Contents

1 Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Completing the square and quadratic equations . . . . . . . . . 2
3 Factorizations and algebraic identities . . . . . . . . . . . . . . 14
4 Factoring expressions involving a b, b c, c a . . . . . . . . 27
5 Factoring a3 + b3 + c3 3abc . . . . . . . . . . . . . . . . . . . 36
6 AM-GM and Hölder’s inequality . . . . . . . . . . . . . . . . . 40
7 Lagrange’s identity and the Cauchy-Schwarz inequality . . . . . 52
8 Making linear combinations . . . . . . . . . . . . . . . . . . . . 65
9 Fixed points and monotonicity . . . . . . . . . . . . . . . . . . 77
10 The floor function . . . . . . . . . . . . . . . . . . . . . . . . . 83
11 Taking advantage of symmetry . . . . . . . . . . . . . . . . . . 95
12 Introductory problems . . . . . . . . . . . . . . . . . . . . . . . 111
13 Advanced problems . . . . . . . . . . . . . . . . . . . . . . . . . 117
14 Solutions to introductory problems . . . . . . . . . . . . . . . . 123
15 Solutions to advanced problems . . . . . . . . . . . . . . . . . . 157
16 Other Books from XYZ Press . . . . . . . . . . . . . . . . . . . 200
Preface 1

1 Preface

The main purpose of this book is to provide an introduction to central topics

in elementary Algebra from a problem-solving point of view. While working
with students who were preparing for various mathematics competitions or
exams, I observed that fundamental algebraic techniques were not part of
their mathematical repertoire. Since algebraic skills are not only critical to
Algebra itself but also to numerous other mathematical fields, a lack of such
knowledge can drastically hinder a student’s performance. Taking the above
observations into account, I put together this introductory book using both
simple and challenging examples which shed light upon essential algebraic
strategies and techniques, as well as their application in diverse meaningful
problems. This work is the first volume in a series of such books.

Regarding the structure of the book, the featured topics are elementary
and classical, including factorizations, algebraic identities, inequalities, alge-
braic equations and systems of equations. More advanced concepts such as
complex numbers, exponents and logarithms, as well as other topics are gen-
erally avoided. Nevertheless, some problems are constructed using properties
of complex numbers which challenge and expose the reader to a broader spec-
trum of mathematics. Each chapter focuses on specific methods or strategies
and provides an ample collection of accompanying problems that graduate in
difficulty and complexity. In order to assist the reader with verifying mastery
of the theoretical component, I included 105 problems in the last sections of
the book, of which 52 are introductory and 53 advanced. All problems come
together with solutions, many employing several approaches and providing the
motivation behind the solutions o↵ered.

Enjoy the problems!

 2 Completing the square and quadratic equations

2 Completing the square and quadratic equations

While the identity
(a + b)2 = a2 + 2ab + b2
is easy to check (simply expand the left hand-side, by writing it as (a+b)(a+b)),
things are a little bit more difficult in real life. Indeed, most of the time we
have to do exactly the opposite: we are given a quadratic expression and we
want to express it as a sum of squares (or some linear combination of squares).
The idea is quite simple: we fix one variable, say x of that expression, such
that the expression becomes a quadratic polynomial in x. Say our expression is
ax2 +bx+c for some real numbers a, b, c (which may themselves be complicated
expressions depending on other real numbers!). Then we complete the square
by writing ✓ ◆
2 2 b c
ax + bx + c = a x + x + =
a a
✓ ◆ ! ✓ ◆
b 2 b2 c b 2
a x+ + =a x+ ,
2a 4a2 a 2a 4a

where = b2 4ac is the discriminant of the expression ax2 +bx+c. This elimi-
b 2
nates the variable x, by including all of its appearances in the term a x + 2a .
Now, 4a may (or may not...) itself be some quadratic expression in di↵erent
variables, so we can apply the same reasoning to write it as a sum of squares.
In particular, the previous discussion applies to the quadratic equation

ax2 + bx + c = 0,

where a, b, c are given real numbers, with a nonzero (if a = 0, then we obtain
a linear equation). The previous paragraph shows that the equation can be
written as ✓ ◆
b 2
x+ = 2.
2a 4a
If the equation has real solutions, then the left hand-side must be nonnegative
(as is the square of any real number). Hence so must be the right hand-side,
which means that 0. In this case, we can solve the previous equation by
taking square roots, and we end up with the solutions
p p
b+ b
x1 = , x2 = ,
2a 2a
which are equal if and only if = 0. Hence we can summarize our discussion
in:
Completing the square and quadratic equations 3

Theorem 2.1. Let a, b, c be real numbers with a 6= 0 and let

= b2 4ac.

