You are on page 1of 83

HANOI NATIONAL UNIVERSITY OF EDUCATION

FALCUTY OF MATHEMATICS AND INFORMATICS

LƯU BÁ THẮNG

LECTURE NOTES

ELEMENTARY ALGEBRA

HANOI, 2022
2
Mục lục

1 Some basis principles 1


1.1 Dirichlet’s principle . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Pigeonhole principle: Simple form . . . . . . . . . . . . . 1
1.1.2 Pigeonhole principle: Strong form . . . . . . . . . . . . . 2
1.2 Discrete extremal principle . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 The existence of the configuration . . . . . . . . . . . . . 5
1.2.2 Find discrete extremes . . . . . . . . . . . . . . . . . . . 5
1.2.3 Create order relation based on equality . . . . . . . . . . 6
1.3 Infinite descent principle . . . . . . . . . . . . . . . . . . . . . . 8
1.3.1 Infinite descent principle in Diophantine equations . . . . 9
1.3.2 Infinite descent principle in Geometry . . . . . . . . . . . 11
1.4 The basis counting principles . . . . . . . . . . . . . . . . . . . 12
1.4.1 Addition, multiplication counting principles . . . . . . . 12
1.4.2 Inclusion – exclusion principle . . . . . . . . . . . . . . . 13
1.5 The issues with extreme views . . . . . . . . . . . . . . . . . . . 17
1.5.1 Applications of some extreme principles in equations and
inequalities . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5.2 Absolute values and minimum, maximum . . . . . . . . . 21

2 Primary problems of the number sequences 25


2.1 The finite sum can’t be represented by algebraic functions . . . 25
2.1.1 Rational function . . . . . . . . . . . . . . . . . . . . . . 25
2.1.2 Algebraic function . . . . . . . . . . . . . . . . . . . . . 26
2.2 Some types of number sequences . . . . . . . . . . . . . . . . . . 27
2.2.1 Affine sequences . . . . . . . . . . . . . . . . . . . . . . 27
2.2.2 Linear recurrence sequences of order two . . . . . . . . . 27
2.2.3 Sequences un+1 = f (un ). . . . . . . . . . . . . . . . . . . 30
2.3 The methods of finite determination . . . . . . . . . . . . . . . 32
2.3.1 Interpolating method . . . . . . . . . . . . . . . . . . . . 32
2.3.2 Difference method . . . . . . . . . . . . . . . . . . . . . 34

i
2.3.3 Derivative and integral methods . . . . . . . . . . . . . . 36
2.3.4 Using complex numbers . . . . . . . . . . . . . . . . . . 37
2.4 Using generating function method . . . . . . . . . . . . . . . . . 39
2.4.1 Ordinary generating functions . . . . . . . . . . . . . . . 40
2.4.2 Exponential generating functions . . . . . . . . . . . . . 40

3 Some focuses of analysis in high school programs 45


3.1 The problems of a tangent . . . . . . . . . . . . . . . . . . . . . 45
3.1.1 Problem 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.2 Problem 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.2 The applications of the mean value theorem . . . . . . . . . . . 47
3.2.1 The mean value theorem and inequalities . . . . . . . . . 47
3.2.2 The mean value theorem and roots of equation . . . . . . 49
3.2.3 The mean value theorem and integral limit . . . . . . . 50
3.3 Methods for finding the maximal and minimal values . . . . . . 52
3.3.1 Derivative method . . . . . . . . . . . . . . . . . . . . . 52
3.3.2 Using inequality . . . . . . . . . . . . . . . . . . . . . . . 55
3.3.3 Using the root conditions of the equation . . . . . . . . . 58
3.3.4 Using the discriminant ∆ . . . . . . . . . . . . . . . . . . 60
3.3.5 Trigonometric method . . . . . . . . . . . . . . . . . . . 62

4 Some applications of polynomials 65


4.1 Some interpolation polynomials . . . . . . . . . . . . . . . . . . 65
4.1.1 Lagrange’s interpolation polynomial . . . . . . . . . . . . 65
4.1.2 Taylor’s Polynomial . . . . . . . . . . . . . . . . . . . . . 66
4.1.3 Newton’s interpolation polynomial . . . . . . . . . . . . 68
4.2 Polynomials of a particular form . . . . . . . . . . . . . . . . . 70
4.2.1 Polynomials with interger coefficients . . . . . . . . . . . 70
4.2.2 Polynomials with real and complex coefficients . . . . . . 72
4.2.3 Arithmetic polynomials . . . . . . . . . . . . . . . . . . . 73
4.2.4 Chebyshev polynomials . . . . . . . . . . . . . . . . . . . 75

Tài liệu tham khảo 79

ii
Chương 1

Some basis principles

1.1 Dirichlet’s principle


In mathematics, the pigeonhole principle states that if n pigeons are put
into m pigeonholes with n > m, then at least one pigeonhole must contain more
than one pigeon. This theorem is exemplified in real-life by truisms like " if
there are 367 people in the room, then there is at least one pair who share the
same birthday, as there are only 366 possible birthdays to choose".
The first formalization of the idea is believed to have been made by a ger-
man mathematician Gustav Lejeune Dirichlet in 1834 under the name Schub-
fachprinzip ("drawer principle" or "shelf principle"). For this reason it is also
commonly called Dirichlet’s box principle, Dirichlet’s drawer principle or sim-
ply "Dirichlet’s principle". Now, we will describe more detail this principle.

1.1.1 Pigeonhole principle: Simple form


Theorem 1. If n + 1 objects are put into n boxes, then at least one box
contains two or more objects.
Example 1.1. Among 13 people there are two who have their birthdays in
the same month
Other principles related to the pigeonhole principle:
• If n objects are put into n boxes and no box is empty, then each box
contains exactly one object.
• If n objects are put into n boxes and no box gets more than one object,
then each box has an object.
The abstract formulation of the three principles: Let X and Y be finite sets
and let f : X → Y be a map.
• If X has more elements than Y , then f is not injective.
• If X and Y have the same number of elements and f is surjective, then
f is one-to-one.
• If X and Y have the same number of elements and f is injective , then
f is surjective.

1
Example 1.2. Given n integers a1 , a2 , ..., an , not necessarily distinct, there
exist integers k and l with 0 ≤ k < l ≤ n such that the sum ak+1 +ak+2 +· · ·+al
is a multiple of n.

Solution. Consider the n integers

a1 , a1 + a2 , a1 + a2 + a3 , ..., a1 + a2 + · · · + an .

Dividing these integers by n, we have

a1 + a2 + · · · + ai = qi n + ri , 0 ≤ ri ≤ n − 1, i = 1, 2, ..., n.

If one of the remainder r1 , r2 , ..., rn is zero, say, rk = 0 then a1 + a2 + · · · + ak


is a multiple of n. If none of r1 , r2 , ..., rn is rezo, then two of them must the
same (since 1 ≤ ri ≤ n − 1 for all i), suppose rk = rl with k < l. This means
that the two integers a1 + a2 + · · · + ak and a1 + a2 + · · · + al have the same
remainder. Thus ak+1 + · · · + al is a multiple of n.

Example 1.3. Given 101 integers in interval [1; 200], there are at least two
integers such that one of them is divisible by the other.

Solution. We can see that any integer can be written in the form 2k a, where
k ≥ 0 and a is odd. The number a can be one of the 100 numbers 1, 3, ..., 199.
Therefore among the 101 numbers chosen, two of them must have the same a
when they are written in the above form, say 2r .a and 2s .a with r ̸= s. If r < s
, then the first one divides the second. If r < s, then the second one divides
the first

1.1.2 Pigeonhole principle: Strong form


Theorem 2. Let q1 , q2 , ..., qn be positive integers. If

q1 + q2 + · · · + qn − n + 1

objects are put into n boxes, then either the 1st box contains at least q1 objects,
or the 2nd box contains at least q2 objects,..., the nth box contains at least qn
objects.

In elementary mathematics the strong form of the pigeonhole principle is


most often applied in the special case when q1 = q2 = · · · = qn = r. In this
case, the principle becomes:

2
• If (n − 1)r + 1 objects are put into n boxes then at least one of the boxes
contains r or more of the objects. Equivalently,
• If the average of n nonnegative integers a1 , a2 , ..., an is greater than r − 1
i.e
a1 + a2 + · · · + an
>r−1
n
then at least one of the integers is greater than or equal to r. Using the Dirich-
let’s principle, we obtain a wonder result following
Theorem 3. Every sequence a1 , a2 , ..., an2 +1 of n2 + 1 distinct real numbers
contains either an increasing subsequence of length n + 1 or a decreasing sub-
sequence of length n + 1.
Proof. Assume there is no increasing subsequence of length n + 1. We suffices
to prove that there must be a decreasing subsequence of length n + 1.
Let lk be the length of the longest increasing subsequence which begins with
ak , 1 ≤ k ≤ n2 + 1. Since it is assumed that there is no increasing subsequence
of length n + 1, we have 1 ≤ lk ≤ n for all k. By strong form of the pigeonhole
principle, n + 1 of the n2 + 1 integers l1 , l2 , ..., ln2 +1 must be equal, say,

lk1 = lk2 = · · · = lkn+1

where 1 ≤ k1 < k2 < · · · < kn+1 ≤ n2 + 1. If there is one ki (1 ≤ i ≤ n) such


that aki < aki+1 , then any increasing subsequence of length lki+1 beginning with
aki+1 will result a subsequence of length lki+1 + 1 beginning with aki by adding
aki in the front; so lki > lki+1 , which is contradictory to lki = lki+1 . Thus, we
must have
ak1 ≥ ak2 ≥ · · · ≥ akn+1 ,
which is a decreasing subsequence of length n + 1.

Exercise
Exercise 1.1. Given 52 integers a1 , a2 , ..., a52 . Prove that thers exist i, j,1 ≤
i < j ≤ 52 such that ai + aj or ai − aj is divisible by 100.
Exercise 1.2. Takes 51 arbitrary points in a square of length of side 1m. Show
1
that there exist 3 points in a circle of length of radius r such that r < m.
7
Exercise 1.3. Given n + 1 distinct integers a1 , a2 , ..., an+1 , there exist integers
k, l with 1 ≤ k < l ≤ n such that ak − al is divisible by n.

3
Exercise 1.4. Show that from a infinite sequence of distinct real numbers, we
can chose either an increasing subsequence of arbitrary length or a decreasing
subsequence of arbitrary length.
Exercise 1.5. Given n + 1 integers in interval [1; 2n], there are at least two
integers such that one of them is divisible by the other.
Exercise 1.6. Show that there exist a positive integer in the form 20122012...2012
which is divisible by 2013.
Exercise 1.7. Show that there exist a positive integer k such that 29k ≡ 1
(mod 1)05 .
Exercise 1.8. Given Fibonacci sequence (Fn )n≥0 defined by
F0 = F1 = 1; Fn+2 = Fn+1 + Fn with n ≥ 0.
Proved that among 100.000.001 first Fibonacci numbers Fj , there exist Fi such
that Fi ≡ 0 (mod 1)04 .

1.2 Discrete extremal principle


Let A be finite nonempty subset of real numbers. Then discrete extremal
principle states that:
ˆ A has a minimal element min A and a maximal element max A.

ˆ Always arrange elements in A by increasing or decreasing order.


Equivalently, the discrete extremal principle can be formulated as follows:
• A real valued function f on a finite set S has a maximal value and a
minimal value.
Example 1.4. Show that the people at a party can be divided into two groups
and sent to two different rooms in such a way that, for every person in either
room, at least half that person’s friends at the party are in the other room
(You may assume that friendship is a symmetric relation).
Solution. Let m the the number of all pairs (P, Q) of people such that P and
Q are in different rooms and P and Q are friends. We may assume that m is
maximal over all possible ways of dividing the people in two groups. Suppose
some person P has aP friends in his own room and bP friends in the other
room. If P would move to the other room then we have to add bP − aP to m.
By our maximality assumption on m, we get that bP ≤ aP for all P.

4
1.2.1 The existence of the configuration
Example 1.5. Show that there are infinite prime numbers in the form 4n +
3, n ∈ N.
Solution. Assume that there is finite prime numbers in the form 4n + 3, say
p1 = 3 < p2 = 7 < .... < pk . Consider a numbers

A = 4p1 p2 ...pk − 1,

then A ≡ 3 (mod 4). Thus A must have a divisor in the form 4n + 3, say q.
Deduce, there exist the pi such that q = pi . From pi | A and pi | A + 1, we
have pi | 1. But this contradicts.
Example 1.6. Prove that 2n − 1 is not divisible by n for every integer n > 1.
Solution. Assume that there exist an integer n > 1 and n | 2n − 1. So, n is
odd. Say p be least prime divisor of n, then p is odd. By Fermat’s Theorem,
p | 2p−1 − 1. Suppose that k is the least positive integer such that p | 2k − 1,
then 1 < k ≤ p − 1 < p. It is easy to see that k|n. Thus, there exist a prime
divisor of n which is less than p. But this contradicts.

1.2.2 Find discrete extremes


Example 1.7. Let m and d be integers with m ≥ d ≥ 2. Assume that
x1 , x2 , ..., xd are positive integers such that x1 + x2 + · · · + xd = m. Find
the minimum of the expression

S = x21 + x22 + · · · + x2d .

Solution. Denote G as a set of all values of S. We see that G ̸= ∅ and


cardinality of G is a finite set. By discrete extremal principle, there exist the
least element of G, say N. Suppose that (a1 , a2 , ..., ad ) are positive integers
such that a1 + a2 + · · · + ad = m and a21 + a22 + · · · + a2d = N. We will prove
that |ai − aj | = 0 or 1 for all 1 ≤ i, j ≤ d.
Assume that a1 − a2 = a > 1, then chose b = a1 − 1, c = a2 + 1. We
have b, c ∈ N∗ , a1 + a2 = b + c and b2 + c2 < a21 + a22 . Thus, we find the
positive integer sequence b, c, a3 , ..., ad such that b + c + a3 + · · · + ad = m and
b2 + c2 + a23 + · · · + a2d < N. This contradicts to chose N.
Without loss of generality, we may assume a1 ≤ a2 ≤ · · · ≤ ad and write
m = dq + r, 0 ≤ r < d. Thus, deduce a1 = · · · = ad−r = q and ad−r+1 = · · · =
ad = q + 1. So, the minimum of S is N = (d − r)q 2 + r(q + 1)2 .

5
Example 1.8. Let m > 3 and d be positive integers. Assume that x1 , x2 , ..., xd
are positive integers such that x1 x2 · · · xd = m. Find the maximum of the
expression
S = x31 + x32 + · · · + x3d .

Solution. Denote A as a set of all values of S. We see that A ̸= ∅ and


cardinality of A is a finite set. By discrete extremal principle, there exist the
greatest element of A, say M. Suppose that (a1 , a2 , ..., ad ) are positive integers
such that a1 a2 · · · ad = m and a31 + a32 + · · · + a3d = M.
Without loss of generality, we may assume a1 ≥ a2 ≥ · · · ≥ ad , then a1 = m
and a2 = · · · = ad = 1. If a1 < m, then a2 > 1. Chose a = a1 a2 and b = 1 then
aba3 ...ad = m. We have

a3 + b3 + a33 + · · · + a3d = (a1 a2 )3 + 1 + a33 + · · · + a3d > a31 + a32 + · · · + a3d = M.

This contradicts to chose M. So a1 = m; a2 = · · · = ad = 1 and therefore


M = m3 + (d − 1).

1.2.3 Create order relation based on equality


Example 1.9. Find the prime numbers a, b, c such that

abc < ab + bc + ac. (1.2.1)

Solution. Without loss of generality, we can assume a ≤ b ≤ c. Then, ab +


bc + ca ≤ 3bc, deduce abc < 3bc ⇒ a < 3, a is prime number, thus a = 2.
Replace a = 2 into the inequality (1.2.1), we have 2bc < 2b + 2c + bc. Thus
1 1 1
bc < 2(b + c) ⇒ + > . Deduce b < 5, b is prime number, so b = 2 or
b c 2
b = 3. If b = 2, then every prime number c is satisfied. If b = 3, then c = 3 or
c = 5.

Example 1.10. Let a, b, c, d be distinct real numbers. Solve the system of


equation following


|a − b|y + |a − c|z + |a − d|t = 1

|b − a|x + |b − c|z + |b − d|t = 1
(1.2.2)


|c − a|x + |c − b|y + |c − d|t = 1
|d − a|x + |d − b|y + |d − c|z = 1.

6
Solution. Due to the role of a, b, c are the same, without loss of generality, we
may assume a > b > c > d, then the system of equations (1.2.2) is transformed
to the system of equations


 (a − b)y + (a − c)z + (a − d)t = 1

(a − b)x + (b − c)z + (b − d)t = 1
(1.2.3)


 (a − c)x + (b − c)y + (c − d)t = 1
(a − d)x + (b − d)y + (c − d)z = 1.

