Polynomials

MOPSS
11 May 2025

Mathematics Olympiad
Problem Solving Sessions

MOPSS

Department of Mathematics
IISER Bhopal

[Link]

Suggested readings
• Evan Chen’s advice On reading solutions, available at [Link]
[Link]/2017/03/06/on-reading-solutions/.
• Evan Chen’s Advice for writing proofs/Remarks on English, available at
[Link]
• Notes on proofs by Evan Chen from OTIS Excerpts [Che25, Chapter 1].
• Tips for writing up solutions by Edward Barbeau, available at https:
//[Link]/barbeau/[Link].
• Evan Chen discusses why math olympiads are a valuable experience for
high schoolers in the post on Lessons from math olympiads, available at
[Link]
List of problems and examples
1.1 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.6 Example (Moscow MO 2015 Grade 9 P6) . . . . . . . . . . . 7
1.7 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.8 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.9 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.10 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.11 Example (Moscow MO 1946 Grades 7–8 P5) . . . . . . . . . 9
1.12 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.13 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.14 Example (Tournament of Towns, Spring 2014, Senior, A Level,
P7 by D. A. Zvonkin) . . . . . . . . . . . . . . . . . . . . . . 11
1.15 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.16 Example (USAMO 1975 P3) . . . . . . . . . . . . . . . . . . 12
1.17 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.18 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.19 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.20 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.21 Example (cf. Moscow MO 1953 Grade 10, India BMath 2006
P3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.22 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.23 Example (Putnam 1999 A2) . . . . . . . . . . . . . . . . . . 16
1.24 Example (India RMO 2017a P3) . . . . . . . . . . . . . . . . 17

§1 Polynomials
For further problems, we refer to [Goy21].

§1.1 Warm up
Example 1.1. Factorize the polynomial x8 + x4 + 1 into factors of at most
the second degree.

Summary — Expressing an expression as a difference of two squares yields a

factorization.

Solution 1. Note that

x8 + x4 + 1

2
1 Polynomials Typos may be reported to jpsaha@[Link].

= x8 + 2x4 + 1 − x4
= (x4 + 1)4 − (x2 )2
= (x4 − x2 + 1)(x4 + 1 + x2 )
= (x4 + 2x2 + 1 − 3x2 )(x4 + 2x2 + 1 − x2 )
√
= ((x2 + 1)2 − ( 3x)2 )((x2 + 1)2 − x2 )
√ √
= (x2 − 3x + 1)(x2 + 3x + 1)(x2 − x + 1)(x2 + x + 1).

Example 1.2. Show that

2a2 b2 + 2b2 c2 + 2c2 a2 − a4 − b4 − c4 = (a + b + c)(a + b − c)(b + c − a)(c + a − b).

Summary — Complete the squares.

Walkthrough —
(a) Try to see what would happen if we were allowed to change the signs!
(b) Change the signs and consider

a4 + b4 + c4 − 2a2 b2 − 2b2 c2 − 2c2 a2 .

This is “almost” (a2 − b2 − c2 )2 !! To be precise

(a2 − b2 − c2 )2 = a4 + b4 + c4 − 2a2 b2 + 2b2 c2 − 2c2 a2 .

(c) Let us continue with the above, and write 2a2 b2 +2b2 c2 +2c2 a2 −a4 −b4 −c4
in terms (a2 − b2 − c2 )2 as follows.

2a2 b2 + 2b2 c2 + 2c2 a2 − a4 − b4 − c4

= 4b2 c2 − (a4 + b4 + c4 − 2a2 b2 + 2b2 c2 − 2c2 a2 ).

(d) Does the above help?

Solution 2. Note that

2a2 b2 + 2b2 c2 + 2c2 a2 − a4 − b4 − c4

= 4b2 c2 − (a4 + b4 + c4 − 2a2 b2 + 2b2 c2 − 2c2 a2 )
= (2bc)2 − (a2 − b2 − c2 )2
= (2bc − (a2 − b2 − c2 ))(2bc + a2 − b2 − c2 )
= (2bc + b2 + c2 − a2 )(a2 − (b2 + c2 − 2bc))
= ((b + c)2 − a2 )(a2 − (b − c)2 )

Some style files, prepared by Evan Chen, have been adapted here. 3
11 May 2025 [Link]

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

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

Solution 3. Write1

a = y + z,
b = z + x,
c = x + y.

