Equations

Adithya B., Brian L., William W., Daniel X.

4/29

§1 Algebraic Manipulations
There are rarely systematic ways to solve general equations or systems of equations, like
those exhibited with linear equations or quadratics. However, many contest equations are
contrived in a way such that they are made easy with careful manipulations. Thus, being
familiar with some of the more common algebraic tricks can go a long way to solving
general classes of equations.

Due to this unit’s open ended nature, we don’t really have many rigorous theorems.
However, there are a number of important ideas to keep in mind when solving algebraic
equations:

• If the condition is symmetric, try to keep it symmetric. Sometimes, making

a substitution to create symmetry is also important. (e.g. Given a quadratic
(x − 5)(x − 1) = 7 we can ”create symmetry” by substituting y = x − 3 to center
the LHS around 0.)

• Often times, an equations problem will not ask you to solve for each variable,
rather an expression containing them, such as their sum. In this case, it is rarely
necessary to actually solve for each individual variable. Use the answer form to
your advantage when solving, and try to transform what you are given to what you
need

• Remember your factorizations so you can spot factorable expressions when working
with big algebraic expressions. (e.g. The expression a3 + b3 + c3 − 3abc looks a lot
more tenable when we realize it is (a + b + c)(a2 + b2 + c2 − ab − ac − bc))

• Know common substitutions associated with certain conditions. For example, when
given xyz = 1, a common substitution would be to set x = ab , y = cb , z = ac . Solving
for a, b, c (up to scaling) can then get us x, y, z.
There exist many other useful, though less common substitutions, such as xyz =
x + y + z + 2 =⇒ ∃ a, b, c such that x = b+c c+a a+b
a , y = b , z = c . It’s not necessarily
worth memorizing these, but just be on the lookout for substitutions like these that
transform messy conditions into easier ones (or eliminate constraints like xyz = 1
entirely).

1
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

§2 Polynomials
We begin with the simplest type of equations: polynomials. While there are several
theorem regarding the roots of polynomials (such as the Rational Root Theorem or
Descartes’ Rule of Signs), we’ll focus on techniques that apply to a broader range of
problems: substitution, factoring, and the like.

Example 2.1
Solve 64x6 + 64x5 − 320x4 − 448x3 − 80x2 + 4x + 1 = 0.

Solution. Note that if we substitute u = 2x, we will decrease all the coefficients while
still keeping them integers. We get the new equation is

u6 + 2u5 − 20u4 − 56u3 − 20u2 + 2u + 1 = 0

Now, note that this sequence of coefficients is symmetric, so we center the degrees of the
coefficients about 0, so we divide by u3 . Now, we get
     
3 1 2 1 1
u + 3 + 2 u + 2 − 20 u + − 56 = 0
u u u

Now, note that each of these terms may be expressed in terms of v = u + u1 . Thus, we
have
(v 3 − 3v) + 2(v 2 − 2) − 20v − 56 = 0
so v 3 + 2v 2 − 23v − 60 = 0. Now, we test possible values of v until we see that

(v + 3)(v + 4)(v − 5) = 0
√
−3± 5
√ √
Thus, we have u + 1
u = −3, −4, 5, so u = 2 , −2 ± 3, 5±2 21 , so
√ √ √
−3 ± 5 −2 ± 3 5 ± 21
x= , , .
4 2 4

Example 2.2 (2000 AIME II # 13)

The equation√ 2000x6 + 100x5 + 10x3 + x − 2 = 0 has exactly two real roots, one of
which is m+r n , where m, n and r are integers, m and r are relatively prime, and
r > 0. Find m + n + r.

