suppose there are t roots r1,…,rt

with multiplicities m1,…,mt. mi >= 1 for i =
1…t. Then:
t  mi  1  n
an     i , j n  ri
j

i 1  j 0 
for all n≥0, where all the α are constants.
Theorem 4: Example

• E.g., Roots of C.E. are 2, 2, 2, 5, 5, & 9.

• Solution:
(α1,0 + α1,1n + α1,2n2).2n + (α2,0 + α2,1n).5n +
α3,09n
Linear NonHomogeneous Recurrence
Relations with Constant Coefficients
• Linear NonHomogeneous RRs with constant coefficients
may (unlike LiHoReCoCos) contain some terms F(n)
that depend only on n (and not on any ai’s). General
form:
an = c1an−1 + … + ckan−k + F(n)

The associated homogeneous recurrence relation

(associated LiHoReCoCo).
 Solutions of LiNoReCoCos

• Theorem 5: If an(p) is a particular solution to the

LiNoReCoCo, then

 k

Then all its solutions a
a =a
n n
are 
n
+a
(p)
n
, (h)
i 1
ci an  i   F (n)
of the form:

where a n is a solution to the associated
(h)

homogeneous RR

 k 
an   ci an  i 
 i 1 
Theorem 5: Proof

• {an(p)} is a particular solution:

• an(p) = c1an-1(p) + c2an-2(p) +..+ckan-k(p) + F(n)  (1)
• Let bn be a 2nd solution:
• bn = c1bn-1 + c2bn-2 +..+ckbn-k + F(n)  (2)
• (2) – (1):
• bn – an(p) = c1(bn-1 - an-1(p)) + c2(bn-2 - an-2(p)) +… +
ck(bn-k - an-k(p))
• Hence, {bn – an(p) } is a solution of the associated
homogeneous RR, say {an(h) }.  bn = an(p) + an(h)
.
Example

• Find all solutions to an = 3an−1+2n. Which solution has

a1 = 3?
• To solve this LiNoReCoCo, solve its associated
LiHoReCoCo equation: an = 3an−1, and its
solutions are an(h) = α3n, where α is a constant.
• By Theorem 5, the solutions to the original
problem are all of the form an = an(p) + α3n.
So, all we need to do is find one an(p) that
works.
Trial Solutions

• Since F(n)=2n, i.e it is linear so a reasonable

trial solution is a linear function in n, say pn = cn
+ d.
Then the equation an = 3an−1+2n becomes,
cn+d = 3(c(n−1)+d) + 2n, (for all n)

Simplifying, we get
(−2c+2)n + (3c−2d) = 0 (collect terms)
So c = −1 and d = −3/2.
So a(p)n = −n − 3/2 is a solution.
Finding a Desired Solution

• From Theorem 5, we know that all general

solutions to our example are of the form:
an = −n − 3/2 + α3n.
Solve this for α for the given case, a1 = 3:
3 = −1 − 3/2 + α31
α = 11/6
• The answer is an = −n − 3/2 + (11/6)3n
Theorem 6

 k 
• Suppose {an} satisfies LiNoRR: an   ci an  i   F (n)
 i 1 
 t t i  n
And F(n)=   bt  i n ;s b t and s are real nos.
 i 0 
• When s is not a root of C.E. of the associated
LiHoRR, there is a particular solution of the form
 t t i  n
  pt  i n  s
 i 0 
Theorem 6 continue…..

• When s is a root of this C.E. and its

multiplicity is m, there is a particular
solution of the form
 t t i  n
n   pt  i n  s
m

 i 0 

• This factor nm ensures that the proposed

particular solution will not already be a
solution of the associated LiHoRR.
Theorem 6: Example

• an = 6an-1 – 9an-2 + F(n); F(n) = 3n

C.E.: r2 – 6r + 9 => (r-3)2 = 0; r = 3.
Solution: By Theorem 6, s = 3,
multiplicity m = 2. Solution is of the form
p0n23n, where p0 is polynomial constant.
§6.3: Divide & Conquer R.R.s

• Many types of problems are solvable by

reducing a problem of size n into some number
a of independent subproblems, each of size
n/b, where a1 and b>1.
• The time complexity to solve such problems is
given by a recurrence relation:
• f(n) = a·f(n/b) + g(n)
where g(n) is the extra operations required.