In the system of equations (1.2.3), the first equation minus the second equation
; the second equation minus the third equation ; the third equation minus the
fourth equation, we obtain the system of equations


 (a − b)(−x + y + z + t) = 0

(b − c)(−x − y + z + t) = 0
(1.2.4)


 (c − d)(−x − y − z + t) = 0
(a − d)x + (b − d)y + (c − d)z = 1.

So,


 −x + y + z + t = 0

−x − y + z + t = 0
(1.2.5)


 −x − y − z + t = 0
(a − d)x + (b − d)y + (c − d)z = 1.

1
Thus, we obtain roots of the system y = z = 0; x = t = .
a−d
Example 1.11. Let a1 , a2 , ..., ad , b1 , b2 , ..., bd be positive real numbers and
d x
P k
x1 , x2 , ..., xd be variable of nonnegative real numbers such that = 1.
k=1 k a
Pd
Find the minimum and the maximum of the expression A = bk x k .
k=1

Solution. Suppose that

a1 b1 = min{a1 b1 , a2 b2 , ..., ad bd }

and
ad bd = max{a1 b1 , a2 b2 , ..., ad bd }.

7
Then,
d d
X xk X a1 b1 xk
a1 b 1 = a1 b 1 =
k=1
ak k=1
ak
d d d
X ak b k x k X ad b d x k X xk
≤ ≤ = ad b d = ad b d .
k=1
ak k=1
ak k=1
ak
Thus, a1 b1 ≤ A ≤ ad bd . Moreover, x1 = a1 , x2 = · · · = xd = 0 then A = a1 b1 ;
x1 = · · · = xd−1 = 0, xd = ad then A = ad bd . So, the minimum of A is a1 b1 and
the maximum of A is ad bd .

Exercise
Exercise 1.9. Show that there are infinite prime numbers of the form 6n +
5, n ∈ N.

Exercise 1.10. Given 7 line segments whose lengths are in interval (10; 100).
Prove that there exist 3 line segments forming the sides of a triangle.

Exercise 1.11. Let m and d be integers with m ≥ d ≥ 2. Assume that


x1 , x2 , ..., xd are positive integers such that x1 + x2 + · · · + xd = m. Find the
minimum and maximum of the expression
d
X
S= kxk .
k=1

Exercise 1.12. Solve in positive integers the equation xyz = 9 + x + y + z.

Exercise 1.13. Find 12 positive integers a1 , ..., a12 such that

a1 + · · · + a12 = a1 a2 ...a12 .

Exercise 1.14. Create or collect more than 5 exercises.

1.3 Infinite descent principle


Infinite descent principle is sometime called the Fermat’s method of infinite
descent (FMID). Fermat was one of the first mathemticians to use a method
"infinite descent" to prore the problems of diophantine equations. Let P be

8
a property concerning the nonnegative integers and let (P (n))n≥1 be the se-
quence of propositions:
P (n): "n satisfies property P."
• P (m) is true for a positive integer m > k, then there must be some
smaller j, m > j ≥ k for which P (j) is true. Then P (n) is false for all n ≥ k.
The method described above is often called the finite descent method.
Fermat’s method of infinite descent can be formulated as follows:
• Let k be a nonnegative integer. Suppose that whenever P (m) is true for
an integer m > k then there must be some smaller integer j, m > j > k, for
which P (j) is true. Then P (n) is false for all n > k.
Two special cases of FMID are particularly useful in the study of Diophan-
tine equations.
FMID 1. There is no infinite sequence of nonnegative integers n1 > n2 >
··· .
We can replace FMID 1 by the following equivalent form: If n0 is the smallest
positive integer n for which P (n) is true, then P (n) is false for all n < n0 .
FMID 2. If the sequence of nonnegative integers (ni )i≥0 satisfies the in-
equalities n1 ≥ n2 ≥ · · · , then there exists i0 such that ni0 = ni0 +1 = · · · .

1.3.1 Infinite descent principle in Diophantine equations


Example 1.12. Solve in nonnegative integers the equation

x3 + 2y 3 = 4z 3 .

Solution. Note that (0, 0, 0) is a solution. We will prove that there √are √ no
other solutions. Assume that (x1 , y1 , z1 ) is a nontrivial solution. Since 3 2, 3 4
are both irrational, it is not difficult to see that x1 > 0, y1 > 0, z1 > 0.
From x31 + 2y13 = 4z13 , it follows that 2|x1 , so x1 = 2x2 , x2 ∈ N∗ . Then 4x22 +
y13 = 2z13 , and hence y1 = 2y2 , y2 ∈ N∗ . Similarly, z1 = 2z2 , z2 ∈ N∗ . We obtain
the new solution (x2 , y2 , z2 ) with x1 > x2 , y1 > y2 , z1 > z2 . Continuing this
procedure, we construct a sequence of positive integer solutions (xn , yn , zn )n≥1
such that x1 > x2 > x3 > · · · . But this contradicts.
Example 1.13. Solve the diophantine equation

8x4 + 4y 4 + 2z 4 = u4 . (1.3.1)

Solution. Note that (0, 0, 0) is a solution. We will prove that there are no
other solutions. Note that if (a, b, c, d) is solution of (1.3.1) then (|a|, |b|, |c|, |d|)
is also solution. So we may solve the equation (1.3.1) in nonnegative integers.

9
Assume that (x1 , y1 , z1 , u1 ) is a nontrivial solution. We may consider u1 > 0,
with u1 is smallest number, the other case is similar. From the equation, we see
that u1 is even, u1 = 2u2 , u2 ∈ N∗ . Then,

8x41 + 4y14 + 2z14 = 16u42 ⇒ 4x41 + 2y14 + z14 = 8u42

and hence z1 = 2z2 , z2 ∈ N. Similarly, x1 = 2x2 , y1 = 2y2 . We obtain a new


solution (x2 , y2 , z2 , u2 ) of (1.3.1) with u2 < u1 . But this contradicts.

Example 1.14. Find the maximal value of m2 + n2 if m and n are integers


between 1 and 1981 satisfying (n2 − mn − m2 )2 = 1.
(22nd IMO)

Solution. Note that (m, n) = (1, 1) satisfies the relation (n2 −mn−m2 )2 = 1.
Moreover, if m = n, then necessarily m = n = 1. Also, if a pair (m, n) satisfies
this relation and 0 < m < n, then m < n ≤ 2m. We have

(n2 − mn − m2 )2 = ((n − m)2 + mn − 2m2 )2


= ((n − m)2 − m(n − m) − m2 )2
= (m2 − m(n − m) − (n − m)2 )2 ,

which show that (n − m, m) satisfies the same relation and 0 < n − m <
m. By FMID 2, the transformation (m, n) 7→ (n − m, m) must terminate
after finite steps and it terminates only when m = n = 1. Thus, all pairs of
numbers satisfying the relation are obtained from (1, 1) by applying the inverse
transformation (m, n) 7→ (n, m + n) several times:

(1, 1) 7→ (1, 2) 7→ (2, 3) 7→ (3, 5) 7→ · · · .

The components of all such pairs are Fibonacci numbers Fn where the sequence
(Fn )n≥0 is defined by

F0 = F1 = 1, and Fn+2 = Fn+1 + Fn , n ≥ 0.

Therefore, all pairs consist of consecutive Fibonacci numbers.The biggest Fi-


bonacci number less than 1981 is F16 = 1597, so the answer to the problem is
2 2
F15 + F16 = 3524578.

10
1.3.2 Infinite descent principle in Geometry
Example 1.15. Let n ≥ 3, n ̸= 4 be an integer. Prove that there is non regular
polygon of n vertices such that all vertices have integer coordinates.
Solution. If n = 3, we show that there does not exist an equilateral triangle
ABC whose vertices are integer coordinates. Assuming the opposite, then area
of ABC can also be expressed in terms of dot products as follows:
1 −→ 2 −→ 2 −→ −→
q
SABC = |AB| |AC| − (AB.AC)2
2
1
= |(xA − xC )(yB − yA ) − (xA − xB )(yC − yA )|.
2

2 3
Then, SABC is rational. However, SABC = AB is irrational from AB 2 ∈ N∗
√ 4
3
and is irrational. This contradicts.
4
Also, if there is a regular polygon of 6 vertices such that all vertices have integer
coordinates, then there exist an equilateral triangle ABC whose vertices are
integer coordinates. This contradicts.
Assume that there exist a regular polygon of n integer vertices n ̸= 3, 4, 6. We
can chose the regular polygon A1 A2 ...An for which length of its edge is the
smallest. Takes points B1 , B2 , ..., Bn such that
−−−→ −−−→ −−−→ −−−→ −−−−−−→ −−−→ −−−→ −−−→
A1 B1 = A2 A3 ; A2 B2 = A3 A4 ; ...; An−1 Bn−1 = An A1 ; An Bn = A1 A2 .
From n ̸= 3, 4, 6, we see that B1 , ..., Bn are distinct points and they have integer
coordinates. Moreover B1 B2 ...Bn is the regular polygon and B1 B2 < A1 A2 , this
contradicts.

Exercise
Exercise 1.15. Solve the diophantine equation
x2 + y 2 + z 2 + u2 = xyzu.
Exercise 1.16. Solve the diophantine equation
x4 + 2y 4 = z 4 .
Exercise 1.17. Solve the diophantine equation
x3 = 3y 3 + 9z 3 .

11
Exercise 1.18. Prove that there are non rational numbers a, b, c such that

a2 + b2 + c2 = 7.

Exercise 1.19. Know that if a function f is continuous on the compact set


M , then f have the minimum and the maximum on M. Prove that the center
of a regular polygon is the point that sum of distances from this points to all
vertices is the smallest value.

Exercise 1.20. Create or collect 10 exercises.

1.4 The basis counting principles


Let A be a finite set. Denote N (A) by the cardinality of A.

1.4.1 Addition, multiplication counting principles


ˆ Addition counting principle states: If A1 , A2 , ..., Am are finite sets and
pairwise disjoint sets. Then
m
X
N (A1 ∪ A2 ∪ · · · ∪ Am ) = N (Ai ).
i=1

ˆ Multiplication counting principle states: If A1 , A2 , ..., Am are finite sets


then
N (A1 × A2 × · · · × Am ) = N (A1 )N (A2 )...N (Am ).

Example 1.16. The United States Postal Service currently uses 5−digit zip
codes in most areas. How many zip codes are possible if there are no restrictions
on the digits used? How many would be possible if the first number could not
be 0?

Solution. There are 10 digits possible for the 1st digit, 10 possible digits for
the 2nd and so on. Therefore there are 10.10.10.10.10 = 100, 000 possible zip
codes. If the first number can not be a 0 then there are only 9 possible for the
1st and 10 possible for the remaining four 9.10.10.10.10 = 90, 000 possible zip
codes.

Example 1.17. Let A be a set with m objects and B be a set with n objects.
Compute the number of functions of A to B?

12
Solution. Assume that A = {a1 , a2 , ..., am }, then there is one to one between
a function from A to B with a element (f (a1 ), f (a2 ), ..., f (am )) ∈ B m . By
multiplication counting principle, the number of functions of A to B is nm .

1.4.2 Inclusion – exclusion principle


Let X be a finite set and A be a subset of X. We have
N (X \ A) = N (X) − N (A).
If A = B ∪ C then N (A) = N (B) + N (C) − N (B ∩ C). Therefore, we obtain
N (X \ A) = N (X) − [N (B) + N (C)] + N (B ∩ C).
However, if A = A1 ∪ A2 ∪ · · · ∪ Am , m ≥ 3 then the formular N (X \ A) is
much more complicate. For instance, if
m
X X
N1 = N (Ai ); N2 = N (Ai ∩ Aj );
i=1 1≤i<j≤m
X
Nk = N (Ai1 ∩ Ai2 ∩ ... ∩ Aik ); k = 3, 4, ..., n
1≤i1 <i2 <...<ik ≤m

then
Lemma 4. m
X
N (A1 ∪ A2 ∪ · · · ∪ Am ) = (−1)i−1 Nk .
i=1
Proof. Pick an element contained in the union of all sets and let A1 , .., Am
be the individual sets containing it. Obviously, m ≥ 1. Since the element is
counted precisely once bt the left-hand side of the equation, we need to show
that it is counted only once by the right-hand side. By the binomial theorem,
       
m m m m m m
(1 − 1) = − + − · · · + (−1) .
0 1 2 m
Note that 0t = 1 and rearranging terms, we have

     
m m m+1 m
1= − + · · · + (−1)
1 2 m
m+1
= N1 − N2 + · · · + (−1) Nm
and so the chosen element is counted only once by the right hand side of the
proposed equation.

13
So, if A1 , A2 , ...., Am are subsets of X, we have the principle following:

Theorem 5 (Inclusion – exclusion principle).

N [X \ (A1 ∪ A2 ∪ · · · ∪ Am )] = N − N1 + N2 − · · · + (−1)m Nm .

Example 1.18. Find the number of integer solutions of the equations

x1 + x2 + x3 + x4 = 15

which satisfy the conditions

2 ≤ x1 ≤ 6, −2 ≤ x2 ≤ 1, 0 ≤ x3 ≤ 6, 3 ≤ x4 ≤ 8.

Solution. Let y1 = x1 − 2, y2 = x2 + 2, y3 = x3 and y4 = x4 − 3. Then the


problem becomes to find the number of nonnegative integer solutions of the
equations
y1 + y2 + y3 + y4 = 12
such that
0 ≤ y1 ≤ 4, 0 ≤ y2 ≤ 3, 0 ≤ y3 ≤ 6, 0 ≤ y4 ≤ 5.
Let S be the set of all nonnegative integer solutions of the equation y1 + y2 +
y3 + y4 = 12. Let

A1 := {y1 ∈ S|y1 ≥ 5}; A2 := {y2 ∈ S|y2 ≥ 4};

A3 := {y3 ∈ S|y3 ≥ 7}; A4 := {y4 ∈ S|y2 ≥ 6}.


Then the problem is to find cardinality N (Ā1 ∩ Ā2 ∩ Ā3 ∪ Ā4 ). In fact,
   
4 + 12 − 1 15
N (S) = = = 455.
3 3
 
10
N (A1 ) = = 120,
3
 
11
N (A2 ) = = 165,
3
 
8
N (A3 ) = = 56,
3
 
9
N (A4 ) = = 84.
3

14
For the intersections of more sets,

N (A1 ∩ A2 ) = 20, N (A1 ∩ A3 ) = 1, N (A1 ∩ A4 ) = 4, N (A2 ∩ A3 ) = 4,

N (A2 ∩ A4 ) = 10, N (A3 ∩ A4 ) = 0.


And
N (A1 ∩ A2 ∩ A3 ) = 0; N (A1 ∩ A2 ∩ A4 ) = 0;
N (A1 ∩ A3 ∩ A4 ) = 0; N (A2 ∩ A3 ∩ A4 ) = 0; N (A1 ∩ A2 ∩ A3 ∩ A4 ) = 0.
Applying the inclusion-exclusion principle, we have

N (Ā1 ∩Ā2 ∩Ā3 ∪Ā4 ) = N (S\(A1 ∪A2 ∪A3 ∪A4 )) = N (S)−N1 +N2 −N3 +N4 = 69.

A well-known application of the inclusion-exclusion principle is to the com-


binatorial problem of counting all derangements of a finite set. A derangement
of a set A is a bijection from A into itself that has no fixed points. Via the
inclusion-exclusion principle we show that if the cardinality of A is n, then the
number of all derangements is

Theorem 6.
n
X (−1)k
Dn = n! .
k=0
k!

Proof. We may assume that A = {1, 2, ..., n}, then a permutation of A is called
a derangement if every integer i(1 ≤ i ≤ n) is not place at the ith position.
Dn is denoted to number of derangements of A.
Let S be the set of all permutation of A, then N (S) = n!. Let Pi be the
property that a permutation of A has the integer i in its ith position and let
Ai be the set of all permutations satisfying the property Pi where 1 ≤ i ≤ n.
Then

Dn = N (S \ (A1 ∪ A2 ∪ · · · ∪ An )) = N (Ā1 ∩ Ā2 ∩ · · · ∪ A¯n ).

For each (i1 , i2 , ..., ik ) such that 1 ≤ i1 < i2 < · · · < ik ≤ n, a permutation
of A with i1 , i2 , ..., ik fixed at the i1 th, i2 th,...,ik th position respectively can be
identified as a permutation of a set A \ {i1 , i2 , ..., ik } of n − k objects. Thus

N (Ai1 ∩ Ai2 ) ∩ ... ∩ Aik ) = (n − k)!.