Note that

2a2 b2 + 2b2 c2 + 2c2 a2 − a4 − b4 − c4

= 2(y + z)2 (z + x)2 + 2(z + x)2 (x + y)2 + 2(x + y)2 (y + z)2
− (y + z)4 − (z + x)4 − (x + y)4
= 2(z 2 + 2yz + y 2 )(z 2 + 2zx + x2 )
+ 2(x2 + 2zx + z 2 )(x2 + 2xy + y 2 )
+ 2(y 2 + 2xy + x2 )(y 2 + 2yz + z 2 )
− (y + z)4 − (z + x)4 − (x + y)4
X
X
=2 x4 + x2 (y 2 + z 2 + 2x(y + z)) + yz(2x + y)(2x + z) − (x + y)4
cyc cyc
X
x4 + x2 y 2 + z 2 x2 + 2x3 (y + z) + 4x2 yz + 2xyz(y + z) + y 2 z 2


=2
cyc
X
− (x + y)4
cyc
X
= 2(x4 + y 4 + z 4 ) + 6(x2 y 2 + y 2 z 2 + z 2 x2 ) + 16xyz(x + y + z) + 4 x3 (y + z)
cyc
X
− (x + y)4
cyc
X
= 2(x4 + y 4 + z 4 ) + 6(x2 y 2 + y 2 z 2 + z 2 x2 ) + 16xyz(x + y + z) + 4 x3 (y + z)
cyc

1 It is good ask the following simple and innocent question: how can one write a, b, c as
stated above? Does it mean that given any three real numbers a, b, c, one can find real
numbers x, y, z such that a = y + z, b = z + x, c = x + y? A crucial point to note is that
one very often deals with indeterminates (aka variables) instead of real numbers. In the
above, a, b, c could be indeterminates instead of being real numbers! What do we do in
that case? Is it so that there are indeterminates x, y, z such that the six indeterminates
a, b, c, x, y, z satisfy a = y + z, b = z + x, c = x + y? As of now, let’s not worry about any
of these!. Just keep in mind the message that at times, we need to be quite careful about
what we do!

4 The content posted here and at this blog by Evan Chen are quite useful.
1 Polynomials Typos may be reported to jpsaha@[Link].

X
− (x4 + 4x3 y + 6x2 y 2 + 4xy 3 + y 4 )
cyc

= 16xyz(x + y + z)
= (a + b + c)(a + b − c)(b + c − a)(c + a − b).

Remark. Please go through the footnote next to the word Write from the above
solution. This footnote would explain that just writing Write a = y + z, b =
z + x, c = x + y requires more care! To address this issue, replace Write
a = y + z, b = z + x, c = x + y in the above solution by the following.
Consider the real numbers x, y, z defined by
1
x= (b + c − a),
2
1
y = (c + a − b),
2
1
z = (a + b − c).
2
Note that
a = y + z, b = z + x, c = x + y
holds.

Example 1.3. Find numbers a, b, c, d for which the equation

2x − 7 a b
= +
4x2 + 16x + 15 x+c x+d
would be an identity.

Walkthrough — Factorize the denominator into linear factors. Then express-

ing the numerator as a linear combination of those factors would provide such
an identity.

Solution 4. Note that

2x − 7 2x − 7
= .
4x2 + 16x + 15 (2x + 3)(2x + 5)

Hence, if 2x − 7 can be expressed as

p(2x + 3) + q(2x + 5),

2x−7
then 4x2 +16x+15 can be expressed as a sum of two fractions, each having a
constant in the numerator and a linear polynomial in the denominator.

Some style files, prepared by Evan Chen, have been adapted here. 5
11 May 2025 [Link]

One way to find if there are any such p, q, is to assume first that there are
such real numbers p and q such that

2x − 7 = p(2x + 3) + q(2x + 5)

holds2 . Substituting x = − 52 , we obtain −2p = −12, which gives p = 6. Next,

substituting x = − 32 , we obtain 2q = −10, which implies q = −5.
Note that

6(2x + 3) + (−5)(2x + 5) = 12x + 18 − 10x − 25 = 2x − 7

holds3 . Using it, we obtain

2x − 7 2x − 7
=
4x2 + 16x + 15 (2x + 3)(2x + 5)
6(2x + 3) + (−5)(2x + 5)
=
(2x + 3)(2x + 5)
6 5
= −
2x + 5 2x + 3
5
3 2
= − .
x + 52 x + 32

Hence, we may take

5 5 3
a = 3, b = − , c = , d = .
2 2 2
■

Exercise 1.4. Are there other choices for a, b, c, d for which the identify would
hold?
Example 1.5. Let n be a positive integer. Show that

xn − 1 = (x − 1)(xn−1 + xn−2 + · · · + x + 1).