Then the quadratic equation

ax2 + bx + c = 0

has either:
• no real solution, if < 0.
• exactly one real solution, if = 0.
• two real solutions if > 0.

Note that the previous discussion also gives a nice way of solving quadratic
inequalities or proving inequalities involving quadratic expressions: since
✓ ◆2 !
2 b
ax + bx + c = a x+ + ,
2a 4a2

we see that the expression ax2 + bx + c has constant sign (equal to that of
a) when  0. On the other hand, if > 0 and if x1  x2 are the real
solutions of the equation ax2 + bx + c = 0, then the inequality ax2 + bx + c  0
is equivalent to a(x x1 )(x x2 )  0. If a > 0, this is in turn equivalent to
x 2 [x1 , x2 ], while if a < 0, this is equivalent to x 2
/ (x1 , x2 ). To summarize,
suppose for the sake of simplicity that a > 0. Then
• if = b2 4ac < 0, then ax2 + bx + c > 0 for all real numbers x.
• If = 0, then ax2 + bx + c 0 for all real numbers x, with equality if
b
and only if x = 2a .
• If > 0, then the equation ax2 + bx + c = 0 has two real roots, say
x1 < x2 , and we have ax2 + bx + c < 0 if and only if x 2 (x1 , x2 ).
A consequence of this discussion is the following important fact (which is
a very special case of a general theorem in real analysis):

Theorem 2.2. Let f (x) = ax2 + bx + c be a quadratic polynomial and let

u  v be real numbers such that f (u)f (v) < 0. Then the equation f (x) = 0
has at least one solution in (u, v).

Proof. Since f changes sign between u and v, its discriminant = b2 4ac

must be positive and so the equation f (x) = 0 has two distinct solutions
x1 < x2 . If none of them belongs to (u, v), then the previous discussion shows
that either f (u), f (v) > 0 or f (u), f (v) < 0 (according to the sign of a). But
this contradicts the hypothesis that f (u) and f (v) have di↵erent signs.
 4 Completing the square and quadratic equations

Actually the previous theorem holds for any polynomial function (and
more generally for continuous functions), but the proof is beyond the scope of
this introductory book. Another very important result concerning quadratic
equations is the following:

Theorem 2.3. (Vieta’s relations for quadratic equations) Let a, b, c be real

numbers, with a 6= 0 and let x1 , x2 be the roots of the equation ax2 +bx+c = 0.
Then
b c
x1 + x2 = and x1 x2 = .
a a
Proof. Since ax2 + bx + c = 0 has roots x1 , x2 , we must have an equality of
polynomials

ax2 + bx + c = a(x x1 )(x x2 ) = ax2 a(x1 + x2 )x + ax1 x2 .

Identifying coefficients yields

b c
x1 + x2 = and x 1 x2 = ,
a a
which is exactly what we wanted to prove.

We remark that the previous theorem holds for complex roots, and roots
with multiplicity, with the same proof.
It is now time for practice: we will see how the above theoretical facts
really apply in practice.

Example 2.1. Solve the equation

(2x 1)2 (3x 1)2 (6x 1)2

+ + = 1.
2 3 6
Solution. Expanding each term and collecting terms according to the suc-
cessive powers of x yields the following equivalent equations
1 1 1
2x2 2x + + 3x2 2x + + 6x2 2x + = 1,
2 3 6
11x2 6x = 0 or x(11x 6) = 0.
6
Hence the solutions are x = 0 and x = 11 .

Example 2.2. Find the greatest integer n for which the equation
1 1 1
+ =0
x 1 nx x + 1
has real solutions.
Completing the square and quadratic equations 5

Solution. Write the equation successively as

1 1 1
+ = ,
x 1 x+1 nx

then x22x 1 = nx
1
, 2nx2 = x2 1 and finally ( 2n + 1)x2 = 1. Since x2 0
for all real numbers x, we deduce that if the equation has real solutions, then
1
2n + 1 > 0, hence n  0. We cannot have n = 0, since then nx wouldn’t
make sense. Hence the largest n is at most 1. And indeed n = 1 gives the
real solutions x = p13 and x = p13 , so the answer is n = 1.