15
By the inclusion-exclusion principle, we have
n
X X
Dn = N (S) + (−1)k N (Ai1 ∩ Ai2 ) ∩ ... ∩ Aik )
k=1 i1 <i2 <···<ik
n
X X
= n! + (−1)k (n − k)!
k=1 i1 <i2 <···<ik
n  
k n
X
= (−1) (n − k)!
k=0
k
n
X (−1)k
= n!
k=0
k!
n!
=≈ where n is enough large.
e

Moreover, the derangement sequence Dn satisfies the recurrence relation


Dn = (n − 1)(Dn−1 + Dn−2 ), n ≥ 3
with D1 = 0, D2 = 1.
Let A be a set with m objects and B be a set with n objects. Then the
number of functions of A to B is nm . The number of injective functions from
A to B is  
n
m! .
m
Let P (m, n) denote the number of surjective functions from A to B. What is
P (m, n)?
n
(−1)k nk (n − k)m .
P 
Theorem 7. P (m, n) =
k=0

Proof. Let S be the set of all functions from A to B. Write B = {b1 , b2 , ..., bn }.
Let Ai be the set of all functions f such that bi ̸∈ f (X) where 1 ≤ i ≤ n. Then
P (m, n) = N (Ā1 ∩ Ā2 ∩ · · · ∪ A¯n ).
For each (i1 , i2 , ..., ik ) such that 1 ≤ i1 < i2 < · · · < ik ≤ n, the set Ai1 ∩ Ai2 ) ∩
... ∩ Aik can be identified to the set of all functions f from A to the component
B − {i1 , i2 , ..., ik }. Therefore
N (Ai1 ∩ Ai2 ) ∩ ... ∩ Aik ) = (n − k)m .

16
By the inclusion-exclusion principle, we have
n
X X
P (m, n) = N (S) + (−1)k N (Ai1 ∩ Ai2 ) ∩ ... ∩ Aik )
k=1 i1 <i2 <···<ik
n
X X
m
=n + (−1)k (n − k)m
k=1 i1 <i2 <···<ik
n  
k n
X
= (−1) (n − k)m .
k=0
k

Note that P (m, n) = 0 fro m < n; we have


n  
k n
X
(−1) (n − k)m = 0 if m < n.
k=0
k

Exercise
Exercise 1.21. Find the formular of Euler phi function φ(n) by the inclusion-
exclusion principle.
Exercise 1.22. Prove the inclusion-exclusion principle by induction.
Exercise 1.23. Prove that for integers m, n ≥ 1,
  X n  
X m k n
= (−1) (n − k)m .
i +i +···+i =m;i ,...,i ≥1
i1 , ..., i n
k=0
k
1 2 n 1 n

Exercise 1.24. There are m teachers and n children, m ≥ n. Each teacher


gives one (random) child a cookie. What is the probability that all n children
get at least one cookie ?
Exercise 1.25. Create or collect 10 exercises.

1.5 The issues with extreme views


In mathematical analysis, we have the famous theorem: Suppose give a
continuous function f : M → R where M ⊂ R is a compact set. Then f
always has a maximum and a minimum.

17
1.5.1 Applications of some extreme principles in equa-
tions and inequalities
Suppose given a real function f : M ⊂ R → R. Then, we have
ˆ If f has a minimal value then
(a) f (x) ≥ c, ∀x ∈ M if and only if min f (x) ≥ c.
x∈M
(b) An inequality f (x) ≤ c has a root in M if and only if min f (x) ≤ c.
x∈M

ˆ If f has a maximal value then


(c) f (x) ≤ c, ∀x ∈ M if and only if max f (x) ≤ c.
x∈M
(d) An inequality f (x) ≥ c has a root in M if and only if max f (x) ≥ c.
x∈M

ˆ If f has a minimal value, a maximal value and f is a continuous function


then the equation f (x) = m has roots if and only if

min f (x) ≤ m ≤ max f (x)


x∈M x∈M

Example 1.19. How is property of a function f (x) = a cos x + b sin x + c with


extreme views?
√ a
Solution. Write f (x) = a2 + b2 cos(x − φ) + c with cos φ = √ . Thus
a + b2
2

√ √
min f (x) = c − a2 + b2 ; max f (x) = c + a2 + b2 .

Since this, we can deduce



(i) f (x)√≥ 0 with every x if and only if min f (x) = c − a2 + b2 ≥ 0, thus
c ≥ a2 + b2 .

(ii) f (x) ≤√0 with every x if and only if max f (x) = c + a2 + b2 ≤ 0, thus
c ≤ − a2 + b 2 .

√ only if c −
(iii) The√equation f (x) = 0 has a root if and a2 + b 2 ≤ 0 ≤
c + a + b . This is equivalent to |c| ≤ a2 + b2 .
2 2

The inequality f (x) ≤√0 has a root if and only if min f (x) = c −
(iv) √
a2 + b2 ≤ 0, thus c ≤ a2 + b2 .

√ inequality f (x) ≥ 0√has a root if and only if max f (x) = c +


(v) The
a2 + b2 ≥ 0, thus c ≥ − a2 + b2 .

18

(vi) f (x)√≡ 0 if and only if min f (x) = max f (x) = 0 ⇔ c − a2 + b 2 =
c + a2 + b2 = 0 ⇔ a = b = c = 0.

Example 1.20. Find m such that the following system of equation has solu-
tions: (
x2 − 3x + 2 ≤ 0,
(1.5.1)
x3 + x2 − 5x + 1 ≤ m.

Solution. The domain of solution of the inequality x2 − 3x + 2 is [1; 2]. So,


the system (1.5.1) has solutions if and only if the inequality

f (x) = x3 + x2 − 5x + 1 ≤ m

has solutions in [1; 2]. This is equivalent to min f (x) ≤ m. We have


x∈[1;2]

f ′ (x) = 3x2 + 2x − 5.
−5 −5
The equation f ′ (x) = 0 has two roots 1 and . Since ̸∈ [1; 2], thus
3 3
min f (x) = min{f (1), f (2)} = −2.
x∈[1;2]

So, m ≥ −2.
Example 1.21. Find a such that the following system of equation has solu-
tions
x3 − 3x2 + x − 3 ≤ a ≤ x3 − 4x2 + 6x − 9.
Solution. The domain of solution of the inequality x3 − 3x2 + x − 3 ≤ x3 −
4x2 + 6x − 9 is [2; 3]. Thus, the system of equation has solutions if and only
if f (x) = x3 − 3x2 + x − 3 ≤ a ≤ x3 − 4x2 + 6x − 9 = g(x) has solutions on
closed domain [2; 3]. This is equivalent to

min f (x) ≤ a ≤ max g(x).


x∈[2;3] x∈[2;3]

Since f (x) and g(x) are increasing functions on [2; 3], so min f (x) = f (2) =
x∈[2;3]
−5 and max g(x) = g(3) = 0. Therefore −5 ≤ a ≤ 0.
x∈[2;3]

Example 1.22. Find m such that cos 2x + m cos x + 5 ≥ 0 for all x.

19
Solution. See that cos 2x + m cos x + 5 = 2 cos2 x + m cos x + 4, so denote
t = cos x, then t ∈ [−1; 1]. We transform this problem to find m such that
f (t) = 2t2 + mt + 3 ≥ 0 for every t ∈ [−1; 1]. This is equivalent to min f (t) ≥
t∈[−1;1]
0.
m m 32 − m2
If − ∈ [−1; 1] ⇔ −4 ≤ m ≤ 4, then min f (t) = f (− ) = . In
4 t∈[−1;1] 4 8
−4 ≤ m ≤ 4
this case, min f (t) ≥ 0 if and only if 32 − m2 , thus −4 ≤ m ≤ 4.
t∈[−1;1]  ≥0
8
(1)
m
If − ̸∈ [−1; 1] ⇔ m ̸∈ [−4; 4], then
4

m ̸∈ [−4; 4]

min f (t) = min{f (−1), f (1)} ⇔ f (−1) ≥ 0
t∈[−1;1] 
f (1) ≥ 0

⇔ 4 ≤ m ≤ 6and − 6 ≤ m ≤ −4. (2)

Since (1) and (2), we obtain −6 ≤ m ≤ 6.

Example 1.23. Find m such that cos6 x + sin6 x + m cos x sin x ≥ 0 for all x.

Solution. Put t = sin 2x, then t ∈ [−1; 1]. We can transform the original
expression

3 m
cos6 x + sin6 x + m cos x sin x = − t2 + t + 1 = f (t).
4 2

The original prolem is transformed to find m such that f (t) ≥ 0 for every
t ∈ [−1; 1]. Since f (t) is the polynomial of degree 2 with leading coefficient
3
a = − < 0, then min f (t) = min f (−1), f (1). Thus
4 t∈[−1;1]

(
f (−1) ≥ 0
min f (t) ≥ 0 ⇔
t∈[−1;1] f (1) ≥ 0.

1 1
Solve this system of inequality, we obtain − ≤ m ≤ .
2 2
20
Exercise
Exercise 1.26. Similarly to Example 1.19, you study a function f (x) = ax2 +
bx + c, a ̸= 0 with extreme views?

Exercise 1.27. Given a function f (x) = cos4 x + sin4 x + a sin 4x + a. Find a


such that

(i) f (x) ≥ 0 for every x.

(ii) The inequality f (x) ≤ 0 has solutions.

Exercise 1.28. Create or collect 20 exercises.

1.5.2 Absolute values and minimum, maximum


Let a, b be real numbers. We have the following formular

(a + b) − |a − b|
min{a, b} = ;
2
(a + b) + |a − b|
max{a, b} = ;
2
Inversion, we also have

a − |b| = min{a + b, a − b}; a + |b| = max{a + b, a − b};

If f (x), g(x) are real functions, defined on a domain M ⊂ R, we have two


following problems:
• Problem 1. Find a maximal values of a function y = g(x) + |f (x)| on
M.
Obviously, y = max{g(x) + f (x), g(x) − f (x)}. Therefore

max y = max[max{g(x) + f (x), g(x) − f (x)}]


x∈M x∈M
= max{max(g(x) + f (x)), max(g(x) − f (x))}.
x∈M x∈M

Example 1.24. Find the maximal value of a function y = 3x3 − 4x2 + 2x −


4 + |x3 − 2x2 + 6x − 6| on closed domain [−10; 10].

Solution. Denote

f (x) = 3x3 − 4x2 + 2x − 4 + (x3 − 2x2 + 6x − 6) = 4x3 − 6x2 + 8x − 10;

21
and

g(x) = 3x3 − 4x2 + 2x − 4 − (x3 − 2x2 + 6x − 6) = 2x3 − 2x2 − 4x + 2.

Then
max f (x) = 3470; and max g(x) = 1762.
x∈[−10;10] x∈[−10;10]

Thus, max y = 3470.


x∈[−10;10]

Example 1.25. Find the maximal value of a function y = cos x + sin x +


|3 cos x + sin x − 1|.

Solution. Denote

f (x) = cos x + sin x + 3 cos x + sin x − 1 = 4 cos x + 2 sin x − 1,

g(x) = cos x + sin x − (3 cos x + sin x − 1) = −2 cos x + 1.


It is easy to see that
√ √
max f (x) = 42 + 22 − 1 = 2 5 − 1 and max g(x) = 3.

Therefore, max y = 2 5 − 1.

Example 1.26. Find the maximal value of a function y = ex +3x+|ex −2x−2|


on closed domain [0; 3].

Solution. Denote

f (x) = ex + 3x + ex − 2x − 2 = 2ex + x − 2

and
g(x) = ex + 3x − (ex − 2x − 2) = 5x + 2.
Since f (x), g(x) are the increasing functions on [0; 3], we have

max f (x) = 2e3 + 1 and max g(x) = 17.


x∈[0;3] x∈[0;3]

Hence, max y = 2e3 + 1.


x∈[0;3]

22
• Problem 2. Find a minimal values of a function y = g(x) − |f (x)| on
M.
Obviously, y = min{g(x) + f (x), g(x) − f (x)}. Therefore

min y = min[min{g(x) + f (x), g(x) − f (x)}]


x∈M x∈M
= min{min(g(x) + f (x)), min(g(x) − f (x))}.
x∈M x∈M

Example 1.27. Find the minimal value of a function y = 3x3 − 4x2 + 2x −


4 − |x3 − 2x2 + 6x − 6| on closed domain [10; 10].

Solution. Denote

f (x) = 3x3 − 4x2 + 2x − 4 + (x3 − 2x2 + 6x − 6) = 4x3 − 6x2 + 8x − 10;

and

g(x) = 3x3 − 4x2 + 2x − 4 − (x3 − 2x2 + 6x − 6) = 2x3 − 2x2 − 4x + 2.

Then
min f (x) = −4690 and min g(x) = −2238.
x∈[−10;10] x∈[−10;10]

Thus, min y = −4690.


x∈[−10;10]

Example 1.28. Find the minimal value of a function y = cos x + sin x −


|3 cos x + sin x − 1|.

Solution. Denote

f (x) = cos x + sin x + 3 cos x + sin x − 1 = 4 cos x + 2 sin x − 1,

g(x) = cos x + sin x − (3 cos x + sin x − 1) = −2 cos x + 1.


It is easy to see that
√ √
min f (x) = − 42 + 22 − 1 = −2 5 − 1 and min g(x) = −1.

Therefore, min y = −2 5 − 1.

Example 1.29. Find the minimal value of a function y = ex +3x−|ex −2x−2|


on closed domain [0; 3].

23
Solution. Denote

f (x) = ex + 3x + ex − 2x − 2 = 2ex + x − 2

and
g(x) = ex + 3x − (ex − 2x − 2) = 5x + 2.
Since f (x), g(x) are the increasing functions on [0; 3], we have

min f (x) = f (0) = 0 and min g(0) = 2.


x∈[0;3] x∈[0;3]

Hence, min y = 0.
x∈[0;3]

Exercise
Exercise 1.29. Find the maximal value of following functions

y = cos 3x + | cos 3x − 2 cos 2x + cos x − 1|;

y = sin 3x + |1 + sin x − sin 3x|.

Exercise 1.30. Find the minimal value of following functions

y = sin 3x + 2 sin x − | sin 3x − sin x − 1|;

y = cos4 x + sin4 x − sin 2x − | cos6 x + sin6 x − 3 sin 2x − 1|.

Exercise 1.31. Create 5 exercises

Exercise 1.32. Find other applications of above prolems.

24
Chương 2

Primary problems of the number


sequences

2.1 The finite sum can’t be represented by al-


gebraic functions
2.1.1 Rational function
Definition 1. Let p(x) and q(x) be polynomials over the field K and q(x) ̸= 0
p(x)
then is said to a rational function.
q(x)

Theorem 8. The expression f (x) of a variable x is not a rational function if


one of the following conditions occurs:
f (x)
(i) lim f (x) = ∞ and lim = 0.
x→∞ x→∞ x
(ii) lim f (x) = 0 and lim xf (x) = ∞.
x→∞ x→∞

(iii) f (x) ∈ Q for all x ∈ Q and lim f (x) is irrational.


x→∞

Pn 1
Example 2.1. For given f (n) = . Prove that f (n) is not the rational
i=1 i
function.

Solution. We have lim f (n) = ∞. By induction, we also prove that f (n) <
n→∞
√ f (n)
2n + 1. So, lim = 0. Apply Theorem 8(i), we deduce that f (n) is not
n→∞ n
the rational function.
Pn (−1)i
Example 2.2. For given g(n) = . Prove that g(n) is not the rational
i=1 i
function.

25
P∞ (−1)i
Solution. We know that the Leibnitz series lim g(n) = = − ln 2
n→∞ i=1 i
and ln 2 is an irrational number. Thus lim g(n) is an irrational number. By
n→∞
Theorem 8(iii), we deduce that g(n) is not the rational function.

Exercise
n
P 1
Exercise 2.1. Prove that the expression f (n) = where a, b are
i=1 ak + b
positive integers is not the rational function.
Exercise 2.2. Create or collect 10 rational sequence (un )n≥1 such that S(n) =
n
P
un is not the rational function.
i=1

2.1.2 Algebraic function


Definition 2. The expression f (x) of a variable x is said to an algebraic
function over the field K if there exists no simultaneously zero polynomials
n
Pi (x)(f (x))i = 0.
P
P0 (x), ..., Pn (x) ∈ K[x] such that
i=0

Definition 3. An algebraic number is a number that is a root of a non-zero


polynomial in one variable with rational coefficients. Numbers such as π that
are not algebraic are said to be transcendental.
Notice that almost all real and complex numbers are transcendental. (Here
"almost all" has the sense "all but a countable set".)
Theorem 9. Let f (x) be an expression of a variable x. If lim f (x) is tran-
x→∞
scendental number, then f (x) is not an algebraic function over the rational
field Q.
Pn 1
Example 2.3. Suppose given f (n) = . Prove that f (n) of a variable n is
i=1 i!
not an algebraic function over the rational field Q.
1 π2 π2
Solution. We have lim f (n) = lim 2 = . Since is a transcendental,
n→∞ x→∞ i 6 6
by Theorem 9 we deduce D(n) is not the algebraic function over Q.
Pn 1
Example 2.4. Suppose given D(n) = 2
. Prove that D(n) is not an alge-
i=1 i
braic function over the rational field Q.