Solution. In general, in order to solve a polynomial we’ll need to factor it completely,

or rewrite it in some clever form (e.g. (3x2 − 1)2 = (2x2 − 1)2 ; can you see why this is
useful?). Given the answer form, the problem seems to suggest that we need a quadratic
factor of the polynomial. But the polynomial seems extremely hard to factor at first
glance; there are no rational roots (since there are only two roots) and no obvious factors.
So let’s try grouping the coefficients and factoring them; maybe we can extract a common
factor this way. We note that the middle three terms form a geometric sequence:

1000x6 − 1
x + 10x3 + 100x5 = x · .
10x2 − 1

2
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

This is great, because the other two terms have a factor of 1000x6 − 1 :

2000x6 − 2 = 2(1000x6 − 1)

so
(1000x6 − 1)(20x2 + x − 2)
 
6 5 3 6 x
2000x +100x +10x +x−2 = (1000x −1) 2 + = .
10x2 − 1 10x2 − 1

This can be rewritten as

(100x4 + 10x2 + 1)(20x2 + x − 2)

so the two real roots of the polynomial come from the second factor, which are
√ √
−1 ± 12 + 4 · 20 · 2 −1 ± 161
= .
40 40
So the answer is −1 + 161 + 40 = 200.

§3 Combining Equations
When faced with systems of equations, it can be useful to consider all of the equations as
a whole. Instead of focusing on each equation individually and trying to solve it, we can
try combining the given equations in a clever way that either directly gives us an answer
or creates cleaner equations to solve.

Example 3.1 (1990 AIME 15)

If a, b, x and y are real numbers such that ax + by = 3, ax2 + by 2 = 7, ax3 + by 3 = 16,
and ax4 + by 4 = 42, find ax5 + by 5 .

Solution. We should first ask ourselves how we are going to get the desired answer. In
this case, multiplying the equations we have may seem like a good idea since we can
easily get x5 , however we also get a coefficient of a2 which is hard to convert to an a.
Instead, since we are given a “sequence” of equations of the form axn + by n , we should
think about how to get from axn + by n to axn+1 + by n+1 . Naively, we can try to multiply
the first expression by (x + y) and see what we get. This gives

(x + y)(axn + by n ) = (axn+1 + by n+1 ) + xy(axn−1 + by n−1 )

This is really good, since we know a lot of values of axn + bxn , so we should be able to
set up a system to find x + y and xy. In particular, substituting in n = 2, 3 gives

7(x + y) = 16 + 3xy and 16(x + y) = 42 + 7xy

Solving, we get x + y = −14, xy = −38. So,

ax5 + bx5 = (x + y)(ax4 + bx4 ) − xy(ax3 + bx3 ) = −14 · 42 + 38 · 16 = 20

3
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

Example 3.2 (Mildorf Mock AIME)

a, b,and c are complex numbers such that

a+b+c=1
a2 + b2 + c2 = 3
a3 + b3 + c3 = 7

Find a7 + b7 + c7 .

Solution. This ideas in solution should seem familiar to the solution of Example 3.1.
Let Sn = an + bn + cn . The first step is to notice that Sy is going to be hard to calcu-
late by manipulating the given equations. We need some sort of general manipulation
we can use to get to S7 . Therefore, the idea is took for some sort of recursion for Sn .
We are given the initial terms S1 , S2 , and S3 , so we can calculate S7 if we have a recursion.

To create a term of the form an+1 , we want to multiply an by a, so one common idea
may be to multiply an + bn + cn by a + b + c. We obtain,

Sn (a + b + c) = Sn+1 + (an b + an c + bn a + bn c + cn a + cn b)

We want to somehow express the second term in terms of previous Si . Then, we notice
that we will get terms of that form if we multiply Sn−1 by ab + bc + ca. Therefore, we
get from expansion

(ab + bc + ca)Sn−1 = (an b + an c + bn a + bn c + cn a + cn b) + an−1 bc + bn−1 ac + cn−1 ab

= (an b + an c + bn a + bn c + cn a + cn b) + abcSn−2

Thus, we have obtained (an b + an c + bn a + bn c + cn a + cn b) in terms of Sn−1 and Sn−2 ,

so we can substitute this back into the first equation.