Example 2.3. Solve the system of equations

(
x y=3
x2 + (x + 1)2 = y 2 + (y + 1)2 + (y + 2)2 .

Solution. We take advantage of the fact that the first equation is very simple
and express x = y + 3, replacing this value in the second equation. This yields
the quadratic equation

(y + 3)2 + (y + 4)2 = y 2 + (y + 1)2 + (y + 2)2 .

Expanding each term and collecting similar terms, we obtain the equivalent
equation

2y 2 + 14y + 25 = 3y 2 + 6y + 5 or y2 8y 20 = 0.

Solving this equation gives y = 2, 10 and since x = y + 3, we obtain the

solutions (x, y) = (1, 2) and (13, 10).

Example 2.4. Evaluate

1 1
p p +p p ,
x+2 x 1 x 2 x 1

where 1  x < 2.

Solution. We start by simplifying each fraction, by completing the squares

at the denominator. We have
p p p
x+2 x 1=x 1+2 x 1+1=( x 1 + 1)2

and
p p p
x 2 x 1=x 1 2 x 1+1=( x 1 1)2 .
 6 Completing the square and quadratic equations

p p
Thus, paying attention to the fact that a2 = |a| and x 1 1 < 0 (since
x < 2), we obtain

1 1 1 1
p p +p p = p + p
x+2 x 1 x 2 x 1 1+ x 1 1 x 1

2 2 2
= p p = = .
(1 + x 1)(1 x 1) 1 (x 1) 2 x

Example 2.5. Solve the system of equations

8 1
>
<x + y = 1
y + z1 = 12
>
:
z + x1 = 2.

Solution. The idea is quite simple: we express everything in terms of one

variable. Namely, from the first equation we can express x in terms of y,
obtaining x = 1 y1 . The second equation gives

2
z= .
1 2y

Replacing these values in the last equation, we obtain

2 y
= 2.
1 2y 1+y

Clearing denominators and simplifying the resulting equation, we arrive at

y + 2y 2 = 0.

Note that y 6= 0, since otherwise y1 wouldn’t make sense. We conclude that

y = 12 . Coming back to x = 1 y1 and z = 1 22y , we obtain x = z = 1,
hence the system has the unique solution (1, 12 , 1).

Example 2.6. Solve the equation

1 1 1
+ + = 1.
3x 1 4x 1 7x 1
Solution. The algebra would be quite nasty if we tried to clear denominators.
Instead, we rewrite the equation as

1 1 1
+ =1
3x 1 4x 1 7x 1
Completing the square and quadratic equations 7

or equivalently
4x 1 + 3x 1 7x 1 1
= .
(3x 1)(4x 1) 7x 1
We remark the common factor 7x 2, which already gives us the solution
x = 27 . Suppose that x 6= 27 is another solution. Then dividing the previous
relation by 7x 2 yields
1 1
= or 12x2 7x + 1 = 7x 1.
(3x 1)(4x 1) 7x 1
This can be further simplified to 6x2 7x+1 = 0. Solving this quadratic equa-
tion yields the other solutions x = 1, 16 of the equation. Hence the equation
has three solutions, given by 27 , 1, 16 .
Example 2.7. Find all pairs (a, b) of positive real numbers such that
9 4
4a + 9b =
+ = 12.
a b
Solution. We write the second equation as
9b + 4a
= 12
ab
and we observe that the numerator equals 12 by hypothesis. Thus ab = 1,
that is b = a1 . Replacing this value of b in the equation 4a + 9b = 12 we
obtain 4a + a9 = 12. Clearing denominators, we obtain a quadratic equation
4a2 12a + 9 = 0, which has the unique solution a = 32 . Going back to the
system, we obtain b = 23 .
If you found the first step (establishing that ab = 1) tricky, we can work
more directly as follows: from the equation 4a + 9b = 12 we express b in terms
of a. We replace this value of b in the equation a9 + 4b = 12, obtaining a
quadratic equation in a, with the unique solution a = 32 .
1 1
Example 2.8. If a is a real number such that a a = 1, find a4 + a4
.
Solution. It is easier to realize what you shouldn’t do in this exercise: you
should not solve the equation a a1 = 1 and then plug in the values you get to
compute a4 + a14 (of course, with a lot of nasty computations one would obtain
the desired answer, but this is far from being an elegant approach). Let us
take the square of the given relation a a1 = 1, and obtain
1 1
a2 + 2
2 = 1, that is a2 + 2 = 3.
a a
Now, all we have to do is to repeat the process: we take the square of the last
relation and obtain
1 1
a4 + 4 + 2 = 9, hence a4 + 4 = 9 2 = 7.
a a
 8 Completing the square and quadratic equations