Solution 5. Note that

(x − 1)(xn−1 + xn−2 + · · · + x + 1)
2 and try to see what conditions get imposed on p, q. It may happen that the conditions
that get imposed, may suggest that there are no such p, q. However, it may also happen
that we would be able to find out which p, q would work!
3 It should be noted that p, q were assumed to exist such that p(2x + 3) + q(2x + 5) = 2x − 7

holds. Under this hypothesis, we obtained p = 6, q = −5. At this point, we cannot

immediately conclude that 6(2x + 3) + (−5)(2x + 5) = 2x − 7 holds (unless we verify it),
because if we do so, then we would do it under the same hypothesis.
• Even then, what would go wrong with that?
• Can a hypothesis (possibly combined with some of its consequences) be a justifica-
tion for itself to hold? Think about this point.

6 The content posted here and at this blog by Evan Chen are quite useful.
1 Polynomials Typos may be reported to jpsaha@[Link].

= x(xn−1 + xn−2 + · · · + x + 1) − (xn−1 + xn−2 + · · · + x + 1)

= xn + xn−1 + · · · + x2 + x − (xn−1 + xn−2 + · · · + x + 1)
= xn − 1.
■
The exercise below relies on Example 1.5.
Example 1.6 (Moscow MO 2015 Grade 9 P6). Do there exist two polynomials
with integer coefficients such that each of them has a coefficient with absolute
value exceeding 2015, but no coefficient of their product has absolute value
exceeding 1?

Summary — Try to come up with enough polynomials g1 (x), g2 (x), g3 (x), . . .

and h1 (x), h2 (x), h3 (x), . . . such that each of the products g1 g2 g3 . . . and
h1 h2 h3 . . . have at least one coefficient which is large in absolute value, and all
the coefficients of the product (g1 g2 g3 . . . )(h1 h2 h3 . . . ) are at most 1 in absolute
value.

Walkthrough —
(a) Try to come up with a polynomial P (x) whose coefficients are at most
1 in absolute value, and it can be written as a product of enough fac-
tors (say f1 (x), f2 (x), . . . ) such that each of such factor fi (x) admits a
decomposition into the product of two polynomials gi (x) and hi (x).
(b) Can you make sure that the product of the gi ’s, and the product of the
hi ’s have to have at least one large coefficient?
(c) For instance, would taking g1 (x) = g2 (x) = g3 (x) = · · · = 1 − x work for
some suitable choice of h1 (x), h2 (x), . . . ?
(d) Does taking

h1 (x) = 1 + x,
h2 (x) = 1 + x + x2 ,
h3 (x) = 1 + x + x2 + x3 ,

etc. work?
(e) Note that the product of enough gi ’s would have a large coefficient
(namely, the coefficient of the second largest power of x). On the other
hand, the product of enough hi ’s would have a large coefficient (namely,
the coefficient of the power of x).
(f ) What can be said about the absolute value of the coefficients of the
product of these two products?

The above seems to work except that having a control on the coefficients of
the product (g1 g2 g3 . . . )(h1 h2 h3 . . . ) seems hard4 .
4 Is it because it fails?

Some style files, prepared by Evan Chen, have been adapted here. 7
11 May 2025 [Link]

Solution 6. Consider the polynomial

2016
P (x) = (1 − x)(1 − x2 )(1 − x4 )(1 − x8 ) · · · (1 − x2 ).

Since
1 + 2 + 22 + 23 + · · · + 2n−1 < 2n ,
it follows that the coefficients of P (x) are at most 1 in absolute value. Note
that
P (x) = Q(x)R(x)
holds where

Q(x) = (1 − x)2017 ,
2016
−1
R(x) = (1 + x)(1 + x + x2 + x3 ) · · · (1 + x + x2 + · · · + x2 ).

The coefficient of x2016 in Q(x) is equal to 2017, and the coefficient of x in

R(x) is equal to 2016. This completes the proof. ■

§1.2 Division algorithm

Example 1.7. Prove that the polynomial x44 + x33 + x22 + x11 + 1 is divisible
by the polynomial x4 + x3 + x2 + x + 1.