This is called a divide-and-conquer recurrence relation

Divide+Conquer Examples

• Binary search: Break list into 1 sub-

problem (smaller list) (so a=1) of size n/2
(so b=2).
• So f(n) = 1.f(n/2)+2 (g(n)=2 constant)
• Merge sort: Break list of length n into 2
sublists (a=2), each of size  n/2 (so b=2),
then merge them, in g(n) = n operations
• So M(n) = 2M(n/2) + n
Fast Multiplication Example

• This algorithm splits each of two 2n-bit

integers into two n bits blocks. Thus, from
multiplications of two 2n-bit integers, it is
reduced to only three multiplications of n
bit integers, plus shifts and additions
• To find the product ab of two 2n-digit-
base 2 numbers, a=(a2n-1a2n-2…a0)2 and
b=(b2n-1b2n-2…b0)2,
first, we break them in half:
a=2nA1+A0, b=2nB1+B0,
Derivation of Fast Multiplication

ab (2 n A1  A0 )(2 n B1  B0 )
2n n (Multiply out
2 A1 B1  2 ( A1 B0  A0 B1 )  A0 B0 polynomials)
2 2 n A B  A B  Zero
1 1 0 0

2 n ( A1 B0  A0 B1  ( A1 B1  A1 B1 )  ( A0 B0  A0 B0 ))
(2 2 n  2 n ) A1 B1  (2 n  1) A0 B0 
2 n ( A1 B0  A1 B1  A0 B0  A0 B1 )
(2 2 n  2 n ) A1 B1  (2 n  1) A0 B0 
2 n ( A1  A0 )( B0  B1 ) (Factor last polynomial)
Three multiplications, each with n-digit numbers
Recurrence Rel. for Fast Mult.

Notice that the time complexity f(n) of the fast

multiplication algorithm obeys the recurrence:
• f(2n)=3f(n)+C(n)
i.e.,
• f(n)=3f(n/2)+C(n) Time to do the needed adds &
So a=3, b=2. subtracts of n-digit and 2n-digit
numbers
Theorem1

• Let f(n) = af(n/b) + c whenever n is divisible by

b, where a1, b is an integer greater than 1, and c
is a positive real number. Then

f (n) is
 if aa> 1
log
O(n b )
O (logif na) = 1
Furthermore, when n =bk, where k is a positive
integer,
f(n) = Clog + C 2.
n1 ba
where C1 = f(1) + c/(a-1) and C2 = -c/(a-1)
Theorem1: Proof
k1 k1
• n = bk; f(n) = akf(1) +  j

j 0
a c a f (1)  c  a j
k

j 0
• a = 1: f(n) = f(1) + ck; Since n = b , k = logbn  f(n) = f(1) +
k

clogbn.
• n is a power of b or not: f(n) is O(logn), when a = 1.
• a > 1: (from Theorem 1, Section 3.2)
• f(n) = akf(1) + c(ak – 1)/(a – 1)
= ak[f(1) + c/(a – 1)] – c/(a – 1)
= C1nlogba + C2, since ak = alogbn = nlogba
•  f(n) is O(nlogba)
MASTER Theorem

• Let f(n) = af(n/b) + cnd whenever n=bk,

where a1, b>1, and c and d are real number,
with c positive and d nonnegative. Then

if a<bd
f(n) is  O ( nd )
if a=bd
O ( n d log nif) a>bd
 o ( n logb a )

Example

• We know that number of comparisons used by the

merge sort to sort a list of n elements is less then M(n),
where M(n)=2M(n/2) + n
• From Master Theorem we find that M(n) is O(nlogn)
§6.4 Generating Functions

• Generating functions: Used to represent

sequences efficiently by coding the terms of a

sequence as coefficients of powers of a


variable x in a formal power series
k
a x
• Definition: Let S = {a0, a1, a2, a3, ...} be an
(infinite) sequence of real numbers. Then the
generating function G(x), of S is the k
series
G(x)=a0+a1x+…+akxk+…=
k 0
Example of GF