Example 2.9. Solve the equation

x4 97x3 + 2012x2 97x + 1 = 0.

Solution. The key point is that the equation is symmetric. Dividing by x2 ,

we obtain the equivalent equation
97 1
x2 97x + 2012 + 2 =0
x x
We reduce this to a quadratic equation by setting
1
x+ = y.
x
1
Then x2 + x2
+ 2 = y 2 , hence the previous equation becomes

y2 97y + 2010 = 0, or (y 30)(y 67) = 0.

Thus y = 30 or y = 67. Now, remember that y = x + x1 , hence we obtain the

quadratic equation x2 xy + 1 = 0. Solving the two equations x2 30x + 1 = 0
and x2 67x + 1 = 0 gives the solutions
p
67 ± 4485 p
x= , and x = 15 ± 224.
2
Example 2.10. Let a, b, c be real numbers such that a b c. Prove that

a2 + ac + c2 3b(a b + c).

Solution. Let us rewrite the inequality as

3b2 3b(a + c) + a2 + ac + c2 0.

This is a quadratic inequality in b. The discriminant is

9(a + c)2 12(a2 + ac + c2 ) = 3(a2 2ac + c2 ) = 3(a c)2  0,

thus the polynomial 3x2 2x(a + c) + a2 + ac + c2 takes only nonnegative

values (its leading coefficient 3 is positive). In particular, its value at b is
nonnegative, and the result follows.
We can also try to complete squares, writing the inequality as

12b2 12b(a + c) + 4(a2 + ac + c2 ) 0,

then
3(2b (a + c))2 + (a c)2 0.
Completing the square and quadratic equations 9

Example 2.11. Prove that 3(x + y + 1)2 + 1 3xy for all x, y 2 R.

Solution. Let us write x + y = a and xy = b. We need to prove that

3(a + 1)2 + 1 3b. Now, the equation t2 at + b = 0 has the real solutions
x, y, hence its discriminant is nonnegative, that is a2 4b (of course, we can
also give a direct proof, since the inequality is equivalent to (x y)2 0).
a2
Thus b  4 and it suffices to prove that

3 2
3(a + 1)2 + 1 a .
4
Multiplying by 4, expanding (a + 1)2 and rearranging terms reduces the in-
equality to
9a2 + 24a + 16 0,
equivalent to (3a + 4)2 0, thus true.
We note that an alternative solution consists in completing the square,
which allows us to rewrite the inequality as
✓ ◆2 ✓ ◆2
1 3
3 x+ y+1 + y+1 0.
2 2

Example 2.12. Find all pairs (x, y) of real numbers such that

4x2 + 9y 2 + 1 = 12(x + y 1).

Solution. Let us separate the variables by writing the equation in the form

4x2 12x + 9y 2 12y + 13 = 0.

Next, we complete the square to obtain

(2x 3)2 + (3y 2)2 = 0.

Since a sum of squares equals zero if and only if each square is zero, it follows
that 2x 3 = 0 and 3y 2 = 0. Thus there is only one solution, given by
x = 32 and y = 23 .

Example 2.13. Prove that if a b > 0, then

(a b)2 a+b p (a b)2

 ab  .
8a 2 8b
Solution. We complete the square to obtain
p p p p 2
a+b p a 2 a b+b ( a b)
ab = = .
2 2 2
 10 Completing the square and quadratic equations