Solution 7. Note that

x44 + x33 + x22 + x11 + 1

= x40 · x4 + x30 · x3 + x20 · x2 + x10 · x + 1
= (x40 − 1)x4 + (x30 − 1)x3 + (x20 − 1)x2 + (x10 − 1)x
+ x4 + x3 + x2 + x + 1.

Hence, to prove that the polynomial x44 + x33 + x22 + x11 + 1 is divisible by
x4 + x3 + x2 + x + 1, it suffices to show that x4 + x3 + x2 + x + 1 divides the
polynomials
x40 − 1, x30 − 1, x20 − 1, x10 − 1.
Since

x5 − 1 = (x − 1)(x4 + x3 + x2 + x + 1),
x10 − 1 = (x5 − 1)(x5 + 1),

it follows that x4 + x3 + x2 + x + 1 divides x10 − 1. Moreover, the polynomial

x10−1 divides all of

x40 − 1, x30 − 1, x20 − 1, x10 − 1.

8 The content posted here and at this blog by Evan Chen are quite useful.
1 Polynomials Typos may be reported to jpsaha@[Link].

Hence, x4 + x3 + x2 + x + 1 divides the polynomials

x40 − 1, x30 − 1, x20 − 1, x10 − 1.

Exercise 1.8. Show that the polynomial x580 +x390 +x326 +x262 +x198 +x134 +1
is divisible by x6 + x5 + x4 + x3 + x2 + x + 1.
Example 1.9. Determine the remainder obtained upon dividing x100 by
x2 − 3x + 2.

Solution 8. Let q(x) (resp. r(x)) denote the quotient (resp. the remainder)
obtained upon dividing x100 by x2 −3x+2. Note that r(x) is a linear polynomial,
i.e. r(x) = ax + b for some real numbers a, b. Then we have

x100 = q(x)(x2 − 3x + 2) + r(x).

Substituting x = 1, it yields

1 = r(1) = a + b.

Similarly, substituting x = 2, it gives

2100 = r(2) = 2a + b.

This shows that

a = 2100 − 1, b = 1 − a = 2 − 2100 .
Hence, the remainder obtained upon dividing x100 by x2 − 3x + 2 is equal to

(2100 − 1)x + 2 − 2100 .

Example 1.10. Factorize the polynomials

x(y − z)3 + y(z − x)3 + z(x − y)3 , xy(x2 − y 2 ) + yz(y 2 − z 2 ) + zx(z 2 − x2 )

into products of linear polynomials.

§1.3 Even and odd polynomials

Example 1.11 (Moscow MO 1946 Grades 7–8 P5). Prove that after completing
the multiplication and collecting the terms

(1 − x + x2 − x3 + · · · − x99 + x100 )(1 + x + x2 + · · · + x99 + x100 )

has no monomials of odd degree.

Some style files, prepared by Evan Chen, have been adapted here. 9
11 May 2025 [Link]

Summary — What happens if x is replaced by −x?

Solution 9. Let P (x) denote the polynomial

(1 − x + x2 − x3 + · · · − x99 + x100 )(1 + x + x2 + · · · + x99 + x100 ).
Note that P (x) = P (−x). By the Claim below, it follows that P (x) has no
monomials of odd degree.

Claim — Let Q(x) be a polynomial satisfying Q(x) = Q(−x). Then

Q(x) has no monomials of odd degree.

Proof of the Claim. Note that

Q(x) + Q(−x) Q(x) − Q(−x)
Q(x) = +
2 2
Q(x)+Q(−x)
holds. Using Q(x) = Q(−x), it follows that Q(x) = 2 . Consequently,
Q(x) has no monomials of odd degree.
■

Remark. The above decomposition of Q(x) is a special case of general phe-

nomenaa .
a Can you think of a few? Which general phenomena is referred to?!

Remark. The above solution is more elegant, and less cumbersome. Moreover,
it also highlights the underlying reason, whereas the solution below obscures
the conceptual viewpoint.
Solution 10. One can multiply the polynomials to note that
1 − x + x2 − x3 + · · · − x99 + x100
= 1 − x + x2 (1 − x) + x4 (1 − x) + x6 (1 − x) + · · · + x98 (1 − x) + x100
= (1 − x)(1 + x2 + x4 + x6 + · · · + x98 ) + x100 .