• Let S = {1, 1, 1, 1, ... }. Then 

G(x) =1+ x + x2 +...+ xk +... =  x
k 0
k

if S=1,1,1,1,1,1
By Theorem 1 of Section 3.2,
(x6-1)/(x-1) = 1+ x + x2 +...+ x5
G(x) =(x6-1)/(x-1) is the generating function
of sequence 1,1,1,1,1,1
Theorems of GF

• Theorem 1:
 
Let f(x) =
 a
k 0
k
and
x k g(x)= k
.
b Then,
k 0
x k


f(x) + g(x)=
k 0
 k k
( a  b and
) x k

f(x)g(x)=  k  k
   a j bk  j  x

k 0  j 0


Example

• A expression for the generating function of the

sequence S= {1, 1, 1, 1, 1, 1, ...} is 1/(1- x)
• 1/(1- x) = 1 + x + x2 + x3 + …

• Let f(x) = 1/(1-x) ; f(x) =
2
 ka xk
k 0
• What are the coefficients of ak?
• From theorem 1, we have (1 + x + x2 + ..) x (1 + x +
x2 + ..)  x  k 

 x  xk
• k 0 k 0
  k  k 
 1/(1- x) = =

2
 1 x
  (k  1) x k
k 0  j 0  k 0
Permutations & Combinations

• Permutation: An ordered arrangement of objects

from a set of distinct objects.
• P(n,r): n – distinct objects; r – number of objects in the
permutation
• P(n,r) = n(n-1)(n-2)…(n-r+1)
•  P(n,r) = n!/(n-r)!
• Example: Selecting first 3 prize winners from 100
different people  100 x 99 x 98 = 970,200.
• Combination: Unordered selection of r elements
from a set.
• C(n,r) = n!/[r!(n-r)!]; n is non-negative, 0 ≤ r ≤ n
• Proof: r-permutation is ordered r-combinations.
Permutations & Combinations …

• Proof: r-permutation is ordered r-combinations.:

•  P(n,r) = C(n,r) x P(r,r)
•  C(n,r) = P(n,r) / P(r,r)
•  C(n,r) = [n!/(n-r)!]/[r!/(r-r)!]
• = n!/[r!(n-r)!]
• Corollary 1: C(n,r) = C(n,n-r) if n & r are non-negative and
r ≤ n.
• Proof: C(n,n-r) = n!/[(n-r)!(n- (n-r))!]
• = n!/[(n-r)!r!]
• Note: C(n,n) = C(n,0) = 1.
• Example: Choosing 5 players from 10 member team.
•  C(10,5) = 10!/(5!5!) = 252.
Binomial Coefficients & Theorem

• C(n,r): denoted by n
• r
• C(n,r): Binomial coefficient  these numbers occur
as co-efficients in the expansion of powers of
binomial expressions such as (a+b)n
• Binomial Theorem: x, y are variables; n – non-
negative integer.
• (x + y)n = ∑j=0 n C(n,j).xn-j.yj
• = C(n,0)xn + C(n,1)xn-1y + C(n,2)xn-2y2
+….C(n,n-1)xyn-1 + C(n,n)yn
• Example: (x + y)2 = (x+y)(x+y) = xx + xy + xy + yy
= x2 + 2xy + y2
Binomial Coefficients & Theorem

• Proof: Count the number of terms of form xn-j.yj

•  Choose xs from the n sums (so that the other terms
are ys)
•  Coefficient of xn-jyj is C(n,n-j) = C(n,j)
• Example 1: Expansion of (x + y)4
• = ∑j=0 4 C(4,j)x4-jyj
• = C(4,0)x4 + C(4,1)x3y + C(4,2)x2y2 + C(4,3)xy3 +
C(4,4)y4
• = x4 + 4x3y + 6x2y2 + 4xy3 + y4
• Example 2: Coeffiecient of x12y13 in (x+y)25
• = C(25,13) = 25!/(13! 12!) = 5,200,300
Binomial Theorem: Corollary 1