On the other hand, we have

p 2 p 2 p p p p
a b= a b =( a b)( a + b).
p p 2
Dividing by ( a b) , we are therefore reduced to proving the inequalities
p p p p
( a + b)2 1 ( a + b)2
  .
8a 2 8b
The inequality on the left is equivalent (after multiplication by 8a and taking
square roots) to p p p
a+ b2 a
and is an immediate consequence of a b. We proceed similarly with the
inequality on the right.
Example 2.14. Simplify the expression
4 12 9
+ 2 .
4x2 2
+ 12x + 9 6x + 5x 6 9x 12x + 4
Solution. Here it is easier to say what you should not do: clear denominators!
Indeed, that would give a terrible mess and chances to solve the exercises with
this approach are close to zero. Instead, let us analyze a little bit each term
in the sum, more precisely its denominator. Each denominator is a quadratic
polynomial in x, so a natural approach would be to see whether it can itself be
factored. Of course, a sum of products is not something very enlightening, but
one might hope that the denominators have a common factor. Well, solving the
quadratic equations 4x2 + 12x + 9 = 0, 6x2 + 5x 6 = 0 and 9x2 12x + 4 = 0,
or by completing the square we end up with
4x2 + 12x + 9 = (2x + 3)2 ,
6x2 + 5x 6 = (2x + 3)(3x 2)
2 2
9x 12x + 4 = (3x 2) .
It turns out that the don’t have a common factor, but they have a quite good
shape: if a = 2x + 3 and b = 3x 2, then our expression is simply
4 12 9 4b2 12ab + 9a2
+ 2 = .
a2 ab b (ab)2
Again, it is easy to recognize that the numerator is the square of (2b 3a)2 .
Since
2b 3a = 2(3x 2) 3(2x + 3) = 13,
we obtain the nice equality
✓ ◆2
4 12 9 13
2 2
+ = .
4x + 12x + 9 6x + 5x 6 9x2 12x + 4 2
6x + 5x 6
Completing the square and quadratic equations 11

Example 2.15. Solve in real numbers the equation

x4 + 16x 12 = 0.

Solution. We will try to find a, b, c such that the left hand-side can be written
as (x2 + a)2 (bx + c)2 . If we can find such numbers a, b, c, then solving the
equation will come down to solving two quadratic equations x2 + a = bx + c
and x2 + a + bx + c = 0.
The identity

x4 + 16x 12 = (x2 + a)2 (bx + c)2

is equivalent to the chain of equalities

2a = b2 , 16 = 2bc, a2 c2 = 12.
b2 8
Thus a = 2, c= b and replacing these in the last equation yields

b4 64
= 12.
4 b2

Let b2 = 4d. The equation becomes 4d2 16 d = 12 and we easily recognize the
2
root d = 1. Thus we can take b = 2 and then a = b2 = 2 and c = 8b = 4.
Now, it remains to solve the equations x2 +2 = 2x 4 and x2 +2 = 2x+4.
The first one has no real solutions since it can be written as (x 1)2 + 5 = 0,
while the second one can be written (x + 1)2 = 3 and has the solutions
p p
x1 = 1 3, x2 = 3 1.

Example 2.16. The equation x4 4x = 1 has two real roots. Find their
product.

Solution. Let us add 2x2 + 1 to both terms, in order to complete the square
in the left hand-side. We obtain the equivalent equation

(x2 + 1)2 = 2x2 + 4x + 2 = 2(x + 1)2 .

p p
This is equivalentpto x2 + 1 p = 2(x + 1) or x2 + 1 = 2(x + 1).
p The first
equation is x 2 2x + 1 2 = 0 and its discriminant is = 4 2 2 > 0.
Hence it hasptwo solutions x1 , x2 and their product is given by Vieta’s formulae:
x1 x 2 = 1 2. Since we already know that the initial equation has two real
roots, they must be x1 , x2 and we are done.
p Of course, it would be very easy
to check that the equation x2 + 1 = 2(x + 1) has no real root,
p since its
discriminant is negative. Thus the answer of the problem is 1 2.
 12 Completing the square and quadratic equations

We can also solve this as follows we look for a, b, c such that

x4 4x 1 = (x2 + a)2 (bx + c)2

for all x, which is equivalent to

2a = b2 , bc = 2, a2 c2 = 1.
b2 2
Replacing a = 2 and c = b in the last equation yields

b4 4
= 1.
4 b2
Setting b2 = 4d, this gives us the third degree equation 4d3 + dp 1 = 0, with
the apparent solution d = 12 . Thus b2 = 2 and we can take b = 2, then a = 1
p
and c = 2. The original p equation
p is therefore reduced
p topthe resolution of
the equations x2 + 1 = 2x + 2 and x2 + 1 = p 2x 2. As above, we
obtain the product of the real roots equals 1 2.
Example 2.17. Solve in real numbers the equation

(x + 1)(x + 2)(x + 3)(x + 4) = 360.