Using this, we obtain

(1 − x + x2 − x3 + · · · − x99 + x100 )(1 + x + x2 + · · · + x99 + x100 )

= (1 − x)(1 + x2 + x4 + x6 + · · · + x98 ) + x100 (1 + x + x2 + · · · + x99 + x100 )

= (1 − x)(1 + x2 + x4 + x6 + · · · + x98 )(1 + x + x2 + · · · + x99 + x100 )

+ x100 (1 + x + x2 + · · · + x99 + x100 )
= (1 + x2 + x4 + x6 + · · · + x98 )(1 − x101 )
+ x100 (1 + x + x2 + · · · + x99 + x100 )

10 The content posted here and at this blog by Evan Chen are quite useful.
1 Polynomials Typos may be reported to jpsaha@[Link].

= (1 + x2 + x4 + x6 + · · · + x98 )(1 − x101 )

+ x100 (x + x3 + x5 + · · · + x99 )
+ x100 (1 + x2 + x4 + · · · + x98 + x100 )
= (1 + x2 + x4 + x6 + · · · + x98 )(1 − x101 )
+ x101 (1 + x2 + x4 + x6 + · · · + x98 )
+ x100 (1 + x2 + x4 + · · · + x98 + x100 )
= 1 + x2 + x4 + x6 + · · · + x98 + x100 (1 + x2 + x4 + · · · + x98 + x100 ),

which has no monomial of odd degree. ■

The following exercise is quite similar to the Claim proved in the solution to
Example 1.11.
Exercise 1.12. Let Q(x) be a polynomial satisfying Q(x) = −Q(−x). Then
Q(x) has no monomials of even degree.
Example 1.13. Let n be an even positive integer, and let p(x) be a polynomial
of degree n such that p(k) = p(−k) for k = 1, 2, . . . , n. Prove that there is a
polynomial q(x) such that p(x) = q(x2 ).

Walkthrough — Note that the polynomial p(x) − p(−x) has degree < n
because n is even. Observe that it has at least n roots.

Remark. What would happen if n is not assumed to be even?

Example 1.14 (Tournament of Towns, Spring 2014, Senior, A Level, P7 by D.

A. Zvonkin). Consider a polynomial P (x) such that
P (0) = 1, (P (x))2 = 1 + x + x100 Q(x),
where Q(x) is also a polynomial. Prove that in the polynomial (P (x) + 1)100 ,
the coefficient of x99 is zero.

Solution 11. Note that

(P (x) + 1)100 + (1 − P (x))100
is a polynomial in P (x)2 of degree 50. Given three polynomials f (x), g(x), h(x)
having complex coefficients, with h(x) ̸= 0, we say that f (x) is congruent
to g(x) modulo h(x) if h(x) divides f (x) − g(x), that is, f (x) − g(x) is the
product of h(x) and a polynomial in x with complex coefficients. Since P (x)2
is congruent to 1 + x modulo x100 , it follows that (P (x) + 1)100 + (1 − P (x))100
is congruent to a polynomial of degree 50 in 1 + x modulo x100 . Using that
P (x) ≡ 1 mod x, we obtain that (P (x) + 1)100 is congruent to a polynomial
of degree 50 in 1 + x modulo x100 . This shows that the coefficient of x99 in
(P (x) + 1)100 is zero. ■

Some style files, prepared by Evan Chen, have been adapted here. 11
11 May 2025 [Link]

§1.4 Factorization and roots

Example 1.15. Let a, b, c be three distinct real numbers. Show that
(a − x)(b − x) (b − x)(c − x) (c − x)(a − x)
+ + = 1.
(a − c)(b − c) (b − a)(c − a) (c − b)(a − b)

Walkthrough — Can a polynomial having degree at most two admit more

than two distinct roots?

Example 1.16 (USAMO 1975 P3). [GA17, Problem 151] A polynomial P (x)
of degree n satisfies
k
P (k) = for k = 0, 1, 2, . . . , n.
k+1
Find P (n + 1).

Solution 12. Note that xP (x + 1) − x is a polynomial of degree n + 1, and it

vanishes at the n + 1 integers 0, 1, 2, . . . , n. It follows that

(x + 1)P (x) − x = cx(x − 1)(x − 2) . . . (x − n)

for some nonzero real number c. Substituting x = −1 yields

1 = (−1)n+1 c(n + 1)!,

(−1)n+1
which gives c = (n+1)! . This implies that

(n + 2)P (n + 1) = n + 1 + (−1)n+1 ,

and consequently,
n + 1 + (−1)n+1
P (n + 1) = .
n+2
■

Example 1.17. Let P (x) be a polynomial of degree ≤ n having rational

coefficients. Suppose P (k) = k1 holds for 1 ≤ k ≤ n + 1. Determine P (0).