• Corollary 1: n is non-negative  ∑k=0 n C(n,k) = 2n

• Proof: 2n = (1 + 1)n = ∑k=0 n C(n,k)1k1n-k
• = ∑k=0 n C(n,k)
•  Basically, ∑k=0 n C(n,k) is the total number of
subsets of a set with n elements
•  2n
Binomial Theorem: Corollary 2

• Corollary 2: n is positive  ∑k=0 n (-1)kC(n,k) = 0

• Proof: 0 = 0n = ((-1) + 1)n = ∑k=0 n C(n,k)(-1)k1n-k
• = ∑k=0 n C(n,k)(-1)k
•  C(n,0) + C(n,2) + C(n,4) + .. = C(n,1) + C(n,3) +
C(n,5) + …
Binomial Theorem: Corollary 3

• Corollary 3: n is positive  ∑k=0 n 2kC(n,k) = 3n

• Proof: 3n = (1+2)n = ∑k=0 n C(n,k)(2)k1n-k
• = ∑k=0 n
C(n,k)2k
Permutations With Repetitions

• Chapter 4.5
• Theorem 1: Number of r-permutations of a set of n
objects with repetitions allowed is nr.
• Theorem 2: There are C(n+r-1, r) r-combinations
from a set with n elements when repetition of
elements is allowed.
Extended Binomial Coefficients

• Definition: Let u be a real number and k a

non-negative integer. Then the extended
binomial coefficient (EBC) is:

 
if k>0,
u u ( u  1)....( u  k 1) / k !
if k=0.
k 1
•
e.g:EBC(-2,3) = (-2)(-3)(-4) / 3! = -4.

EBC(0.5,3)=(0.5)(-0.5)(-1.5) / 3! = 1/16
 Example

When the top parameter (u) is negative

integer, the EBC can be expressed in terms
of ordinary binomial coefficients. Using
earlier definition:
EBC
 n r =n  r(-n)(-n-1)….(-n-r+1)/r!
 1
   1   r 1 C (n  r  1, r ) n, r  Z 
r

 r   = r(-1)  n(n+1)…(n+r-1)/r!
= (-1)r(n+r-1)!/(r!*(n-1)!). Therefore,
The Extended Binomial Theorem

• Let x be a real number with |x| < 1

and let u be a real number. Then


(1+x) =
  x
u u k
k
k 0
Example

• Find GF for (1+x)-n and (1-x)-n

By EB Theorem, it follows that

(1+x) =
-n
 k n x k
 
k 0
We know that n
( 1) r C (n  r  1, r )
r

(1+x) = -n
 (
k 0
 1) k
C ( n  k  1, k ) x k

Replacing x with –x we get


(1-x)-n =
 C ( n
k 0
 k  1, k ) x k
Counting Problems and Generating
Functions
• Generating Functions can be used to solve a wide
variety of counting problems. In particular they can be
used to count the number of combinations of various
types.
Example

• Use GF to find the number of ways to select r

objects of n different kinds {ar} if we must select
at least one object of each kind?
• Chapter 4, theorem.
• Each of the n kinds of objects contributes the
factor (x+x2+x3…+) to the GF G(x), Hence
• G(x)= (x+x2+x3…+)n=xn(1+ x+x2+x3…+) =xn/(1-x)n

• Using EB Theorem, we have
G(x) = x (1-x) = x
n. -n n 


r 0
=x r
n
r
 C (n  r  1, r ) x
( x)
n

r 0
r


G(x)=
 C (n  r  1, r ) x
r 0
n r
Continue…

• Substituting t = n+r, we get


G(x)=  C (t  1, t  n ) x t

• Replacing t tn by r as the index of summation,

we get

G(x)=

r n
C ( r  1 , r  n ) x r

Hence, there are C(r-1,r-n) ways to select r

objects of n different kinds if we must select
at least one object of each kind.
Using GF to Solve Recurrence Relations

• We can find the solution to a recurrence relation and its

initial conditions by finding an explicit formula for the
associated generating function.
Example

• Solve the R.R., ak=3ak-1 for k=1,2,3… and initial

condition a0=2.

ak 3ak  1 a0 2.
  