Sn (a + b + c) = Sn+1 + (ab + bc + ca)Sn−1 − abcSn−2

Sn+1 = (a + b + c)Sn − (ab + bc + ca)Sn−1 + abcSn−2

Now, our recursion is almost complete. We just have to determine the values of ab+bc+ca
and abc. The former is quite simple. Note that

(a + b + c)2 − (a2 + b2 + c2 ) 1−3

ab + bc + ca = = = −1
2 2
For the latter, we can use the following factorization:

a3 + b3 + c3 − 3abc = (a + b + c)(a2 + b2 + c2 − ab − bc − ca)

Substituting what we know,

7 − 3abc = 4 =⇒ abc = 1.

Now, our recursion becomes Sn+1 = Sn + Sn−1 + Sn−2 . From the given values, we have
S4 = 11, S5 = 21, S6 = 39, and S7 = 71.

4
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

Example 3.3 (2014 HMMT Algebra 9)

Given that a, b, c are complex numbers satisfying

a2 + ab + b2 = 1 + i
b2 + bc + c2 = −2
c2 + ca + a2 = 1

compute (ab + ac + bc)2 .

Solution. Like the previous question, we should note that we are asked to find (ab+ac+bc)2
instead of a, b, c. What this most likely means is that a, b, c are quite difficult to find.
Expanding, we see that we want to find

(ab + ac + bc)2 = a2 b2 + a2 c2 + b2 c2 + 2a2 bc + 2ab2 c + 2abc2

The easiest way to get an expression with the a2 bc terms is to just multiply two of the
expressions we are given. For example,

(b2 + bc + c2 )(c2 + ca + a2 ) = a2 b2 + a2 c2 + b2 c2 + c4 + bc3 + ac3 + (a2 bc + ab2 c + abc2 )

Recall, though, that we want to preserve symmetry, since both our expressions, and our
desired result are symmetrical. We can do this by adding this result with our other two
similar products. Namely,

(a2 + ab + b2 )(a2 + ac + c2 ) + (a2 + ab + b2 )(b2 + bc + c2 ) + (a2 + ac + c2 )(b2 + bc + c2 )

= (a4 +b4 +c4 )+3(a2 b2 +a2 c2 +b2 c2 )+(a3 b+b3 a+a3 c+c3 a+b3 c+c3 b)+3(a2 bc+ab2 c+abc2 )
Now, let’s try to address the annoying a4 and a3 b terms. Note that these terms all appear
in
(a2 + ab + b2 )2 = a4 + b4 + 2a3 b + 2ab3 + 3a2 b2
So, if we try a similar symmetric sum as before, note that

(a2 + ab + b2 )2 + (a2 + ac + c2 )2 + (b2 + bc + c2 )2

= 2(a4 + b4 + c4 ) + 2(a3 b + b3 a + a3 c + c3 a + b3 c + c3 b) + 3(a2 b2 + a2 c2 + b2 c2 )

Aha! The coefficients work out such that if I multiply the first expression by 2 and
subtract the second expression, I eliminate everything I don’t want! Doing this, we get
X X
2 (a2 + ab + b2 )(a2 + ac + c2 ) − (a2 + ab + b2 )2

= 3(a2 b2 + a2 c2 + b2 c2 ) + 6(a2 bc + ab2 c + abc2 ) = 3(ab + ac + bc)2

So, our final answer is

2((1 + i)(−2) + (1 + i)(1) + (−2)(1)) − ((1 + i)2 + (−2)2 + (1)2 ) 11 4

=− − i
3 3 3

5
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

Note: It looks like the last step was a bit magical, as not only did everything we didn’t
want cancel out, but we were left with exactly our final answer. Of course, we knew from
the beginning this would happen, as this is a contest question, and thus there must exist
a certain level of artificiality in order for this question to work, since general systems of
quadratics are quite difficult to solve.

In this case, however, there is a deeper idea behind just the random algebraic iden-
tity we obtained. For the interested reader, note that a2 + ab + b2 looks a lot like Law
of Cosines when the angle is 120◦ . See if you can find a geometric interpretation of the
question.