Solution 13. Note that xP (x) − 1 is a polynomial of degree at most n + 1,

and it vanishes at 1, 2, . . . , n + 1. This implies that

xP (x) − 1 = c(x − 1)(x − 2) . . . (x − n − 1)

holds for some rational number c. Substituting x = 0, we obtain

(−1)n
c= .
(n + 1)!

12 The content posted here and at this blog by Evan Chen are quite useful.
1 Polynomials Typos may be reported to jpsaha@[Link].

Differentiating and dividing by xP (x) − 1, we obtain

xP ′ (x) + P (x) 1 1 1
= + + ··· + ,
xP (x) − 1 x−1 x−2 x−n−1
which yields
1 1 1
P (0) = 1 + + + ··· + .
2 3 n
■

Example 1.18. Let g(x) and h(x) be polynomials with real coefficients such
that
g(x)(x2 − 3x + 2) = h(x)(x2 + 3x + 2)
and f (x) = g(x)h(x) + (x4 − 5x2 + 4). Prove that f (x) has at least four real
roots.

Solution 14. Note that g(x) and h(x) satisfy

g(x)(x − 1)(x − 2) = h(x)(x + 1)(x + 2),
which shows that
g(−1), g(−2), h(1), h(2)
are equal to 0. Also note that
x4 − 5x2 + 4 = (x2 − 1)(x2 − 4)
= (x − 1)(x + 1)(x + 2)(x − 2).
Hence, the polynomials g(x)h(x) and x4 − 5x2 + 4 vanish at 1, −1, 2, −2.
Consequently, f also vanishes at these four points. ■

Example 1.19. Let P (x) be a polynomial with real coefficients such that
P (sin α) = P (cos α) for all α ∈ R. Show that P (x) = Q(x2 − x4 ) for some
polynomial Q(x) with real coefficients.

Walkthrough —
(a) Show that P (x) = P (−x) for any −1 ≤ x ≤ 1, and hence P (x) = f (x2 ).
(b) Deduce that f (x) = f (1 − x) for any 0 ≤ x ≤ 1.
(c) Using induction or otherwise, prove that f (x) = g(x − x2 ) for some
polynomial g(x) with real coefficients.

Example 1.20. [WH96, Problem 27] Let p1 , . . . , pn denote n ≥ 1 distinct

integers. Show that the polynomial
(x − p1 )2 (x − p2 )2 · · · (x − pn )2 + 1
cannot be expressed as the product of two non-constant polynomials with
integral coefficients.

Some style files, prepared by Evan Chen, have been adapted here. 13
11 May 2025 [Link]

Solution 15. On the contrary, let us assume that the polynomial

P (x) := (x − p1 )2 (x − p2 )2 · · · (x − pn )2 + 1

can be expressed as the product of two non-constant polynomials f (x), g(x)

with integral coefficients.
Let us first establish the following Claims.

Claim — Replacing f, g by −f, −g respectively (if necessary), we may

assume that f, g take positive values at all real arguments.

Proof of the Claim. Note that the polynomial P (x)−1 vanishes at x = p1 , . . . , pn .

Since the product of the leading coefficients of f (x) and g(x) is equal to the
leading coefficient of P (x), we may replace f (x), g(x) by −f (x), −g(x) respec-
tively (if necessary) to assume that the leading coefficients of f (x), g(x) are
positive. Since P = f g and P does not have a real root, it follows that the
polynomials f, g do not have any real roots. At large enough real arguments,
the polynomials f, g take positive values. Since f, g have no real roots, we
conclude that they take positive values at all real arguments.

Claim — The polynomials f, g are of degree n. Moreover, these polyno-

mials are equal.

Proof of the Claim. On the contrary, let us asssume that the degrees of f, g
are not equal. Interchanging f, g if necessary, we assume that deg(f ) < deg(g).
Since the sum of the degrees of f, g is equal to 2n, it follows that deg(f ) < n.
For any 1 ≤ i ≤ n, the integers f (pi ), g(pi ) are equal to 1 or −1. Since
f, g take positive values at all real arguments, we obtain f (pi ) = 1 for any
1 ≤ i ≤ n. This shows that the polynomial f − 1 has at least n distinct roots.
Using deg(f ) < n, we conclude that f − 1 is the zero polynomial, which is
impossible since f is a non-constant polynomial. Therefore, the hyothesis that
the degrees of f, g are not equal is not tenable. This completes the proof of
the first part of the Claim.
Note that f, g are polynomials of degree n with equal leading coefficients.
This shows that the polynomial f (x) − g(x) has degree less than n and it
vanishes at the n distinct points p1 , . . . , pn . It follows that f = g.