Solution. Expanding the product is out of the question, so there is certainly

some trick here. We try to pair the factors in the product defining the left
hand-side. If we pair the first two and then the last two we obtain the equation
(x2 +3x+2)(x2 +7x+12) = 360, which is not simpler. The same thing happens
if we pair the first and the third factor, but a miracle happens if we pair the
first and the last factor: we obtain the equation

(x2 + 5x + 4)(x2 + 5x + 6) = 360,

which is fourth degree equation in x, but a quadratic equation in y = x2 +

5x. And while solving quartic equations is hard, solving quadratic ones is
straightforward. Namely, the equation (y + 4)(y + 6) = 360 is equivalent to
y 2 + 10y 336 = 0, with solutions y = 14 and y = 24. Next, we have to
solve the equations x2 + 5x = 14 and x2 + 5x = 24. The discriminant of the
second one is negative, so it does not give real solutions. On the other hand,
the equation x2 + 5x = 14 has solutions x = 7 and x = 2. Hence these two
numbers are the solutions of the initial equation.
Example 2.18. Find all n > 1 such that

x21 + x22 + ... + x2n xn (x1 + x2 + ... + xn 1)

for all real numbers x1 , ..., xn .

Completing the square and quadratic equations 13

Solution. We write the inequality as

x21 x1 xn + x22 x2 xn + ... + x2n 1 xn 1 xn + x2n 0.

We complete the squares to get the equivalent inequality

⇣ xn ⌘2 ⇣ xn ⌘2 n 1
x1 + ... + xn 1 x2n + x2n 0,
2 2 4
that is ⇣ xn ⌘2 ⇣ xn ⌘2 n 5 2
x1 + ... + xn 1 xn .
2 2 4
If n  5, then the right hand-side is nonpositive and the left hand-side is
nonnegative, hence the inequality holds. On the other hand, if n > 5, then we
can choose
xn xn
x1 = , ..., xn 1 = , xn = 1
2 2
and the inequality is no longer true. Hence the answer is n = 2, 3, 4, 5.
 14 Factorizations and algebraic identities

3 Factorizations and algebraic identities

There are a few classical algebraic identities that play a crucial role in almost
all branches of mathematics. In this section we recall some of them and
give many examples of applications to factorization of algebraic expressions.
Being able to recognize factorizations of (sometimes complicated) algebraic
expressions is fundamental, since this often plays an important role in solving
equations, systems of equations or proving inequalities.
A first fundamental identity is

a2 b2 = (a b)(a + b).

This holds for all real numbers a, b (and is actually much more general than
that, but we will stick to real numbers from now on) and follows easily by ex-
panding the right hand-side and canceling the terms ab and ba. Though very
simple, this identity is crucial when factoring or simplifying more complicated
algebraic expressions. It is also a special case of a more general, and also very
handy identity, which describes a partial factorization of an bn . Note that
an bn vanishes when a = b, hence it must have a factor of a b. We actually
have
an bn = (a b)(an 1 + an 2 b + ... + abn 2 + bn 1 )
for all real numbers a, b and all positive integers n. Indeed, we have

(a b)(an 1
+ an 2
b + ... + abn 2
+ bn 1
)=

a(an 1
+ an 2
b + ... + bn 1
) b(an 1
+ ... + bn 1
)=
an + a n 1
b + ... + abn 1
an 1
b ... abn 1
b n = an bn ,
by canceling equal terms. For instance, if n = 3 we get the very useful identity

a3 b3 = (a b)(a2 + ab + b2 )

and for n = 4 we obtain

a4 b4 = (a b)(a3 + a2 b + ab2 + b3 ).

One may wonder if we can still factor a2 + ab + b2 and a3 + a2 b + ab2 + b3 . This

is not the case for a2 + ab + b2 , but the answer is positive for a3 + a2 b + ab2 + b3 ,
since

a3 + a2 b + ab2 + b3 = a2 (a + b) + b2 (a + b) = (a + b)(a2 + b2 ).

We obtain therefore

a4 b4 = (a b)(a + b)(a2 + b2 ).