§4 More Systems

Example 4.1 (2005 Iberoamerican/1)

Determine all triples of real numbers (x, y, z) such that

xyz = 8
x y + y z + z 2 x = 73
2 2

x(y − z)2 + y(z − x)2 + z(x − y)2 = 98.

Solution. The ugliest expression in this system is the last one, so let’s expand it to see if
it is simplifiable. Expanding, we get

x2 y + y 2 z + z 2 x + x2 z + y 2 x + z 2 y − 6xyz = 98

Using the other two equations, we get x2 z + y 2 x + z 2 y = 98 + 6 · 8 − 73 = 73. So, we have

x2 y + y 2 z + z 2 x = xy 2 + yz 2 + zx2 =⇒ (x − y)(x − z)(y − z) = 0

As the equations are cyclic, we can assume that x = y. Now, the first two equations
become
x2 z = 8, x3 + x2 z + z 2 x = 73
The second equation rearranges as z 2 = 65 2
x − x , and substituting into the first equation
yields 65x − x = 64. This is a quadratic in x3 which yields
3 6

x3 = 1, 64 =⇒ x = 1, 4

So, our solution set is (1, 1, 8), 4, 4, 12 , and cyclic permutations.

Note: Though this example is easier than the other ones, it demonstrates the importance
of knowing factorizations. After we get x2 y + y 2 z + z 2 x = xy 2 + yz 2 + zx2 , there are ways
to show that we must have two of them are equal with inequalities, but you can save
yourself a lot of time by quickly recognizing that the expression is actually factorable.

§5 A Tricky Example

6
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

Example 5.1 (2014 AIME 14)

Let m be the largest real solution to the equation
3 5 17 19
+ + + = x2 − 11x − 4
x − 3 x − 5 x − 17 x − 19
p √
There are positive integers a, b, and c such that m = a + b + c. Find a + b + c.

Solution. The key to this problem is to try to preserve symmetry for as long as possible.
First, note that adding 1 to each of the fractions on the left makes them a lot simpler, so
let’s break up the −4 on the RHS to do that. We get
x x x x
+ + + = x(x − 11)
x − 3 x − 5 x − 17 x − 19
Zero definitely isn’t going to be the largest real solution to this equation, so we can safely
divide by x as well. Now, the resultant expression is a quintic, which is unfortunately
unsolvable generally. However, in this case, we have symmetry on our side. Note that
both the LHS and RHS are ”centered” around x − 11. So, if we make the substitution
y = x − 11, our equation becomes
1 1 1 1
+ + + =y
y+8 y+6 y−6 y−8
Now, if we group together similar terms on the left, we get
2y 2y 1 1 1
+ 2 = y =⇒ 2 + 2 =
y2 − 64 y − 36 y − 64 y − 36 2
This is a quadratic now, which we can solve, but let’s continue our symmetry argument.
The LHS is now symmetric around y 2 − 50, so if we let z = y 2 − 50, we get
1 1 1
= + =⇒ z 2 − 196 = 4z
2 z − 14 z + 14
√ p √
Quadraticpformula gives z = 2 + 200, so back substituting gives y = 52 + 200 =⇒
√
x = 11 + 52 + 200. So, our answer is 200 + 52 + 11 = 263.

7
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

§6 Problems
Problem 6.1 (2020 CMIMC Algebra #1). Suppose x is a real number such that
x2 = 10x + 7. Find the unique ordered pair of integers (m, n) such that x3 = mx + n.

Problem 6.2 (Brilliant). Suppose that a and b are real numbers such that ab = 3 and
7 6 −a6 b+b7
a2 + b2 = 8. Find the value of a −aba+b .

Problem 6.3 (2000 AIME I #7). Suppose that x, y, and z are three positive numbers
that satisfy the equations xyz = 1, x + z1 = 5, and y + x1 = 29. Then z + y1 = m
n , where
m and n are relatively prime positive integers. Find m + n.