Using the above Claim, note that

2
f (x)2 − ((x − p1 )(x − p2 ) · · · (x − pn )) = 1,

or equivalently,



f (x) + (x − p1 )(x − p2 ) · · · (x − pn ) f (x) − (x − p1 )(x − p2 ) · · · (x − pn )

14 The content posted here and at this blog by Evan Chen are quite useful.
1 Polynomials Typos may be reported to jpsaha@[Link].

= 1,

which implies that the polynomials

f (x) + (x − p1 )(x − p2 ) · · · (x − pn ), f (x) − (x − p1 )(x − p2 ) · · · (x − pn )

are constant polynomials, and both of them are equal. Consequently, the
polynomial (x−p1 )(x−p2 ) · · · (x−pn ) is the zero polynomial, which is impossible.
This shows that the hypothesis that the given polynomial can be expressed as
the product of two non-constant polynomials with integral coefficients is not
tenable. This completes the proof. ■

Example 1.21 (cf. Moscow MO 1953 Grade 10, India BMath 2006 P3).
[AE11, §1.7] Let n be a positive integer. Find the roots of the polynomial

X X(X + 1) X(X + 1) · · · (X + n − 1)
Pn (X) = 1 + + + ··· + .
1! 2! n!

Walkthrough — Note that P1 (X) = X + 1 has −1 as its root, P2 (X) =

1
2
(X + 1)(X + 2) has −1, −2 as its roots. Check that

1
P3 (X) = (X + 1)(X + 2)(X + 3).
3!
What happens for general n?

Solution 16. We claim that the roots of Pn (X) are −1, −2, . . . , −n for any
integer n ≥ 1. Note that the claim holds for n = 1. Suppose the claim holds
for some integer n. Comparing leading coefficients, it follows that
1
Pn (X) = (X + 1)(X + 2) · · · (X + n).
n!
Observe that
X(X + 1) · · · (X + n)
Pn+1 (X) = Pn (X) +
(n + 1)!
1 X(X + 1) · · · (X + n)
= (X + 1)(X + 2) · · · (X + n) +
n! (n + 1)!
1
= (X + 1)(X + 2) · · · (X + n)(n + 1 + X).
(n + 1)!

Hence the roots of Pn+1 are −1, −2, . . . , −(n + 1). The claim follows by
induction. ■

Example 1.22. Show that any odd degree polynomial with real coefficients
has at least one real root.

Some style files, prepared by Evan Chen, have been adapted here. 15
11 May 2025 [Link]

Example 1.23 (Putnam 1999 A2). Show that for some fixed positive integer
n, we can always express a polynomial with real coefficients which is nowhere
negative as a sum of the squares of n polynomials.

Walkthrough —
(a) Show that the real roots of P have even multiplicity.
(b) Conclude that P can be expressed as a product of monic quadratic
polynomials with real coefficients having nonreal roots, and even powers
of linear polynomials with real coefficients.
(c) Show that a monic quadratic polynomial with real coefficients having
nonreal roots is the sum of the squares of two polynomials with real
coefficients.

Solution 17. Note that if P is a constant polynomial, then it is clear.

Henceforth, let us assume that P is a nonconstant polynomial.

Claim — The polynomial P can be written as the product of polynomials,

each of which can be expressed as the sum of the squares of two polynomials
with real coefficients.

Proof of the Claim. Since P has real coefficients, it follows that if α ∈ C \ R is

a root of P , then so is α. Thus, the nonreal complex roots of P form pairs of
complex conjugates. Note that

(x − α)(x − α) = (x − Re(α))2 + Im(α)2 .