26
1
Solution. We have lim D(n) = lim = e. Since e is a transcendental, by
n→∞ x→∞ i!
Theorem 9 we deduce f (n) is not the algebraic function over Q.

Exercise
n 1
P (−1)i
Exercise 2.3. Suppose given U (n) = 4
, V (n) = lim . Prove that
i=1 i x→∞ i!
U (n) and V (n) are not algebraic functions over the rational field Q.
Exercise 2.4. Create or collect 10 rational sequence (un )n≥1 such that S(n) =
n
P
un is not an algebraic function.
i=1

Exercise 2.5. Prove that the function cos x is a transcendental expression


over the real field R (i.e not an algebraic function over the real field R).

2.2 Some types of number sequences


2.2.1 Affine sequences
Definition 4. A sequence of number (un )n≥0 defined by un+1 = aun + b, a, b ∈
K ⊂ C is said to be an affine sequence.
Proposition 10. Given (un )n≥0 be an affine sequence.
a b
(i) un = u0 + nb if a = 1 and un = an (u0 + )− if a ̸= 1.
b−1 a−1
n
P un+1 − nb − u0
(ii) Put S(n) = un , then S(n) = if a ̸= 1.
i=1 a−1

2.2.2 Linear recurrence sequences of order two


Definition 5. A linear real sequence (un )n≥0 of order two is defined by un+2 =
aun+1 + bun , n ≥ 0 (∗ ) where a, b ∈ R.
Proposition 11. For given Da,b = {(un )n≥0 |un+2 = aun+1 + bun } where a, b
are real constant , then Da,b is the R− vector space of dimension 2.
Proof. We see that a sequence (0)n≥0 ∈ Da,b , thus Da,b ̸= ∅. Pick x, y ∈ R and
(un )n≥0 , (vn )n≥0 ∈ Da,b , then

xun+2 +yun+2 = x(aun+1 +bun )+y(aun+1 +bun ) = a(xun+1 +yun+1 )+b(xun +yvn ).

27
Hence, x(un )n≥0 + y(vn )n≥0 ∈ Da,b , so Da,b (
is the R− vector space.
U0 = 1, U1 = 0
Suppose (Un )n≥0 , (Vn )n≥0 ∈ Da,b such that . Then, if
V0 = 0, V1 = 1
(
xU0 + yU1 = 0
x(Un )n≥0 + y(Vn )n≥0 = (0)n≥0 , x, y ∈ R ⇒ .
xV0 + yV1 = 0
Deduce x = y = 0. Thus, (Un )n≥0 , (Vn )n≥0 are linear independence in Da,b .
Pick an arbitrary sequence (un )n≥0 ∈ Da,b , by induction it is easy to check
(un )n≥0 = u0 (Un )n≥0 + u1 (Vn )n≥0 . Thus (Un )n≥0 , (Vn )n≥0 is a base of Da,b .
Now, apply Proposition 11, we may find all sequences satisfying (∗ ). Con-
sider an following equations
t2 − at − b = 0 (∗∗ ),
this equation is said to an characteristic equation of (∗ ). Discriminant of this
equation is ∆ = a2 + 4b.
If ∆ > 0, then (∗∗ ) has two distinct real roots α1 and α2 . It is easy to check
that (α1n )n≥0 and (α2n )n≥0 is an base of Da,b . So the general solution of (∗ ) is
un = xα1n + yα2n , n ≥ 0 where x, y are arbitrary real number. Note that x, y
will be defined if know u0 and u1 .
If ∆ = 0, then (∗∗ ) has double real roots α. It is easy to check that (αn )n≥0
and (nαn−1 )n≥0 is an base of Da,b . So the general solution of (∗ ) is un =
xαn + ynαn−1 , n ≥ 0 where x, y are arbitrary real number.
If ∆ < 0, then (∗∗ ) has two complex roots, no real numbers, z1 and z2 where
z1 and z2 are conjugated. Now, we consider (∗ ) over the complex field C and
denote A by the set of all these complex sequences. We also deduce that A is
a vector space of dimension two over C. We also obtain a general solution of
(∗ ) in C is (xz1n + yz2n )n≥0 , x, y ∈ C. Put z = z1 , we will show that
Da,b = {xz n + xz n |x ∈ C}.
Obviously, {xz n + xz n |x ∈ C} is the set of real sequences of A, thus {xz n +
xz n |x ∈ C} ⊂ Da,b . Inversion, assume that (un )n≥0 ∈ Da,b , then (un )n≥0 ∈ A.
So, there exists complex numbers c, d such that
un = cz1n + dz2n = cz n + dz̄ n , n ≥ 0.
Since u0 , u1 ∈ R, deduce c + d, cz + dz̄ ∈ R. We obtain d = c̄. Therefore
un = cz n + cz n for every n ≥ 0. So,
Da,b = {cz n + cz n |x ∈ C} = {2 Rez(xz n )|x ∈ C}.

28
1
Write x = (p + iq), p, q ∈ R, z = r(cos α + i sin α) where r = |z|, α = Arg(z),
2
then

2 Rez(xz n ) = Rez((p + iq)rn (cos nα + i sin nα)) = rn (p cos nα − q sin nα).

Thus,
Da,b = {rn (p cos nα − q sin nα)|p, q ∈ R}.
So, un = rn (p cos nα − q sin nα) with (p, q) ∈ R2 and n ≥ 0.

Example 2.5. Let (Fn )n≥0 be a Fibonacci sequence defined by F0 = 1; F1 = 1


and Fn+2 = Fn+1 + Fn , n ≥ 0. Find the genenal term Fn ?

2 1± 5
Solution. The characteristic equation t −t−1 = 0 has two real roots ,
2
so √ !n √ !n
1+ 5 1− 5
Fn = x +y for every n ≥ 0.
2 2

x + y √ !

√ !
=0
Since F0 = 1, F1 = 1, we deduce 1+ 5 1− 5
x

2
+y
2
= 1.
 √
x = 5

Solve this system of equation, we obtain 5√ . So, the genenal term
y = − 5

√ √ !n √ 5!n !
5 1+ 5 1− 5
of sequence is Fn = − .
5 2 2

Example 2.6. Let (Qn )n≥0 be a sequence defined by Q0 = 1; Q1 = 1 and


Qn+2 = 4Qn+1 − Qn , n ≥ 0. Find the genenal term Qn ?

Solution. The characteristic equation t2 − 4t + 4 = 0 has double real roots


t = 2, so
Qn = x2n + yn2n−1 for every n ≥ 0.
(
x =1
Since Q0 = Q1 = 1, we deduce Solve this system, we obtain
2x + y = 1.
x = 1 and y = −1. Thus, Qn = 2n − n2n−1 for n ≥ 0.

29
Example 2.7. Let (Pn )n≥0 be a sequence defined by P0 = 0; P1 = 1 and
Pn+2 = −2Pn+1 − 4Pn , n ≥ 0. Find the genenal term Pn ?
Solution. The characteristic equation t2 + 2t + 4 = 0 has two complex roots
√ √ 2π
z = −1 + 3i and z̄ = −1 − 3i. Since |z| = 2 and Arg z = , thus the
3
2π 2π
genenal term of the original sequence is Pn = 2n (x cos + y sin ) for n ≥ 0.
 3 3
x = 0
Since P0 = 0 and P1 = 1, we have 2π 2π . Solve this
2(x cos + y sin ) = 1
√ 3 √ 3
3 3 2π
system, we obtain x = 0 and y = . So, Pn = 2n sin for n ≥ 0.
3 3 3

Exercise
Exercise 2.6. Let (un )n≥0 be a sequence defined by u0 = 0; u1 = 1 and
2un+2 = 2un+1 − un , n ≥ 0. Find the genenal term un ?

p Let (un )n≥0 be a sequence defined by u0 = a > 0; u1 = b > 0


Exercise 2.7.
and un+2 = 3 u2n un+1 , n ≥ 0. Find the genenal term un ?
Exercise 2.8. Let (un )n≥0 be a sequence defined by u0 = a > 0; u1 = b > 0
2un un+1
and un+2 = , n ≥ 0. Find the genenal term un ?
un + un+1
Exercise 2.9. Let (un )n≥1 be a sequence defined by u1 = u2 = u3 = 1 and
1 + un+1 un+2
un+3 = , n ≥ 0. Prove that un is a possitive integer for all n ≥ 1.
un
Exercise 2.10. Create or collect 20 exercises.
Exercise 2.11. Prove Proposition 11 without using vector space.

2.2.3 Sequences un+1 = f (un ).


In this section, we study some sequences formed un+1 = f (un ) where f :
M → M is an increasing function or a decreasing function or a continuous
function on M .
ˆ If f is an increasing function on M then

– (un )n≥0 is an increasing sequence if u1 ≥ u0 ,

30
– (un )n≥0 is an decreasing sequence if u1 ≤ u0 .

ˆ If f is a decreasing function on M then f 2 = f ◦ f is an increasing


function, so

– (u2n )n≥0 is an increasing sequence and (u2n+1 )n≥0 is an decreasing


sequenceif u2 ≥ u0 ,
– (u2n )n≥0 is an decreasing sequence and (u2n+1 )n≥0 is an increasing
sequenceif u2 ≤ u0 .

ˆ Let f be continuous on M and M be compact in R. Assume the sequence


(un )n≥0 converges to α then

– α ∈ M and lim f (un ) = f (α).


n→∞

– α = f (α) i.e α ∈ M is a root of the equation f (x) = x.


un
Example 2.8. Consider the sequence (un )n≥0 with u0 = 1 and un+1 =
u2n+1
for n ≥ 0.

Solution. It is easy to see that un > 0 for n ≥ 0. On the other hand, since

un u3n
un+1 − un = − un = − < 0,
u2n + 1 u2n + 1

thus (un )n≥0 is the strictly increasing sequence. See that un ∈ [0; 1], thus there
a
exists lim un = a ⇒ a = 2 . So a = 0.
n→∞ a +1
Example 2.9. Consider the sequence (un )n≥0 with u0 = a > 0 and un+1 =
1 2
(u + 8) for n ≥ 0.
6 n
1 2
Solution. Put f (x) = (x + 8), then the equation f (x) = x has two roots
6
x = 2 and x = 4. Since f is the increasing function from [0; ∞) → [0; ∞)
and f ([0; 2]) ⊂ [0; 2]; f ([2; 4]) ⊂ [2; 4]; f ([4; ∞)) ⊂ [4; ∞). So, we consider the
following cases:
1
If a ∈ [0; 2), then u1 = (a2 + 8) ≥ a = u0 . Thus, it is easy to see that (un )n≥0
6
is an increasing sequence. So lim un = 2.
n→∞
1 2
If a ∈ [2; 4), then u1 = (a + 8) ≤ a = u0 . Thus, it is easy to see that (un )n≥0
6
31
is a decreasing sequence. So lim un = 2.
n→∞
If a = 4, then un = 4 for n ≥ 0. So lim un = 4.
n→∞

1 2
If a ∈ [4; ∞), then u1 = (a + 8) > a = u0 . Thus, it is easy to see that
6
(un )n≥0 is an increasing sequence. So, if lim un = b then b > 4 and f (b) = b.
n→∞
This contradicts. So (un )n≥0 is not to converge.

Exercise
Exercise 2.12. Consider sequences (un )n≥0 defining by
1 b2
1) u0 = a > 0 and un+1 = (un + ) with b > 0.
6 un
2
2) u0 = 1 and un+1 = 1 − .
un
3) u0 = a and un+1= √
3
7un −6 .

6
4) u0 = a ≥ 0 and un+1 = .
2 + u2n

2.3 The methods of finite determination


2.3.1 Interpolating method
n
P
Theorem 12. For give a real sequence (un )n≥1 and S(n) = ui . Then S(n)
i=1
is a polynomial of degree d > 0 with variable n if and only if un is a polynomial
of degree d − 1 with variable n.

Proof. We have S(n) − S(n − 1) = un .


(⇒) : Assume that S(n) = ad nd +G(n)(deg G(n) < d) is a polynomial of degree
d > 0. Then, un = S(n) − S(n − 1) = aad nd−1 + H(n)(deg H(n) < d − 1) is a
polynomial of degree d − 1.
(⇐) : We will prove that

Gk (n) = 1k + 2k + · · · + nk , k ∈ N (∗ )

is a polynomial of degree k + 1 by induction.


It is easy to check that (∗ ) is true with k = 0 and k = 1. Assume that (∗ ) is

32
true for every i ≤ k. We have
ik+1 − (i − 1)k+1 = ak ik + ak−1 ik−1 + · · · + a1 i + a0 ,
thus n
X
k+1
n = [ik+1 − (i − 1)k+1 ]
i=1
= ak Gk (n) + ak−1 Gk−1 (n) + · · · + a1 G1 (n) + a0 n.
Since deg nk+1 = k + 1 and deg(ak−1 Gk−1 (n) + · · · + a1 G1 (n) + a0 n) ≤ k, we
deduce deg Gk (n) = k + 1.
Now, assume that un = bd−1 nd−1 + · · · + b1 n + b0 , bd−1 ̸= 0, then
n
X
S(n) = ui = bd−1 Gd−1 (n) + · · · + b1 G1 (n) + b0 n.
i=1

From above proof, we have deg Gd−1 (n) = d and deg Gi (n) = i with i < d.
Thus, deg S(n) = d.
Theorem 13. If Pd (n) is a polynomial of degree d with variable n, then
d+1 d

(n − 1)...(n − d)(n − d − 1) X i−1
Pd (n) = (−1)d+1−i Pd (i).
d! i=1
n−i
Proof. Consider as an exercise.

n
i2 .
P
Example 2.10. Compute the sum S(n) =
i=1

Solution. By Theorem 12, we have deg S(n) = 3. From S(1) = 1, S(2) =


5, S(3) = 14, S(4) = 30, by Theorem 13, we obtained
n(n + 1)(2n + 1)
S(n) = .
6

n
i4 .
P
Example 2.11. Compute the sum S(n) =
i=1

Solution. By Theorem 12, we have deg S(n) = 5. Compute S(1), S(2), ..., S(6)
respectively and apply Theorem 13, we obtained
n(n + 1)(2n + 1)(3n2 + 3n − 1)
S(n) = .
30

33
Exercise
Exercise 2.13. Use the above method, compute the following sums:
n
X n
X n
X
3 5
i; i; i6 .
i=1 i=1 i=1

Exercise 2.14. Compute the following sums:


n
X n
X n
X
2
(2i + 1) ; i(i + 1)(i + 2); i(i + 1)(i + 2)(i + 3).
i=1 i=1 i=1

Exercise 2.15. Create or collect 10 exercises.

2.3.2 Difference method


Purpose of difference methods is to find a function f (n) of variable n such
that
un = f (n) − f (n − 1).
n
P Pn
Therefore, S(n) = ui = S(n) = (f (i) − f (i − 1)) = f (n) − f (0).
i=1 i=1

n
P 1
Example 2.12. Compute the sum .
j=1 j(j + 1)(j + 2)

1
Solution. Consider the function f (j) = , then
2j(j + 1)

1
f (j) − f (j + 1) = .
j(j + 1)(j + 2)

Therefore n n
X 1 X
= [f (j) − f (j + 1)]
j=1
j(j + 1)(j + 2) j=1

n3 + 3n2 + 2n − 2
= f (1) − f (n) = .
4n(n + 1)(n + 2)

n
P
Example 2.13. Compute the sum sin kx.
k=1

34
1 1 x
Solution. We have cos(k + )x − cos(k − )x = −2 sin kx sin . Thus, put
2 2 2
1
cos(k + )x
2
f (x) = x , then f (k) − f (k − 1) = sin kx. So
2 sin
2
n
X
S= [f (k) − f (k − 1)] = f (n) − f (0)
k=1
x 1 n+1 n
cos − cos(n + )x sin x sin x
= 2 2 = 2 2 .
x x
2 sin sin
2 2

Exercise
Exercise 2.16. Prove that
n
X n(n + 1)(n + 2)(n + 3)
k(k + 1)(k + 2) = .
k=1
4

1
Exercise 2.17. Given a function f (n) such that f (n) − f (n − 1) = for
n
n ≥ 1. Prove that f (n) is no rational function.
1
Exercise 2.18. Given a function f (n) such that f (n) − f (n − 1) = for
n!
n ≥ 1. Prove that f (n) is no algebraic expression over Q.
1
Exercise 2.19. Given a function f (n) such that f (n) − f (n − 1) = for
n2
n ≥ 1. Prove that f (n) is no algebraic expression over Q.
Exercise 2.20. By finding f (n) such that f (n) − f (n − 1) = un , compute the
n
P
following sum uk where
k=1

uk = k(k + 1)(k + 2)(k + 3)(k + 4); uk = cos(kx + a);


1
uk = .
k(k + 1)(k + 2)(k + 3)(k + 4)
Exercise 2.21. Create 20 exercises.