G ( x) a0   ak x a0   3ak  1 x a0  3x  a j x j a0  3xG ( x ) 
k k

k 1 k 1 j 0
 
a0
G ( x)  a0  3 x  (2 3k ) x k
k k

1  3x k 0 k 0


• Uses identity1/(1-ax) =  x
a k k

k 0

• Consequently, ak=2 . 3k
§6.5 The principle of Inclusion-Exclusion

If A, B and C are finite sets then

|A  B| = |A| + |B| – |A  B|
Continue…

• If A, B and C are finite sets then

|A  B  C| = |A| + |B| + |C| - |A  B| - |B  C| - |A  C| - |
ABC|

|A  B|

|A| |B|

|A  C| |B  C|

|C|
General Inclusion-Exclusion

• Theorem 1: Let A1,…,An be finite sets.

Then | A1  A2 …… An | =

 | Ai | 
1i n
 | Ai  Aj | 
1i j n
 i j k
| A  A
1i j k n
 A |  ...  (  1) n 1
| A1  A2 ...  An |
General Theorem: Proof

• | A1  A2 …… An |: Show an element in the union is

counted only once in the right hand side.
• Let a be a member of exactly r sets (A1, …).
• a is counted C(r,1) times by 1st summation, C(r,2) by 2nd
summation of the intersections, and C(r,m) times by mth
summation involving m sets Ai.
• a is counted exactly C(r,1) – C(r,2) + C(r,3) - … + (-
1)r+1C(r,r).
• By Corollary 2 of Section 4.4, C(r,0) - C(r,1) + C(r,2) -
C(r,3) - … + (-1)rC(r,r) = 0.
•  C(r,0) = C(r,1) – C(r,2) + C(r,3) - … + (-1)r+1C(r,r).
•  a is counted only once.
§6.6 Applications of Inclusion-Exclusion

• Alternative Form: Let Ai be the subset containing

the elements that have Property Pi. Writing these
quantities in terms of sets, we have
| Ai1  Ai 2 ...  Aik | N ( Pi1 Pi 2. ..Pik )
• If the number of elements with none of the
properties P1,…,Pn is denoted by N(P1’ P2’…Pn’)
then, N(P1’ P2’…Pn’)=N-
| A1  A2  ...  An |
Continue

• From the inclusion-

exclusion principle,
we see that
N(P1’ P2’…Pn’)=N
  N | P |   N (P P )  
1i n
i
1i j n
i j N ( Pi Pj Pk )  ...
1i j k n

 ( 1) n N ( P1 P2 ...Pn )
The Number of Onto Functions

• Let m and n be positive integers with (m  n).

How many ways to assign m different jobs to n
different employees if every employee is
assigned at least one job?
• Assume m = 5, n = 4, following Theorem 1 (next
slide):
• 45 – C(4,1).35 + C(4,2).25 – C(4,3)15 = 240
The Number of Onto Functions

• Theorem: Let m and n be positive integers with

(m  n). Then, there are

nm - C (n,1)·(n -1)m + C (n,2)·(n -2)m+…

+(-1)iC (n,i )·(n-i )m +…+ (-1)n-1C (n,n-1)·1m

onto functions from a set with m elements to a

set with n elements
Derangements

• A derangement is a permutation of objects

that leaves no object in its original position.
• Theorem: The number of derangements of a
set with n elements is

 1 1 1 1 n 1 
Dn n! 1         1 
 1! 2! 3! 4! n! 
Theorem Proof

• Property Pi: A permutation that fixes element i.

• Number of derangements: number of permutations having
none of the properties Pi for i = 1,2,…,n.
• Dn = N(P1’,P2’,…,Pn’) (from inclusion-exclusion)
= N -  N ( Pi )  N ( Pi Pj ) - ...
i i j

• N: number of permutations of n elements. So, N = P(n).

• Similarly: N(Pi) = P(n-1); N(Pi,Pj) = P(n-2).
• Inserting these values in the expression for Dn, we get the
formula.