Decomposing P over the pairs of nonreal complex conjugate roots, and the
real roots, it follows that P can be expressed as the product
Y
cf (x) (x − a)ma ,
a∈A

where c denote the leading coefficient of P , f (x) denotes the product of (possibly
no) quadratic polynomials of the form (x − a)2 + b2 with a ∈ R, b ∈ R \ {0}, and
A denotes the set of real roots of P , and for an element a ∈ A, the multiplicity
of a is denoted by ma . P
Evaluating P at a suitable real number (for instance, at 1 + a∈A a (resp.
1) if A is nonempty (resp. empty)), it follows that c > 0.
Let a be an element of A. Since A is finite, there exists a real number ε > 0
such that the interval (a − ε, a + ε) contains no real roots of P other than a.
If ma were odd, then the sign of P (x) would not remain constant as x ranges
over in (a − ε, a + ε) \ {a}. Hence, it follows that ma is even.
Since c > 0 amd ma is even for any a ∈ A, the Claim follows.

16 The content posted here and at this blog by Evan Chen are quite useful.
References Typos may be reported to jpsaha@[Link].

Claim — Let f1 (x), g1 (x), f2 (x), g2 (x) be polynomials with real coeffi-
cients. Then the following holds.

(f1 (x)2 + g1 (x)2 )(f2 (x)2 + g2 (x)2 )

= (f1 (x)f2 (x) − g1 (x)g2 (x))2 + (f1 (x)g2 (x) − f2 (x)g1 (x))2

Proof of the Claim. Note that

(f1 (x)2 + g1 (x)2 )(f2 (x)2 + g2 (x)2 )

= f1 (x)2 f2 (x)2 + g1 (x)2 g2 (x)2 − 2f1 (x)f2 (x)g1 (x)g2 (x)
+ f1 (x)2 g2 (x)2 + f2 (x)2 g1 (x)2 + 2f1 (x)g2 (x)f2 (x)g1 (x)
= (f1 (x)f2 (x) − g1 (x)g2 (x))2 + (f1 (x)g2 (x) + f2 (x)g1 (x))2 .

Combining the above Claims, and using induction, the result follows. ■

x
Example 1.24 (India RMO 2017a P3). Let P (x) = x2 + +b and Q(x) = x2 +
2
cx + d be two polynomials with real coefficients such that P (x)Q(x) = Q(P (x))
for all real x. Find all real roots of P (Q(x)) = 0.

Solution 18. Let α be a root of P (x) in C. Substituting x = α in P (x)Q(x) =

Q(P (x)), we obtain Q(0) = 0, which shows that d = 0. This gives

P (x)(x2 + cx) = P (x)(P (x) + c),

which yields P (x) + c = x2 + cx. It follows that c = 12 , b = − 12 .

Note that the roots of P (x) are −1, 12 . Hence, any root β of P (Q(x)) = 0
satisfies Q(β) = −1 or Q(β) = 12 . Since the discrimint of Q(x) + 1 is negative,
it follows that if β is a real root of P (Q(x)) = 0, then Q(β) = 12 , which gives
β = −1 or β = 1/2.
Note that P ( 12 ) = 0 and −1, 12 are the roots of Q(x) = 1/2. This shows that
the real roots of P (Q(x)) = 0 are −1, 12 . ■

References
[AE11] Titu Andreescu and Bogdan Enescu. Mathematical Olympiad
treasures. Second. Birkhäuser/Springer, New York, 2011, pp. viii+253.
isbn: 978-0-8176-8252-1; 978-0-8176-8253-8 (cited p. 15)
[Che25] Evan Chen. The OTIS Excerpts. Available at https : / / web .
[Link]/[Link]. 2025, pp. vi+289 (cited p. 1)

Some style files, prepared by Evan Chen, have been adapted here. 17
11 May 2025 [Link]

[GA17] Răzvan Gelca and Titu Andreescu. Putnam and beyond. Second.
Springer, Cham, 2017, pp. xviii+850. isbn: 978-3-319-58986-2; 978-
3-319-58988-6. doi: 10.1007/978- 3- 319- 58988- 6. url: https:
//[Link]/10.1007/978-3-319-58988-6 (cited p. 12)
[Goy21] Rohan Goyal. “Polynomials”. Available at [Link]
com / s / yo31nat6z5ggaue / Polynomials . pdf ? dl = 0. 2021 (cited
p. 2)
[WH96] Kenneth S. Williams and Kenneth Hardy. The Red Book of
mathematical problems. Corrected reprint of the ıt The Red Book:
100 practice problems for undergraduate mathematics competitions
[Integer Press, Ottawa, ON, 1988]. Dover Publications, Inc., Mineola,
NY, 1996, pp. x+174. isbn: 0-486-69415-1 (cited p. 13)

18 The content posted here and at this blog by Evan Chen are quite useful.