35
2.3.3 Derivative and integral methods
n
P
These methods base on the truth following: Suppose S(x) = ui (x), then
i=1

n
X Z b n Z
X b

S (x) = u′i (x) and Sn (x)dx = ui (x)dx.
i=1 a i=1 a

Example 2.14. Compute the sums


n n  
X
i−1
X n i−1
S(x) = ix ; U (x) = i x .
i=1 i=1
i

Solution. We have
n n
!′ ′
xn+1 − 1 nxn+1 − (n + 1)xn + 1
X X 
i−1 i
S(x) = ix = x = = ;
i=1 i=1
x−1 (x − 1)2

n   n  
!′
X n i−1 X n i
U (x) = i x = x = ((x + 1)n )′ = n(x + 1)n−1 .
i=1
i i=1
i

Example 2.15. Compute the following sums:


n   n  
X 1 n k+1 X 1 n k+2
S(x) = x ; U (x) = x .
k=0
k+1 k k=0
(k + 1)(k + 2) k

R 1
Solution. Since xk dx = xk+1 + C, we have
k+1
n   n  Z
X 1 n k+1 X n
S(x) = x = xk dx
k=0
k+1 k k=0
k
n  
(x + 1)n+1
Z X Z
n k
= x dx = (x + 1)n dx = + C.
k=0
k n + 1
1 1
We have 0 = S(0) = +C ⇒C =− . So,
n+1 n+1
(x + 1)n+1 1 (x + 1)n+1 − 1
S(x) = − = .
n+1 n+1 n+1

36
Continue, we have
n   Z
X 1 n k+2
U (x) = x = S(x)dx
k=0
(k + 1)(k + 2) k

(x + 1)n+1 − 1 (x + 1)n+2 − x
Z
= dx = + D.
n+1 (n + 1)(n + 2)
1 1
We have 0 = U (0) = +D ⇒D =− . So,
(n + 1)(n + 2) (n + 1)(n + 2)

(x + 1)n+2 − x − 1
U (x) = .
(n + 1)(n + 2)

Exercise
Exercise 2.22. Compute the following sums:
n   n
X n k X
S(x) = k 3 ; U (x) = k 2 cos kx;
k=1
k k=1

n
(−1)k
 
X n k+2
V (x) = x .
k=0
(k + 1)(k + 2)(k + 3) k

Exercise 2.23. Create or collect 30 exercises

2.3.4 Using complex numbers


We can write a complex number z = a + bi, a, b ∈ R by two way:
n
[2]   [ n−1
2
]  
X n n−2k 2k X n
z n = (a + bi) = (−1)k a b +i (−1) k
an−2k−1 b2k+1
k=0
2k k=0
2k + 1

where [x] is the greatest integer less than or equal to x.

z = r(cos φ + i sin φ) where r = |z|, φ = argz


then
z n = rn (cos nφ + i sin nφ).

37
So, we obtained
n
[2]  
n
X
k n n−2k 2k
r cos nφ = (−1) a b ,
k=0
2k

[ n−1
2
]  
n
X
k n
r sin nφ = (−1) an−2k−1 b2k+1 .
k=0
2k + 1

Example 2.16. Prove that


n n−1
[2]   [ 2 ]  
X
k n n nπ X
k n n nπ
(−1) = 2 cos
2 and (−1) = 2 2 sin .
k=0
2k 4 k=0
2k + 1 4

√ π π
Solution. Chose z = 1 + i = 2(cos + i sin ), we have
4 4
n n−1
[2]   [ 2 ]  
n n
X
k n X
k n
z = (1 + i) = (−1) +i (−1)
k=0
2k k=0
2k + 1

√ n π π n nπ n nπ
zn = 2 (cos + i sin )n = 2 2 cos + i2 2 sin .
4 4 4 4
Thus, we obtain
n n−1
[2]   [ 2 ]  
X
k n n nπ X
k n n nπ
(−1) = 2 2 cos and (−1) = 2 2 sin .
k=0
2k 4 k=0
2k + 1 4

Example 2.17. Prove that


n
[2]  
X
k n
cos nφ = (−1) cosn−2k φ sin2k φ
k=0
2k

[ n−1
2
]  
X
k n
sin nφ = (−1) cosn−2k−1 φ sin2k+1 φ.
k=0
2k + 1

Solution. Chose z = cosφ + i sin φ, it is easy to prove.

38
Exercise
Exercise 2.24. Prove that
 
2m 1−2m 1 2m m−1
P 2m

(i) cos φ = 2 m
+ k
cos 2(m − k)φ .
2 k=0

(−1)m
 m−1

2m
2m
(−1)k 2m
m 1−2m
 P 
(ii) sin φ = (−1) 2 m
+ k
cos 2(m − k)φ .
2 k=0

 m

−2m 2m+1
2m+1
P 
(iii) cos φ=2 k
cos(2m + 1 − 2k)φ .
k=0

 m

(iv) sin2m+1 φ = (−1)m 21−2m (−1)k 2m+1
P 
k
sin(2m + 1 − 2k)φ .
k=0

Exercise 2.25. Compute the following sums


n
X n
X n
X
3
cos(a + kb); sin kx; ak sin3 kx.
k=0 k=0 k=0

Exercise 2.26. Compute the following products


n n
Y kπ Y kπ
cos ; sin .
k=1
2n + 1 k=1 2n + 1

2.4 Using generating function method


In mathematics, a generating function is a formal power series in one inde-
terminate, whose coefficients encode information about a sequence of numbers
an that is indexed by the natural numbers. Generating functions were first in-
troduced 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 sev-
eral natural numbers.There are various types of generating functions, including
ordinary generating functions, exponential generating functions, Lambert se-
ries, Bell series, and Dirichlet series. In this section, we only present ordinary
generating functions and exponential generating functions

39
2.4.1 Ordinary generating functions
The ordinary generating function of a number sequence (un )n≥0 is

X
2
F (t) = u0 + u1 t + u2 t + · · · = ui ti
i=0

2.4.2 Exponential generating functions


The exponential generating function of a number sequence (un )n≥0 is

u1 u2 X un
E(t) = u0 + t + t2 + · · · = ti
1! 2! i=0
n!

Generating functions are used to

ˆ Find a closed formula for a sequence given in a recurrence relation. For


example consider Fibonacci numbers.

ˆ Find recurrence relations for sequences—the form of a generating func-


tion may suggest a recurrence formula.

ˆ Find relationships between sequences—if the generating functions of two


sequences have a similar form, then the sequences themselves may be
related.

ˆ Explore the asymptotic behaviour of sequences.

ˆ Prove identities involving sequences.

ˆ Solve enumeration problems in combinatorics and encoding their solu-


tions. Rook polynomials are an example of an application in combina-
torics.

ˆ Evaluate infinite sums.


n
n 2n
 
)2 =
P
Example 2.18. Prove that ( k n
.
k=0

Solution. The generating function of a finite sequence uk = nk , 0 ≤ k ≤ n is




     
n n n n
F (t) = + t + ··· + t = (1 + t)n .
0 1 n

40
We have       2
2 n n n n
F (t) = + t + ··· + t
0 1 n
and      
2 2n 2n 2n 2n 2n
F (t) = (1 + t) = + t + ··· + t .
0 1 2n
By comparing the term of degree n of F (t)2 , we obtaine
  X n   
2n n n
= .
n k=0
k n−k
n
n n n 2n
  P 2 
Notice that k
= n−k
, 0 ≤ k ≤ n, therefore ( k
) = n
.
k=0

Example 2.19. Find the sequence (Fn )n≥0 of Fibonacci numbers is defined
by the recurrence relation

F0 = 1, F1 = 1, Fn+2 = Fn+1 + Fn for all n ≥ 0.

Proof . The generating function of a infinite sequence (Fn )n≥0 is



X
2
F (t) = F0 + F1 t + F2 t + · · · = Fn tn .
n=0

We have ∞
X
2 3
tF (t) = F0 t + F1 t + F2 t + · · · = Fn tn+1
n=0

X
2 2 3 4
t F (t) = F0 t + F1 t + F2 t + · · · = Fn tn+2 ,
n=0

then, we deduce

tF (t) + t2 F (t) = F (t) − 1 ⇒ F (t)(1 − t − t2 ) = 1


1
⇒ F (t) = .
1 − t − t2
√ √
2 1− 5 1+ 5
Write (1 − t − t ) = (1 − at)(1 − bt) with a = and b = , we
2 2
obtain ∞
an+1 − bn+1 n
  X
1 a b
F (t) = − = t .
a − b 1 − at 1 − bt n=0
a−b

41
By comparing the coefficients of series F (t), we have
√ √ !n+1 √ !n+1
 
5 1+ 5 1− 5
Fn = − .
5 2 2

Example 2.20. Find a general formular of sequence (an )n≥0 defining by the
recurrence relation

a0 = −1, a1 = 3, an+2 = 5an+1 − 6an for all n ≥ 0.

Solution. The generating function of a infinite sequence (an )n≥0 is



X
2
F (t) = a0 + a1 t + a2 t + · · · = an tn .
n=0

From hypothesis, we have



X
F (t) = a0 + a1 t + (5an−1 − 6an−2 )tn
n=2
∞ ∞
X X .
= a0 + a1 t + 5t an tn − 6t2 an tn
n=1 n=0
2
= −1 + 3t + 5t(F (t) + 1) − 6t F (t)

Hence (1 − 5t + 6t2 )F (t) = −1 + 8t, implies



−1 + 8t 5 6 X
F (t) = = − = (5.3n − 6.2n )tn .
1 − 5t + 6t2 1 − 3t 1 − 2t n=0

So, Fn = 5.3n − 6.2n , n ≥ 0.


Example 2.21. Find a general formular of sequence (an )n≥0 defining by the
recurrence relation

a0 = 1, an a0 + an−1 a1 + · · · + an a0 = 1 for all n ≥ 1.

Solution. The generating function of a infinite sequence (an )n≥0 is



X
2
F (t) = a0 + a1 t + a2 t + · · · = an tn .
n=0

42
We have
F (t)2 = a20 + (a1 a0 + a0 a1 )t + · · · + (an a0 + · · · + a0 an )tn + · · ·
= 1 + t + t2 + · · · + tn + · · · +
1
= .
1−t
Thus,
1 1.3 t2 1.3....(2n − 1) tn
F (t) = (1 − t)−1/2 = 1 + t + 2 + ··· + + ··· .
2 2 2! 2n n!
1.3....(2n − 1)
So, an = .
2n n!
Example 2.22. Compute the number of nonnegative integer solutions of the
following equation
x1 + x2 + · · · + xd = n.
Solution. Denote Un as the number of nonnegative integer solutions of the
equation x1 + x2 + · · · + xd = n. Then, the generating function of the sequence
(Un )n≥0 issues
∞ ∞  
X
n 2 d 1 X k+d−1 k
F (t) = Un t = (1 + t + t + · · · ) = = t .
n=0
(1 − t)d n=0
k

Hence, Un = n+d−1

n
.

Exercise
Exercise 2.27. Find a general formular of sequence (an )n≥0 defining by the
recurrence relation
a0 = 1, an+1 = 2an + 1 for all n ≥ 1.
Exercise 2.28. Find a general formular of sequence (an )n≥0 defining by the
recurrence relation
a0 = 1, a1 = 0 and an+2 = 2an+1 − an + 3.2n for all n ≥ 0.
Exercise 2.29. Let (an )n≥0 and (bn )n≥0 be sequences satisfied a0 = 1, b0 = 2
and (
an+1 = 3an + 2bn
for all n ≥ 0.
bn+1 = 3bn + 2an
Find a general formular of sequence (an )n≥0 and (bn )n≥0 .

43
Exercise 2.30. Compute the number of nonnegative integer solutions of the
following inequality
x1 + x2 + · · · + x5 ≤ 29
with x1 ≥ 1.

Exercise 2.31. Compute the number of nonnegative integer solutions of the


following equation
x1 + x2 + x3 = 23
with 4 ≤ x1 ≤ 7, 5 ≤ x2 ≤ 8, 6 ≤ x3 ≤ 9.

Exercise 2.32. Prove that


n     
X 2n + 1 m + k 2m + 1
=
k=0
2k 2n 2n

u

with convention v
= 0 if u < v.

Exercise 2.33. Prove that


n     
X m n m+n
=
k=0
r−k k r

u

where m, n, r are positive integers such that r ≤ m + n and convention v
=0
if u < v.

Exercise 2.34. Let n be a nonnegative integer. Compute the number of poly-


nomial P (x) of coefficients in {0, 1, 2, 3} such that P (2) = n.

Exercise 2.35. Create or collect 10 exercises.

44
Chương 3

Some focuses of analysis in high school


programs

3.1 The problems of a tangent


Differentiation is a method to compute the rate at which a dependent
output y changes with respect to the change in the independent input x. This
rate of change is called the derivative of y with respect to x. In more precise
language, the dependence of y upon x means that y is a function of x. This
functional relationship is often denoted y = f (x), where f denotes the function.
If x and y are real numbers, and if the graph of y is plotted against x, the
derivative measures the slope of this graph at each point. The slope of the
tangent line at point (x0 , f (x0 )) is equal to f ′ (x0 ).

3.1.1 Problem 1
For given a graph (C) : y = f (x), then the equation of the tangent line at
point M (x0 , f (x0 )) is

y = f ′ (x0 )(x − x0 ) + f (x0 ).

Corollary 14. Given a Cartesian coordinate xOy and a curve (C) : y = f (x),
then Ox is the tangent line of the curve C at (x0 , f (x0 )) if and only if f (x0 ) =
f ′ (x0 ) = 0.

Therefore, we obtain the following result: Ox is the tangent


( line of the curve
f (x) = 0,
(C) if and only if the equation system following has roots
f ′ (x) = 0.

Definition 6. Let P (x) be a polynomial in one variable and coefficients in R.


An element x0 is called a root of multiplicity k of P (x) if there is a polynomial
Q(x) such that Q(x0 ) ̸= 0 and P (x) = (x − x0 )k Q(x). If k = 1, then x0 is
called a simple root. If k = 2, then x0 is called a double root.

45
Multiplicity can be thought of as "How many times does the solution appear
in the original equation?".
It is easy to see that the polynomial P (x) has a root x0 of multiplicity k if
and only if P (x0 ) = P ′ (x0 ) = · · · = P (k−1) (x0 ) = 0 and P (k) (x0 ) ̸= 0.

Proposition 15. Let P (x) be a polynomial of positive degree, then x-axis is


the tangent line of the graph y = P (x) at the point x = x0 if and only if x0 is
a root of multiplicity k, k ≥ 2 of P (x).

Corollary 16. Let P (x) be a polynomial of positive degree, then x-axis is the
tangent line of the graph y = P (x) if and only if P (x) has a root of multiplicity
k, k ≥ 2.

3.1.2 Problem 2
For given a curve (C) : y = f (x) and M (a, b). Fine all tangent lines of (C)
passing the point M.

To solve this problem, we can do the following step:

Step 1. Solve the equation f ′ (t)(a − t) + f (t) = b, we obtain the roots


t1 , ..., ts .
Step 2. Equation of tangent line are y = f ′ (ti )(a − ti ) + f (ti ), i = 1, 2, ..., s.

x2 − x + 1
Example 3.1. Find all tangent lines of the curve (C) : y = passing
x−1
the point O(0, 0).

x2 − 2x
Solution. We have y ′ (x) = . So, if (d) is a tangent line of (C) at
(x − 1)2
x-axis t and to pass the point O(0, 0), then t is a root of the equation

t2 − 2t t2 − t + 1
(0 − t) + = 0.
(t − 1)2 t−1

1
Solve this equation, we obtain t = . Therefore, the tangent line is y = −3x.
2

46
3.2 The applications of the mean value theo-
rem
Theorem 17. (Lagrange’s Theorem) If a function f (x) is continuous on
the closed interval [a; b], where a < b, and differentiable on the open interval
(a; b), then there exists a point c ∈ (a; b) such that

f (b) − f (a)
f ′ (c) = .
b−a
Lagrange’s theorem is said to be the mean value theorem. The mean value
theorem is a generalization of Rolle’s theorem

Lemma 18. (Rolle’s Theorem) If a function f (x) is continuous on the


closed interval [a; b], where a < b, differentiable on the open interval (a; b) and
f (a) = f (b), then there exists a point c ∈ (a, b) such that f ′ (c) = 0 i.e the
equation f ′ (x) = 0 has a root in (a; b).