Problem 6.4 (2010 AIME I #9). Let (a, b, c) be the real solution of the system of
equations x3 − xyz = 2, y 3 − xyz = 6, z 3 − xyz = 20. What is the greatest possible value
of a3 + b3 + c3 ?

Problem 6.5 (2018 CMIMC Algebra #5). Suppose that a,b, and c are nonzero real
numbers such that
1 2 7 1
bc + = ca + = ab + = .
a b c a+b+c
Find a + b + c.

Problem 6.6 (1989 AIME #8). Assume that x1 , x2 , . . . , x7 are real numbers such that

x1 + 4x2 + 9x3 + 16x4 + 25x5 + 36x6 + 49x7 = 1

4x1 + 9x2 + 16x3 + 25x4 + 36x5 + 49x6 + 64x7 = 12
9x1 + 16x2 + 25x3 + 36x4 + 49x5 + 64x6 + 81x7 = 123.

Find the value of 16x1 + 25x2 + 36x3 + 49x4 + 64x5 + 81x6 + 100x7 .

Problem 6.7 (AoPS). Let α satisfy α3 − α2 − 2α + 1 = 0. If α = x + x1 , then find

x64 − 2x52 + 3x43 + 2x38 − 2x29 + 5x17 + 5x10 − 7x7 + 7.

Problem 6.8 (2020 CMIMC Algebra #8). Compute the positive difference between the
two real solutions to the equation
√
(x − 1)(x − 4)(x − 2)(x − 8)(x − 5)(x − 7) + 48 3 = 0.

Problem 6.9. Let a, b, c, and d be complex numbers such that a + b + c + d = 1,

ab + bc + cd + da + ac + bd = 1, abc + acd + abd + bcd = 2, and abcd = 3. Find the value
of a6 + b6 + c6 + d6 .

Problem 6.10 (AoPS). Let a, b, c be the real roots of x3 − 4x2 − 32x + 17 = 0. Solve
for x in √ √ √
3 3
x − a + x − b + 3 x − c = 0.

Problem 6.11 (2010 HMMT Algebra #5). Suppose that x and y are complex numbers
such that x + y = 1 and x20 + y 20 = 20. Find the sum of all possible values of x2 + y 2 .

Problem 6.12 (2019 CMIMC Algebra #6). Let a,b, and c be the distinct solutions to
the equation x3 − 2x2 + 3x − 4 = 0. Find the value of
1 1 1
+ + .
a(b2 + c − a ) b(c + a − b ) c(a + b2 − c2 )
2 2 2 2 2 2

8
Adithya B., Brian L., William W., Daniel X. (4/29) Equations

Problem 6.13 (2010 HMMT Algebra #7). Let x, y, z, a, b, c be nonzero complex numbers
such that x + y + z = 67 and xy + yz + xz = 2010. Given that
b+c c+a a+b
= a, = b, = c,
x−3 y−3 z−3
Find the value of xyz.

Problem 6.14 (2013 HMMT Algebra #8). Let x, y be complex numbers such that
x2 +y 2 x4 +y 4 x6 +y 6
x+y = 4 and x3 +y 3 = 2. Find all possible values of x5 +y 5 .

Problem 6.15 (EGMO 2019/1). Find all triples (a, b, c) of real numbers such that
ab + bc + ca = 1 and

a2 b + c = b2 c + a = c2 a + b.

Problem 6.16 (Mexico 2011/3). Let n be a positive integer. Find all real solutions
(a1 , a2 , . . . , an ) to the system:

a21 + a1 − 1 = a2
a22 + a2 − 1 = a3
..
.
a2n + an − 1 = a1

Problem 6.17 (2006 Putnam B1). Show that the curve x3 + 3xy + y 3 = 1 contains only
one set of three distinct points, A, B, and C, which are vertices of an equilateral triangle,
and findits area.

Problem 6.18 (USAMO 2015/1). Solve in integers the equation

 3
x+y
x2 + xy + y 2 = +1 .
3