Lemma 19. If a function f (x) is continuous on the closed interval [a; b](a < b),
1 Rb
then there exists a point c ∈ (a, b) such that f (c) = f (x)dx.
b−a a

3.2.1 The mean value theorem and inequalities


Example 3.2. Show that ex > 1 + x for x > 0.

Solution. Let f (x) = ex − x − 1, then f ′ (x) = ex − 1 > 0 for x > 0. Thus f (x)
is increasing on interval (0, +∞). So we can conclude that f (x) > f (0) = 0 for
x > 0. i.e ex > 1 + x for x > 0.
π
Example 3.3. Show that for 0 < α < , then we have the following inequal-
2
ities

(i) sin α < α < tan α.


tan α + sin α
(ii) > α.
2
Solution. (i) Consider f (x) = sin x on [0; α]. One has f ′ (x) = cos x < 1 for
x ∈ (0; α). By Lagrange Theorem, there exists a c ∈ (0; α) such that

f (α) − f (0)
= f ′ (c) = cos c < 1,
α−0

47
sin α
this implies < 1. Thus sin α < α.
α
Consider g(x) = tan x on [0; α]. One has g ′ (x) = 1 + tan2 x > 1 for x ∈ (0; α).
By Lagrange Theorem, there exists a b ∈ (0; α) such that
g(α) − g(0)
= g ′ (b) = 1 + tan2 b > 1,
α−0
tan α
this implies > 1. Thus tan α > α.
α
sin x + tan x 1 1
(ii) Consider h(x) = on [0; α]. One has h′ (x) = (cos x+ 2 ) ≥
2 2 cos x
1
√ > 1 for x ∈ (0; α). By Lagrange Theorem, there exists a d ∈ (0; α)
cos x
such that
h(α) − h(0)
= h′ (d) > 1,
α−0
sin α + tan α
this implies > 1.
2
Example 3.4. Show that ea (b − a) < eb − ea < eb (b − a) for a < b.
Solution. Let f (x) = ex , then f ′ (x) = ex . By Lagrange Theorem, there exists
f (b) − f (a) eb − ea
a c ∈ (a; b) such that ec = f ′ (c) = = . Since a < c < b, we
b−a b−a
eb − ea
have ea < ec < eb . Hence, ea < < eb .
b−a

Exercise
π
Exercise 3.1. Prove that for 0 < α < , then we have the following inequal-
2
ities
tan α + 2 sin α
> α.
3
n xk
Exercise 3.2. Prove that ex >
P
for x > 0.
k=0 k!

Exercise 3.3. Prove that


(i) | arcsin a − arcsin b| ≥ |a − b|.
(ii) | arctan a − arctan b| ≤ |a − b|.
b−a b b−a
Exercise 3.4. Prove that < ln < for 0 < a < b.
b a a
Exercise 3.5. Create or collect 20 exercises.

48
3.2.2 The mean value theorem and roots of equation
Example 3.5. Prove that for a < b < c, the following equation has always
two different roots

(x − a)(x − b) + (x − a)(x − c) + (x − b)(x − c) = 0.

Solution. Let f (x) = (x − a)(x − b)(x − c), one has

f ′ (x) = (x − a)(x − b) + (x − a)(x − c) + (x − b)(x − c).

Since f (a) = f (b) = 0, by Rolle Theorem, there exists a c1 ∈ (a; b) : f ′ (c1 ) = 0.


Since f (b) = f (c) = 0, by Rolle Theorem, there exists a c2 ∈ (b; c) : f ′ (c2 ) = 0.
The intervals (a; b) and (b; c) are separate, thus the equation

(x − a)(x − b) + (x − a)(x − c) + (x − b)(x − c) = 0

has two different roots c1 and c2 .

Example 3.6. Let a, b, c ∈ R, then the following equation

a cos 3x + b cos 2x + c cos x + sin x = 0

has always roots.

Solution. Consider a function


1 1
f (x) = a sin 3x + b sin 2x + c sin x − cos x.
3 2
One has

f ′ (x) = a cos 3x + b cos 2x + c cos x + sin x and f (0) = −1 = f (2π).

By Rolle Theorem, there exists a c ∈ (0; 2π) such that f ′ (c) = 0. Thus,
a cos 3x + b cos 2x + c cos x + sin x = 0 has always on (0; 2π).

a b c d
Example 3.7. Let + + + + e = 0. Prove that the following equation
5 4 3 2
has roots
a cos8 x + b cos6 x + c cos4 x + d cos2 x + e = 0.

49
Solution. Let t = cos2 , then t ∈ [0; 1]. The original equation is equivalent to
the equation
at4 + bt3 + ct2 + dt + e = 0.
a b c d
Consider a function f (t) = t5 + t4 + t3 + t2 + et, we have
5 4 3 2
a b c d
f ′ (t) = at4 + bt3 + ct2 + dt + e andf (0) = 0; f (1) = + + + + e = 0.
5 4 3 2
By Rolle Theorem, the equation f ′ (t) = 0 has a root on (0; 1). Thus, the
equation a cos8 x + b cos6 x + c cos4 x + d cos2 x + e = 0 has roots.

Exercise
Exercise 3.6. Given a < b < c < d, prove that the equation

(x−a)(x−b)(x−c)+(x−a)(x−b)(x−d)+(x−a)(x−c)(x−d)+(x−b)(x−c)(x−d) = 0

has only three different roots.

Exercise 3.7. Given n different reals a1 , a2 , ..., an (n > 1), prove that the
equation
n
X 1
=0
i=1
x − ai
has only n − 1 different roots.
a b c
Exercise 3.8. Given a, b, c, m(m > 0) such that + + = 0.
m+2 m+1 m
Prove that the equation ax2 + bx + c = 0 has a root on (0; 1).
n
P
Exercise 3.9. Prove that the equation ai (bi cos x + ci sin x) = 0 has roots.
i=1

Exercise 3.10. Create or collect 20 exercises.

3.2.3 The mean value theorem and integral limit


1
Rn ex + sin x
Example 3.8. Given a sequence (un )n≥1 defined by un = n dx.
−1
x+1
n
Compute lim un .
n→∞

50
ex + sin x
Solution. Consider a function f (x) = on [− n1 ; n1 ], then f (x) is
x+1
continuous on [− n1 ; n1 ]. Thus, there exists a yn ∈ (− n1 ; n1 ) such that
1 1
Zn x Zn
1 e + sin x n ex + sin x un
f (y) = dx = dx = .
1
n
− −1
n
x+1 2 x+1 2
1 1
−n −n

So, lim un = 2 lim f (yn ) = 2 lim f (y) = 2f (0) = 2.


n→∞ n→∞ y→0

R 2x + tan x
1 1+x
Example 3.9. Compute lim dx.
x→0 x 1−x sin x + 1

2x + tan x
Solution. Consider a function f (x) = on [1−x; 1+x], then f (x) is
sin x + 1
continuous on [1 − x; 1 + x] with |x| < 1 . Thus, there exists a y ∈ (1 − x; 1 + x)
such that
Z1+x Z1+x
1 1
f (y) = f (x)dx = f (x)dx.
(1 + x) − (1 − x) 2x
1−x 1−x

So,

Z1+x Z1+x
1 2x + tan x 1
lim dx = 2 lim f (x)dx
x→0 x sin x + 1 x→0 2x
1−x 1−x

2(2 + tan 1)
= 2 lim f (y) = 2 lim f (y) = .
x→0 y→1 1 + sin 1

Exercise
x πx n+1
R 3 + tan 4 n
Exercise 3.11. Compute lim n dx.
n→∞ n−1 ln x + x
n

πx
R 2
1+x+x x7 + cot
Exercise 3.12. Compute lim (1 + cot2 x) 3 dx.
x→0
1+x−x2 |x| + 1
Exercise 3.13. Create 5 exercises.

51
3.3 Methods for finding the maximal and min-
imal values
3.3.1 Derivative method
Theorem 20. If f (x) is a continuous function on [a; b] and have derivative on
(a; b) then there exists the maximal and minimal values of f (x) on [a, b].

Maximal and minimal values on the interval

Step 1. Solve the equation f ′ (x) = 0 and choose the roots in (a; b). Denote
these roots as x1 , x2 , ..., xn . (Notice that it may has no root in (a; b)).

Step 2. Compute f (a), f (b) and f (x1 ), f (x2 ), ..., f (xn ).

Step 3. max f (x) = max{f (a), f (b), f (x1 ), f (x2 ), ..., f (xn )},
x∈[a;b]

min f (x) = min{f (a), f (b), f (x1 ), f (x2 ), ..., f (xn )}.
x∈[a;b]

Maximal and minimal values on the infinite domain


It is possible that there is no global maximum or minimum. It is difficult,
and not particularly useful, to express a complete procedure for determining
whether this is the case. Generally, the best approach is to gain enough un-
derstanding of the shape of the graph to decide i.e investigate the slope table
of function. Fortunately, only a rough idea of the shape is usually needed.

Example 3.10. Given a positive integer n > 1, find the maximal and minimal
values of the following function

y = cos2n x + sin2n x.

Solution. Let t = cos2 x, then t ∈ [0; 1] and

cos2n x + sin2n x = tn + (1 − t)n .

It is easy to see that the maximal and minimal values of y is also the maximal
and minimal values of f (t) = tn + (1 − t)n on [0; 1]. We have

1
f ′ (t) = ntn−1 − n(1 − t)n−1 , f ′ (t) = 0 ⇔ t = ∈ (0; 1).
2
52
1 1
Since f (0) = f (1) = 1; f ( ) = n−1 < 1, we obtained
2 2
1
max y = 1; min y = .
2

Example 3.11. Let a, b be nonnegative reals, prove that


n
an + b n

a+b
≥ for n ≥ 2.
2 2
Solution. The original inequality is equivalent to the following inequality
 n  n  n
a b 1
+ ≥2 .
a+b a+b 2
a
Put x = , then x ∈ [0; 1]) and f (x) = xn + (1 − x)n . We have
a+b
1
f ′ (x) = nxn−1 − n(1 − x)n−1 , f ′ (x) = 0 ⇔ x = ∈ [0; 1].
2
1 1 1
Since f (0) = f (1) = 1; f ( ) = 2 n < 1, deduce min f (x) = 2 n i.e f (x) ≥
2 2 x∈[0;1] 2
1
2 n for all x ∈ [0; 1]. So
2
n
an + b n

a+b
≥ .
2 2

Example 3.12. Let c > 0 and n > 1 be positive integer, find the minimal
value of f (x) = xn + (c − x)n . Since that, prove that if a + b ≥ 0, then
n
an + b n

a+b
≥ for n ≥ 2
2 2
c
Solution. We have f ′ (x) = nxn−1 − n(c − x)n−1 , and f ′ (x) = 0 ⇔ x = ;
2
c c
f ′ (x) > 0 ⇔ x > ; f ′ (x) < 0 ⇔ x < . Thus the minimal value of f (x) is
2 2
c c
f ( ) = 2( )n . Thus
2 2
c
xn + (c − x)n ≥ 2( )n . (∗ )
2
53
Now, we show that
n
an + b n

a+b
≥ for n ≥ 2.
2 2
n
an + b n an + (−a)n

a+b
If a + b = 0, then a = −b, thus = ≥0= .
2 2 2
If a + b > 0, put c = a + b. By (∗ ), we have
n
an + b n an + (c − a)n

c n a+b
= ≥( ) = .
2 2 2 2

Example 3.13. Given a > 0; b > 0 and x > y > 0. Prove that

(ax + bx )y < (ay + by )x .

Solution. The original inequality is equivalent to


 y  x
xy b x xy b y
a 1+( ) <a 1+( )
a a
 y  x
b x b y
⇔ 1+( ) < 1+( )
a a
 1  1
b x x b y y
⇔ 1+( ) < 1+( ) .
a a
b
Put x = and takes ln two sides, we obtain the equivalent inequality
a
ln(1 + ex ) ln(1 + ey )
< .
x y

ln(1 + et )
Consider the functions f (t) = on (0; +∞), we have f ′ (t) < 0 for
t
t ∈ (0; +∞). Hence, f (t) is decreasing on (0; +∞). Since x > y > 0, we can
conclude that f (x) < f (y), so

(ax + bx )y < (ay + by )x .

54
Exercise
Exercise 3.14. Find the maximal and minimal values of the following func-
tions

1−x2
y = ax + a
on [0; 1] where a is positive constant, a ̸= 1.

Exercise 3.15. Find the maximal and minimal values of the following func-
tions

y = 2| sin x| + 2| cos x|

Exercise 3.16. Find the maximal and minimal values of the following func-
tions

(i) y = | cos x|α + | sin x|α with a constant α > 0 ;

(ii) y = cos 3x − 5 cos 2x + cos x + 1;


√ √
(iii) 2n x + 2n a − x with a > 0 and n is a possitive integer.
√ √
(iv) 2n x − 2n a − x with a > 0 and n is a possitive integer.

Exercise 3.17. Create or collect 20 exercises.

3.3.2 Using inequality


Example 3.14. Given n be a possitive integer, find the maximal and minimal
values of the following function
√ √
(i) y = 2n cos x − 2n sin x.

(ii) y = cosn x − sinn x with n ≥ 2.

Solution. (i) We have



2n √ √
2n √
−1 ≤ − sin x ≤ 2n cos x − sin x ≤ 2n cos x ≤ 1,

hence
−1 ≤ y ≤ 1.

55
π
Moreover, y(0) = 1, y( ) = −1, this implies min y = −1 and max y = 1.
2
(ii) We have (
− cos2 x ≤ cosn x ≤ cos2 x
.
− sin2 x ≤ − sinn x ≤ sin2 x
pi
Plus side to side, we obtain −1 ≤ y ≤ 1. Moreover, y(0) = 1, y( ) = −1, this
2
implies min y = −1 and max y = 1.
Example 3.15. Find the minimal value of the following functions
1 1
(i) y = +
| cos x| | sin x|
1 n 2 1 n
(ii) y = (cos2 x + 2 ) + (sin x + ) with n ∈ N∗ .
sin x cos2 x
Solution. (i) By AM-GM inequality,

1 1 2 2 2 √
y= + ≥p =p ≥ 2 2.
| cos x| | sin x| | cos x|| sin x| | sin 2x|
π √ √
Moreover, y( ) = 2 2, we obtain min y = 2 2.
4
(ii) Since the inequality in Example 3.11, we have
n
an + b n

a+b
≥ ,
2 2
thus
1 n 1 n
y = (cos2 x + 2
2 ) + (sin x + )
sin x cos2 x
1 1 1 n
≥ n−1 (cos2 x + sin2 x + 2 + )
2 sin x cos2 x
1 1 1 n 1 4
= n−1 (1 + 2 + ) = n−1 (1 + )n
2 sin x cos x 2 2 sin2 2x
5
≥ 2( )n .
2
π 5 n 5
Since y( ) = 2( ) , we obtain min y = 2( )n .
4 2 2

Example 3.16. Find the minimal value of a function


√ √
y = x2 − x + 1 + x2 + x + 1.

56
√ √
r r
1 2 1
Solution. We have y = (x − ) + ( 32) + (x + )2 + ( 32)2 . Takes
2
2 2
two vector
1 √ 1 √
⃗u = ( − x, 32) and ⃗v = ( + x, 32).
2 2
Then √
⃗u + ⃗v = (1, 3) and y = |⃗u| + |⃗v | ≥ |⃗u + ⃗v | = 2.
Moreover y(0) = 2, hence min y = 2.
Example 3.17. Given a, b with a > b, find the minimal and maximal values
of a function √ √
y = a − x + x − b.
√ √ √ √
Solution. We have y = a − x + x − b ≥ a − x + x − b = a − b.
By Cauchy - Schwarz inequality, we also have
√ √
y 2 = ( a − x + x − b)2 ≤ 2(a − x + x − b) = 2(a − b).
√ p √ a+b
Thus, a − b ≤ y ≤ 2(a − b). Moreover y(a) = a − b and y( ) =
p √ p 2
2(a − b), so min y = a − b and max y = 2(a − b).
Example 3.18. Given a, b, c be positive constants and x, y, z be positive vari-
a b c
ables such that + + = 1. Find the minimal value of S = x + y + z.
x y z
Solution. Apply Cauchy - Schwarz inequality, we have
a b c
S = x + y + z = (x + y + z)( + + )
x y z
r 2 s !2 r 
2
√ 2 √ 2 √ 2 a b c
= [( x) + ( y) + ( z) ][ + + ]
x y z
s r !2
√ b √ √ √ √
r
a √ c
≥ x + y + z = ( a + b + c)2 .
x y z

Equality occurs if and only if


s
√ b √
r r
a √ c
x: = y: = z: .
x y z
√ √ √ √ √ √ √ √ √ √ √ √
Thus, x = a( a + b + c); y = √b( a√ + b+√ c); z = c( a + b + c).
So, we can conclude that min S = ( a + b + c)2 .

57
Exercise
Exercise 3.18. Find the minimal and maximal values of the following func-
tions

(i) y = | cos x|α − | sin x|α with α > 0.


√ √
(ii) y = 2n a − x ± 2n x − b with a > b and n ∈ N∗ .

(iii) y = 3| cos x|α − 4| sin x|α with α > 0.

Exercise 3.19. Find the minimal values of the following functions


1 1
(i) y = α
+ with α > 0.
| cos x| | sin x|α
1 2n 1
(ii) y = (cos2n x + m
2n ) + (sin x + )m with m, n ∈ N∗ .
sin x cos2n x
Exercise 3.20. Given ai , bi , i = 1, 2, ..., n be positive constants and xi , i =
n n b
P P i
1, 2, ..., n be positive variables. Put A = ai xi and B = .
i=1 i=1 xi

(i) Find the minimal value of A if B = 1.

(ii) Find the minimal value of B if A = 1.

Exercise 3.21. Create or collect 20 exercises.

3.3.3 Using the root conditions of the equation


Let f (x) be a function determining on the domain M. Denote as G := f (M )
then y ∈ G if and only if there exists x ∈ M such that f (x) = y. That means
the equation y = f (x) has a root in M. Therefrom, we can find G.

Example 3.19. Find the set of values the function f (x) = a cos x + b sin x + c.

Solution. Denote D as the set of values the function f (x), then y ∈ D if and
only if the equation a cos x + b sin x + c − y = 0 has a root. This is equivalent
to a2 + b2 ≥ (c − y)2 . Thus,
√ √
c − a2 + b 2 ≤ y ≤ c + a2 + b 2 .
√ √
So, D = [c − a2 + b2 ; c + a2 + b2 ].

58
Example 3.20. Find the minimal and maximal values of the functions
2 cos x − sin x + 1
f (x) = .
cos x + sin x + 2
Solution. Since cos x + sin x + 2 > 0 for all x, the function f (x) is defined
over R. Let G be a set of values f (x), then y ∈ G if and only if the equation
2 cos x − sin x + 1
y= has a root.This is equivalent to the following equation
cos x + sin x + 2
has a root:
(2 − y) cos x − (1 + y) sin x + (1 − 2y) = 0.
Implies (2 − y)2 + (1 + y)2 ≥ (1 − 2y)2 . Solve this inequality, we obtain −1 ≤
y ≤ 2 i.e G = [−1; 2]. So, min f (x) = −1 and max f (x) = 2.

Example 3.21. Find the minimal and maximal values of the functions
x+2
y= .
x2 −x+1
1 3
Solution. Since x2 − x + 1 = (x − )2 + > 0 for all x, the function f (x)
2 4
is defined over R. Let G be a set of values f (x), then m ∈ G if and only if
x+2
the equation m = 2 has a root. This is equivalent to the following
x −x+1
equation has a root:

mx2 − (m + 1)x + (m − 2) = 0

. If m = 0 then x = −2 is a root, so 0 ∈ G. If m ̸= 0, then the equation has


a root if and only if ∆ = (m + 1)2 − 4m(m
√ − 2) ≥ 0. Solve
√ this inequalit and
5−2 7 5+2 7
combine with m = 0, we obtain ≤m≤ . This implies that
√ 3√ 3
5−2 7 5+2 7
min y = and max y = .
3 3

Exercise
Exercise 3.22. Find the minimal and maximal values of the following func-
tions
cos2 x − cos x sin x
(i) y = .
cos2 x + 1

59
cos4 x + sin4 x
(ii) y = .
cos6 x + sin6 x + cos2 2x
2x + 1
(iii) y = .
x2 + 2x + 10
sin 2x
(iv) y = .
cos x − sin x − 2
a cos x + b sin x + c
Exercise 3.23. Find the conditions such that y = has
p cos x + q sin x + r
the minimal and maximal values where p2 + q 2 ̸= 0 and the equation a cos x +
b sin x + c = 0 and p cos x + q sin x + r = 0 have not common roots.
Exercise 3.24. Create or Collect 20 exercises.

3.3.4 Using the discriminant ∆


Let f (x) be a polynomial ax2 + bx + c, a ̸= 0 with the discriminant ∆ =
2
b − 4ac. Then
(i) f (x) has a root if and only if ∆ ≥ 0.

(ii) f (x) ≥ 0 for all x if and only if a > 0 and ∆ ≤ 0.

(iii) f (x) ≤ 0 for all x if and only if a < 0 and ∆ ≤ 0.


Example 3.22. Show that for all x, y, z

A = 2x2 + 2y 2 + 3z 2 + 2zy − 2xz + 2x + 2y + 2z + 3 > 0

Solution. Rewrite A = 2x2 + 2(1 − z)x + (2y 2 + 3z 2 + 2zy + 2y + 2z + 3), then


A > 0 for all x, y, z if and only if

∆′1 = (1 − z)2 − 2(2y 2 + 3z 2 + 2zy + 2y + 2z + 3) < 0

for all x, y. Rewrite

∆′1 = −4y 2 − 4(1 + z)y − (5z 2 + 6z + 5)

then
∆′1 < 0 for all y, z ⇔ ∆′2 < 0 for all z
where ∆′2 = 4(z + 1)2 − 4(5z 2 + 6z + 5). This is always true. So, A > 0.

60
Example 3.23. Find the minimal value of the expression
A = x2 + 2y 2 + 3z 2 + 2zy − xz + 2x − 2y
where x, y, z are variables.
Solution. m is a minimal value of A if and only if A ≥ m for all x, y, z and
there exists x0 , y0 , z0 such that A(x0 , y0 , z0 ) = m. Thus, we consider a problem
A − m = x2 + 2y 2 + 3z 2 + 2zy − xz + 2x − 2y − m
= 3z 2 + (2y − x)z + (x2 + 2y 2 + 2x − 2y − m) ≥ 0
for all x, y, z. This implies that
∆1 = (2y − x)2 − 12(x2 + 2y 2 + 2x − 2y − m)
= −20y 2 − 4(x − 6)y − (11x2 + 24x − 12m) ≤ 0
for all x, y. Thus,
∆′2 = −24[9x2 + 22x − (6 + 10m)] ≤ 0
for all x. This occurs only
−35
∆′3 = 121 + 9(6 + 10m) = 0 ⇔ m = .
18
−35
So, min A = .
18

Example 3.24. Find m such that


x2 + y 2 + z 2 + mzy − xz + x − y + 1 ≥ 0
for all x, y, z.
Solution. We have x2 + y 2 + z 2 + mzy − xz + x − y + 1 = x2 + (1 − z)x +
(y 2 + z 2 + mzy − y + 1) ≥ 0 for all x, y, z if and only if
∆1 = (1−z)2 −4(y 2 +z 2 +mzy −y +1) = −4y 2 −4(mz −1)y −(3z 2 +2z +3) ≤ 0
for all y, z. Thus
∆2 = 4(mz − 1)2 − 4(3z 2 + 2z + 3) = 4[(m2 − 3)z 2 − 2(m + 1)z − 2] ≤ 0 for all z.
This is equivalent to
(
m2 − 3 <0
2 2
(m + 1) + 2(m − 3) ≤ 0.
−5
Solve this system of equation, we obtain ≤ m ≤ 1.
3
61
3.3.5 Trigonometric method
Suppose that we find the maximal and minimal values of a function f (x, y)
with x, y satisfying the equation G(x, y) = 0. In elementary mathematics, we
only solve some special cases of this problem such as: If from G(x, y) = 0, we
obtain y = U (x) then the problem will be transformed to find extremal values
of one variable function f (x, U (x)). In other cases,(if G(x, y) = 0 is equivalent
(x − x0 )2 (y − y0 )2 x = x0 + a sin t
to 2
+ 2
= 1, then we pull out Hence,
a b y = y0 + b cos t.
the problem will be transformed to find extremal values of the trigonometric
function f (x0 + a sin t, y0 + b cos t).

Example 3.25. For given x, y ∈ R such that x2 + y 2 = 1. Find maximum and


minimum of the expression following

A = 3x2 − 2y 2 + 4xy + 1.

Solution. From x2 + y 2 = 1, we can write x = cos t, y = sin t. Then,


5 3
A = 3 cos2 t − 2 sin2 t + 4 cos t sin t + 1 = cos 2t + 2 sin 2t + .
2 2
Therefore, we deduce
√ √
3− 41 3 + 41
min A = ; max A = .
2 2

Example 3.26. For given x, y ∈ R such that 4x2 + y 2 − 8x − 2y + 1 = 0. Find


the maximum and minimum of the expression following

A = 3x + 2y + 1.

Solution. We have 4x2 + y 2 − 8x − 2y + 1 = 0 ⇔ (x − 1)2 + ( y−1


2
)2 = 1. Hence,
we can write
x = 1 + cos t; y = 1 + 2 sin t.
Deduce A = cos t + 4 sin t + 6. Therefore, we obtain easily the maximum and
minimum of A
min A = 1; max A = 11.

62
Example 3.27. For given x, y ∈ R such that 4x2 + y 2 − 8x − 2y + 1 = 0. Find
the maximum and minimum of the expression following
x − 2y + 1
B= .
x+y+1

Solution. We have 4x2 + y 2 − 8x − 2y + 1 = 0 ⇔ (x − 1)2 + ( y−1 2


)2 = 1. Hence,
we can write
x = 1 + cos t; y = 1 + 2 sin t.
cos t − 4 sin t
Deduce B = . From cos t + 2 sin t + 3 > 0 for t ∈ R, B is
cos t + 2 sin t + 3
defined on R. Now, denote M as the set of values of B, then m ∈ M if and
only if the equations
cos t − 4 sin t
=m
cos t + 2 sin t + 3
has roots. Equivalently, the equation

(m − 1) cos t + 2(m + 2) sin t + 3m = 0

has roots. Deduce


√ (m − 1)2 + 2 2
√4(m + 2) ≥ (3m) . Solve this inequality, we
7 − 134 7 + 134
obtain ≤m≤ . So,
5 5
√ √
7 − 134 7 + 134
min B = ; max B = .
5 5

63
64
Chương 4

Some applications of polynomials

4.1 Some interpolation polynomials


4.1.1 Lagrange’s interpolation polynomial
A polynomial of n-th degree is uniquely determined, given its values at
n + 1 points. So, suppose that P (x) is an n-th degree polynomial and that
P (xi ) = yi in different points x0 , x1 , ..., xn . There exist unique polynomials
E0 (x), E1 (x), ..., En (x) of n-th degree such that Ei (xi ) = 1 and Ei (xj ) = 0 for
j ̸= i. Then the polynomials

P (x) = y0 E0 (x) + y1 E1 (x) + · · · + yn En (x)

has the desired properties:


n
X
P (xi ) = yj Ej (xi ) = yi Ei (xi ) = yi .
j=0

It remains to find the polynomials E0 (x), E1 (x), Q ..., En (x). A polynomial that
vanishes at the n points xj , j ̸= i, is divisible by j̸=i (x − xj ), from which we
Q (x − xj )
easily obtain Ei (x) = j̸=i .
(xi − xj )
This shows that

Theorem 21. (Lagrange’s interpolating polynomial). For given numbers


y0 , y1 , ..., yn and distinct x0 , x1 , ..., xn , there is an unique polynomial P (x) of
degree n such that P (xi ) = yi for i = 0, 1, ..., n. This polynomial is given by
the formula
X n Y (x − xj )
P (x) = yi .
i=0 j̸=i
(x i − x j )

The polynomial P (x) is called Lagrange’s interpolation polynomial and the


points x0 , x1 , ..., xn are called the interpolation nodes.

65
4.1.2 Taylor’s Polynomial
Given a nice function f : D ⊂ R → R, a point a in the interior of the
domain D, and an interger n, find a polynomial P of degree ≤ n such that

P (a) = f (a), P ′ (a) = f ′ (a), ..., P (n) (a) = f (n) (a).

We simply verify that a polynomial

f ′ (a) f ”(a) f (n) (a)


P (x) = f (a) + (x − a) + (x − a)2 + · · · + (x − a)n
1! 2! n!
does the job. Moreover, it is also easy to see that this polynomial is the only
polynomial of degree ≤ n that does the job.
In particular, f (x) is a polynomial of degree n then for a point a ∈ R, we
have

f ′ (a) f ”(a) f (n) (a)


f (x) = f (a) + (x − a) + (x − a)2 + · · · + (x − a)n .
1! 2! n!
This formula is called the Taylor expansion of the polynomial f (x).

Example 4.1. Find the cubic polynomial P (x) such that P (i) = 2i for i =
0, 1, 2, 3.

Solution. Apply the Lagrange’s interpolation polynomial, we have

(x − 1)(x − 2)(x − 3) 2x(x − 2)(x − 3) 4x(x − 1)(x − 3)


P (x) = + +
−6 2 −2

8x(x − 1)(x − 2) x3 + 5x + 6
+ = .
6 6

Example 4.2. Let x0 , x1 , ..., xn be distinct real numbers and f (x) = (x −


x0 )(x − x1 )...(x − xn ). Then, for a polynomial

p(x) = a0 xn + a1 xn−1 + · · · + an , a0 ̸= 0,
n p(x )
P i
we have a0 = ′
.
i=0 f (xi )

66
Solution. Apply the Lagrange’s interpolation polynomial, we have
n
X p(xi ) f (x)
p(x) = .
i=0
f ′ (xi ) x − xi

Compare the leadding coefficient of two polynomials in left side and right side,
we obtained n
X p(xi )
a0 = .
i=0
f ′ (xi )

Example 4.3. Let x0 < x1 < · · · < xn be integers and p(x) = xn + a1 xn−1 +
n!
· · · + an ∈ R[x] then there exists at least xi such that |p(xi )| ≥ n .
2
Solution. Let f (x) = (x − x0 )(x − x1 )...(x − xn ), since the above example, we
have n
X p(xi )
1= ′ (x )
.
i=0
f i

Put
α = max{|p(x0 )|, |p(x1 )|, ..., |p(xn )|},
then
n n n n
X p(xi ) X p(xi ) X 1 X 1
1=| ′
|≤ | ′ |≤α ′
≤α .
i=0
f (xi ) i=0
f (xi ) i=0
|f (xi )| i=0
i!(n − i)!

So n
X 1 α2n
1≤α = .
i=0
i!(n − i)! n!
n!
Hence α ≥ .
2n

Example 4.4. Given a polynomial p(x) of degree ≤ 2n, n be a positive integer


and |p(i)| ≤ 1 for every integer i ∈ [−n; n]. Show that |p(x)| ≤ 22n for x ∈
[−n; n].
Solution. Apply the Lagrange’s interpolation polynomial, we have
n n
X Y x−k
p(x) = p(i) .
i=−n k̸=i,k=−n
i−k

67
Hence,
n n n n
X Y x − k X Y x − k
|p(x)| ≤ |p(i)| i−k ≤

i − k .

i=−n k̸=i,k=−n i=−n k̸=i,k=−n

We also have
n
Y
|x − k| ≤ (2n)! textf or x ∈ [−n; n].
k̸=i,k=−n

The, we obtained
n n n
X Y 1 X (2n)!
|p(x)| ≤ (2n)! = = 22n .
i=−n k̸=i,k=−n
|i − k| i=−n
(n − i)!(n + i)!

4.1.3 Newton’s interpolation polynomial


It is sometimes useful to find several approximating polynomials P1 (x),
P2 (x), ..., PN (x) and then choose the one that suits our needs. If the Lagrange’s
interpolation polynomials are used, there is no constructive relationship be-
tween Pn−1 (x) and Pn (x). Each polynomial has to be constructed individually,
and the work required to compute the higher-degree polynomials involves many
computations. We take a new approach and construct Newton’s interpolation
polynomials that have the recursive pattern:
P0 (x) = a0 ,
P1 (x) = a0 + a1 (x − x0 ),
P2 (x) = a0 + a1 (x − x0 ) + a2 (x − x0 )(x − x1 ),
P3 (x) = a0 + a1 (x − x0 ) + a2 (x − x0 )(x − x1 ) + a3 (x − x0 )(x − x1 )(x − x2 ),
.................................................
Pn (x) = a0 + a1 (x − x0 ) + a2 (x − x0 )(x − x1 ) + · · · + an (x − x0 )...(x − xn−1 ).
Here the polynomial Pn (x) is obtained from Pn−1 (x) using the recursive rela-
tionship
Pn (x) = Pn−1 (x) + an (x − x0 )(x − x1 )...(x − xn−1 ).
The polynomial Pn (x) is said to be a Newton polynomial with n centers
x0 , x1 , ..., xn . We can find ai much more quickly by solving the equations re-
spectively
P (x0 ) = y0 , P (x1 ) = y1 , ...., P (xn ) = yn .

68
This shows that

Theorem 22. (Newton’s interpolation polynomial). For given a poly-


nomial P (x) ∈ R[x] of degree n > 0 and x0 , x1 , ..., xn ∈ R, there exist
a0 , a1 , ..., an ∈ R such that

P (x) = a0 +a1 (x−x0 )+a2 (x−x0 )(x−x1 )+· · ·+an (x−x0 )(x−x1 )...(x−xn−1 ).

Exercise
Exercise 4.1. Let n be a positive integer. Find the polynomial f (x) ∈ R[x]
1
of the smallest degree such that f (i) = for i = 1, 2, ..., n.
i
Exercise 4.2. Solve the system of following equation by using Lagrange’s
interpolation polynomial
n−1

a0 + a1 x1 + · · · + an−1 x1
 = b1
a + a x + · · · + a xn−1

= b2
0 1 2 n−1 2


 ........................................
a0 + a1 xn + · · · + an−1 xn−1

n = bn

Exercise 4.3. Let f ∈ C n+1 [a, b] and P (x) be Lagrange’s interpolation poly-
nomial for f with nodes x0 , x1 , ..., xn ∈ [a, b]. Then

M
max |P (x) − f (x)| ≤ max |(x − x0 )(x − x1 )...(x − xn )|
a≤x≤b (n + 1)! a≤x≤b

where M = max |f (n+1) (x)|.


a≤x≤b

Exercise 4.4. Assume that the polynomial P (x) of degree n takes the value
1 at points 0, 2, 4, .., 2n. Compute P (−1).

Exercise 4.5. Assume that α1 , ..., αn are n roots of polynomial p(x). Then
n
X αis
= 0 for s = 0, 1, ..., n − 2.
i=1
f ′ (αi )

Exercise 4.6. Let f (x)R[x] and a be a real. Prove that a is the root of mul-
tiplicity n of f (x) if and only if f (m) (a) = 0 for m < n and f (n) (a) ̸= 0.

69
4.2 Polynomials of a particular form
4.2.1 Polynomials with interger coefficients
Consider a polynomial P (x) = an xn + · · · + a1 x + a0 ∈ Z[x]. The difference
P (x) − P (y) can be written in the form

an (xn − y n ) + · · · + a2 (x2 − y 2 ) + a1 (x − y)

in which all terms are multiples of polynomial x − y. This leads to the simple
though important arithmetic property of polynomial.

Theorem 23. If P is a polynomial with interger coefficients then a − b |


P (a) − P (b) for all distinct integers a and b. In particular, all integer roots of
P divide P (0).

Theorem 24. If a rational number p/q (p, q ∈ Z, q ̸= 0) and (p, q) = 1 is a


root of polynomial P (x) then p | a0 and q | an .

Example 4.5. Polynomial P (x) ∈ Z[x] takes values ±1 at three different


integer points. Prove that it has no integer solutions.

Solution. Suppose to the contrary, that a, b, c, d are integers with P (a), P (b), P (c) ∈
{−1, 1} and P (d) = 0. Then the integers a − d, b − d and c − d all divide 1.
This contradicts.

Definition 7. A polynomial P (x) ∈ Z[x] is called irreducible over Z if it


cannot be represented as a product of two polynomials of possitive degree
with integer coefficients. Otherwise, P (x) is called reducible.

Lemma 25. (Gauss’ lemma) A polynomial P (x) ∈ Z[x] is irreducible over


Z if and only it is irreducible over Q, also.

One of the best known irreducible criteria of polynomials is the following


Eisenstein’s Criterion

Theorem 26. (Eisenstein’s Criterion) Let P (x) = an xn + · · · + a1 x + a0


be a polynomial in Z[x]. If there exist a prime number p such that

p | a0 , a1 , ..., an−1 , p ∤ an ; p2 ∤ a0

then P (x) is irreducible over Z.

70
Theorem 27. (Extended Eisenstein’s Criterion) Let P (x) = an xn +· · ·+
a1 x+a0 be a polynomial in Z[x]. If there exist a prime number p and an integer
k ∈ {0, 1, ..., n − 1} such that

p | a0 , a1 , ..., ak , p ∤ ak+1 ; p2 ∤ a0

then P (x)have an irreducible factor of a degree greater than k.


In particular, if p can be taken so that k = n − 1 then P (x) is irreducible
over Z.

Example 4.6. Let p be a prime and let q be not divisible by p. Then xm − pq


is irreducible over Z.

Example 4.7. If p is a prime, then P (x) = xp−1 + xp−2 + · · · + 1 is irreducible


over Z.

Solution. One can apply Eisenstein’s criterion to the polynomial

(x + 1)p − 1
   
p−1 p p−2 p
P (x + 1) = =x + x + ··· + .
(x + 1) − 1 1 p−1

Thus, P (x + 1) is irreducible over Z. Deduce P (x) is irreducible over Z.

Example 4.8. For any positive integer n, the polynomial

x2 xn
P (x) = 1 + x + + ··· +
2! n!
is irreducible.

Solution. We have to prove that the polynomials

n!P (x) = xn + nxn−1 + · · · + n!

is irreducible over Z. To this end, it suffices to find the prime p such that n! is
divisible by p but is not divisible by p2 i.e p ≤ n ≤ 2p.
Let n = 2m or n = 2m + 1. Bertrand’s postulate states that
for m > 1, there exists at least one prime p such that m < p < 2m.
For n = 2m, the inequalities p ≤ n < 2p are obvious. For n = 2m + 1, we
obtain the inequalities p ≤ n − 1 and n − 1 < 2p. But in this case the number
n − 1 is even and hence the inequality n − 1 < 2p implies n < 2p. It is also
clear that p ≤ n − 1 < n.

71
Theorem 28. (Osada’s Criterion) Let P (x) = xn + an−1 x + · · · + a1 x ± p
be a polynomial in Z[x] and p be a prime number. If

p > 1 + |an−1 | + · · · + |a1 |

then P (x) is irreducible.

Example 4.9. If p is a odd prime, then P (x) = xp−1 + xp−2 + · · · + x ± p is


irreducible over Z.

4.2.2 Polynomials with real and complex coefficients


Theorem 29. The polynomial of degree n has exactly n complex roots with
their multiplicities.

Theorem 30. A real polynomial P (x) of degree n has a unique factorization


(up to the order) of the form

P (x) = a(x − α1 )...(x − αk )(x2 − p1 x + q1 )...(x2 − pl x + ql )

where a, αi , pj , qj are real numbers with p2j − 4qj < 0 and k + 2l = n.

Example 4.10. Find possitive integers n such that (x+1)n −xn −1 is divisible
by x2 + x + 1.

Solution. Let f (x) = (x √+ 1)n − xn − 1 and g(x) = x2 + x + 1. See that


−1 ± 3i
g(x) has two roots , thus f (x) is divisible by g(x) if and only if
√ 2
−1 ± 3i
f( ) = 0. Rewrite
2

−1 ± 3i 2π 2π
= cos ± i sin
√2 3 3.
−1 ± 3i π π
+ 1 = cos ± i sin
2 3 3
Hence,

−1 ± 3i π π 2π 2π
f( ) = 0 ⇔ (cos ± i sin )n − (cos ± i sin )n − 1 = 0.
2 3 3 3 3

Deduce cos = 0, equivalent to n = ±1 + 6k, k ∈ N∗ .
3
72
4.2.3 Arithmetic polynomials
A real polynomial P (x) is said to be an arithmetic polynomial if the value of
the polynomial P (x) is integer at all integer points x. Note that a polynomial
that takes integer values at all integer points does necessarily have integer
x(x − 1)
coefficients as seen on the polynomial . The arithmetic polynomial is
2
also called integer valued polynomial.
Theorem 31. If the value of a polynomial P (x) of degree n > 0 is integer at
all integers x, then there exist integers a0 , a1 , ..., an such that
     
x x x
P (x) = an + an−1 + · · · + a0 ,
n n−1 0
where    
x x x(x − 1)...(x − k + 1)
= 1; = , k = 1, 2, ..., n.
0 k k!
The converse is true, also.
Proof. We use induction on n. The case n = 1 is trivial; Now, assume n > 1.
Polynomial Q(x) = P (x+1)−P (x) is of degree n−1 and takes integer values at
all integer points, so by the inductive hypothesis there exists b0 , b1 , ..., bn−1 ∈ Z
such that    
x x
Q(x) = bn−1 + · · · + b0 .
n−1 0
For integer x > 0, we have P (x) = Q(0) + Q(1) + · · · + Q(x − 1). Using the
identity        
x x−1 x−2 0
= + + ··· +
k+1 k k k
for every integer k, we obtain the desired representation of P (x)
     
x x x
bn−1 + bn−2 + · · · + b0 + P (0).
n n−1 1

Corollary 32. If a polynomial P (x) ∈ R[x] is an arithmetic polynomial then


P (x) ∈ Q[x].
Theorem 33. Given a real polynomial Qn (x) of degree n > 0. If the value of
Q(x) is integer for n + 1 consecutive integers x, then Qn (x) is an arithmetic
polynomial.

73
Proof. Assume Qn (x) has integer values at x = m, m + 1, ..., m + n. The poly-
nomials
xn
       
x x x x(x − 1) x
= 1, = x, = , ..., = + ···
0 1 2 2 n n!
form a basis in the space of polynomials of degree not greater than k, and
hence    
x x
Qn (x) = cn + cn−1 + · · · + c0 ,
n n−1
where c0 , c1 , ..., cn are some numbers. It only remains to prove that these num-
bers are integers.
We use induction on n. For n = 0, the polynomial Q0 (x) = c0 , then c0 =
Q0 (m) ∈ Z. Assume the required statement is true for all polynomials of degree
not greater than n. Let the polynomial
   
x x
Qn+1 (x) = cn+1 + cn + · · · + c0 ,
n+1 n
take integer values at x = m, m + 1, .., m + n + 1. Then the polynomial
   
x x
∆Qn+1 (x) = Qn+1 (x + 1) − Qn+1 (x) = cn+1 + cn + · · · + c1
n n−1
takes integer values at x = m, m+1, ...., m+k. Therefore cn+1 , ...,c1 are integers
m
− cn m − c1 m1 .
 
and hence so is c0 = Qn+1 (m) − cn+1 n+1 n
− · · ·
Example 4.11. Let m be a positive integer and P (x) = an xn + · · · + a1 x + a0
be a real polynomial such that P (x) is an integer divisible by m whenever x
is an integer. Prove that n!an is divisible by m.
1
Solution. Apply Theorem 31 on polynomial P (x), we have
m
   
1 x x
P (x) = bn + bn−1 + · · · + b0 , ai ∈ Z, i = 0, 1, .., n.
m n n−1
By comparing the leading coefficient of two polynomials, we obtained
an bn b0
= + ··· + .
m n! 1
n!an
Therefore = bn +(n−1)bn−1 +· · ·+n!b0 ∈ Z. The statement is proved.
m
74
4.2.4 Chebyshev polynomials
The Chebyshev polynomials of the first kind are defined by the recurrence
relation
T0 (x) = 1; T1 (x) = x; Tn+1 (x) = 2xTn (x) − Tn−1 (x).
The conventional generating function for Tn is

X 1 − tx
Tn (x)tn = .
n=0
1 − 2tx + t2

Theorem 34. For each positive integer n, there exist a polynomial Tn of


degree n such that Tn (cos α) = cos(nα).

We can dedude that Tn (x) is Chebyshev polynomial of the first kind. So


that the Chebyshev polynomials of the first kind can be defined as the unique
polynomials satisfying
Tn (cos α) = cos(nα).
The Chebyshev polynomials of the second kind are defined by the recur-
rence relation

U0 (x) = 1; U1 (x) = 2x; Un+1 (x) = 2xUn (x) − Un−1 (x).

The generating function for Un is



X 1
Un (x)tn = .
n=0
1 − 2tx + t2

Theorem 35. For each positive integer n, there exist a polynomial Un of


sin((n + 1)α)
degree n such that Un (cos α) = .
sin α
The fact that Tn (x) = cos nα for x = cos α directly implies that |Tn (x)| ≤ 1
for x ≤ 1. The above recurrence implies that

Tn (x) = 2n−1 xn + a1 xn−1 + · · · + an ,

where a1 , ..., an are integers.


One of the most important property of Chebyshev polynomials was discov-
ered by Chebyshev himself. It consists of the following.

75
Theorem 36. Let Pn (x) = xn + a1 xn−1 + · · · + an be a monic polynomial of
1 Tn (x)
degree n such that |Pn (x)| ≤ n−1 for |x| ≤ 1. then Pn (x) = n−1 . In other
2 2
Tn (x)
words, the polynomial n−1 is the monic polynomial of degree n that has the
2
least deviation from zero on [−1; 1].

Proof. Consider the polynomial

Tn (x)
Q(x) = − Pn (x).
2n−1
Tn (x)
Then deg Q(x) ≤ n − 1 since the leading term of n−1 and Pn (x) are equal.
2
1 kπ
Since |Pn (x)| ≤ n−1 for |x| ≤ 1, it follows that at the point xk = cos ,k =
2 n
0, 1, ..., n the sign of Q(xk ) coincides with the sign of Tn (xk ). Note that Tn (xk ) =
(−1)k . Therefore at the end points of each segment [xk+1 ; xk ], the polynomial
Q(x) take values of opposite signs, and hence Q(x) has a root on each of these
segments.
The number of segments [xk+1 ; xk ] is equal to n and thus the polynomial
Q(x) has at least n roots. For a polynomial of degree not greater than n − 1,
Tn (x)
this mean that it is identically zero, i.e., Pn (x) = n−1 .
2
Example 4.12. Prove that Tm (Tn (x)) = Tn (Tm (x)) for m, n ≥ 0.

Solution. Let x = cos φ, then Tn (x) = cos nφ = y and Tm (y) = cos m(nφ),
thus Tm (Tn (x)) = cos mnφ. Similarly, Tn (Tm (x)) = cos mnφ. Hence the iden-
tity Tn (Tm (x)) = Tm (Tn (x)) holds for |x| < 1, and therefore hold for all x.

hni
Example 4.13. Let m = . Prove that
2
m  
X n n−2j 2
Tn (x) = x (x − 1)j .
j=0
2j

Solution. If z = cos φ+i sin φ, then z +z −1 = 2 cos φ and z n +z −n = 2 cos nφ.


Therefore
z + z −1 z n + z −n
Tn ( )= .
2 2
76
z + z −1 z − z −1
Let x = and x = . Then y 2 = x2 − 1 and
2 2
n  
n −n n n
X n
z + z = (x + y) + (x − y) = (1 + (−1)i )xn−i y i
i=0
i
m   m   .
X n n−2j 2j X n n−2j 2 j
=2 x y =2 x (x − 1)
j=0
2j j=0
2j

Hence,
m  
X n
Tn (x) = xn−2j (x2 − 1)j
j=0
2j
holds for |x| ≤ 1, and therefore hold for all x.

Exercise
Exercise 4.7. Let P (x) be a polynomial with integer coefficients. Prove that
if
P (P (...P (x))) = x
for some integer x (where P is iterated n times), then P (P (x)) = x.
Exercise 4.8. Prove that P (x) = (x − a1 )(x − a2 )...(x − an ) + 1 is irreducible
over Z where n ≥ 5 and a1 , ..., an are different integers.
Exercise 4.9. Prove that P (x) = (x−a1 )2 (x−a2 )2 ...(x−an )2 +1 is irreducible
over Z where a1 , ..., an are different integers.
Exercise 4.10. Let R(x) be a rational function over R which takes integer
values at all integers x. Then R(x) is an arithmetic polynomial.
Exercise 4.11. Find necessary and sufficient conditions such that a polyno-
mial
f (x) = ax3 + bx2 + cx + d ∈ R[x]
is an arithmetic polynomial.
Exercise 4.12. If both α and cos απ are rational, then 2 cos απ is an inte-
1
ger,i.e., cos απ = 0, ± or ±1.
2
Exercise 4.13. Let p be an odd prime. Prove that
Tp (x) ≡ T1 (x) (mod p)

77
78
Tài liệu tham khảo

[1] Dương Quốc Việt - Đàm Văn Nhỉ, Giáo trình Đại số sơ cấp, NXB ĐHSP-
2007

[2] Dương Quốc Việt - Đàm Văn Nhỉ, Cơ sở lí thuyết số và đa thức, NXB
ĐHSP-2008.

[3] Jerome E. Kaufmann and Karent L. Schwitters, Elementary Algebra,


Ninth Edition, Brooks/Cole Cengage Learning, USA 2007.

[4] Terry H. Wesner, Elementary Algebra with Applications, McGraw-Hill


Companies; 3rd edition

[5] Victor.V. Prasolov, Polynomials, Translated from the Russian by Dimitry


Leites, Springer 2004.

79

You might also like