0% found this document useful (0 votes)
8 views

Polynomials s Li Dec22

The document is a lecture on polynomial problems, covering various topics such as curve fitting, Descartes' rule of signs, cyclotomic polynomials, and irreducibility. It includes theorems like Lagrange interpolation and Gauss' lemma, along with proofs and propositions related to polynomial properties. Additionally, it presents several problems and warm-ups for further exploration of polynomial concepts.

Uploaded by

jonahlauje
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
8 views

Polynomials s Li Dec22

The document is a lecture on polynomial problems, covering various topics such as curve fitting, Descartes' rule of signs, cyclotomic polynomials, and irreducibility. It includes theorems like Lagrange interpolation and Gauss' lemma, along with proofs and propositions related to polynomial properties. Additionally, it presents several problems and warm-ups for further exploration of polynomial concepts.

Uploaded by

jonahlauje
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Polynomials (S)

Charles Li

9th December 2022

Polynomial problems often take the form, “here’s some problem about a function f . blah blah blah. oh and by
the way, this only works if f is a polynomial haha. so maybe assume that?”. The focus of today’s lecture is
what the ‘oh and by the way’ tells you. I guess the usual tactics for dealing with generic functions could still
be helpful.

1 Revision/Outline
You’ll already be familiar with the following tricks/results:

• Degree arguments and equal too often • Polynomial division, remainder theorem
• Roots and the fundamental theorem of algebra • Factorisations of xn ± y n
• Intermediate value theorem • a − b | P (a) − P (b) for integer polynomials

• Viete’s formulas and symmetric polynomials • Looking modulo a polynomial

Instead of boring you and covering that again1 , today I want to look at:

• Curve fitting and Lagrange interpolation • Reducibility


• Descarte’s rule of signs – Gauss’ lemma
n
• Factorisations of x − 1 and cyclotomics – Upstairs-downstairs and Eisenstein

Angelo’s syllabus says I should also remind you of the following:

• Root focus – “Focusing on the properties, e.g. • Double, triple roots


size, of a real or complex root is often helpful.”
This is probably most useful when dealing with • Iteration
a polynomial functional equation. • Minimal polynomials2

2 Curve fitting
You know that any n distinct roots uniquely determine a degree n polynomial, up to scaling. You can think of
the ‘up to scaling’ as fixing some other point (c, f (c)) you want f to pass through, where f (c) 6= 0. So basically,
fixing n + 1 points of a particular form gives you a unique polynomial.
What about if I take n + 1 arbitrary points? You still can get a unique polynomial, but it will have degree ≤ n.
Theorem (Lagrange interpolation). Fix x0 , x1 , . . . , xn distinct reals, and let y0 , y1 , . . . , yn ∈ R. Then there is
a unique polynomial P of degree at most n passing through each (xi , yi ).
1 Please
ask if you’re not familiar with one of them!
2 For
every root α of a monic integer polynomial, there’s a unique monic integer polynomial mα (x) of minimal degree with α as
a root. This polynomial divides every other polynomial with α as a root.

1
Proof. Define
n
Y x − xi
P̂j (x) =
x − xi
i=0 j
i6=j

for each 0 ≤ j ≤ n. Then P̂j (x) is 0 at each x = xi with i 6= j, and is 1 at x = xj . So


n+1
X
P (x) = yj P̂j (x)
j=1

works. Moreover, it’s the unique such polynomial: If P̃ also works, P − P̃ has degree ≤ n but has n+1 roots.

Huh?? What was that? It was a mathematical proof, of the variety you’ll find at uni: slick and polished, with
all traces of intuition removed. Optimized for rote memorisation. Ugh.
An easier way to remember how to construct such polynomials is to follow the following steps – at each step
find a polynomial P satisfying:

• P (x0 ) 6= 0, P (x1 ) = P (x2 ) = P (x3 ) = · · · = P (xn ) = 0.


• P (x0 ) = 1, P (x1 ) = P (x2 ) = P (x3 ) = · · · = P (xn ) = 0.
• P (x0 ) = y0 , P (x1 ) = P (x2 ) = P (x3 ) = · · · = P (xn ) = 0.
• P (x0 ) = 0, P (x1 ) = y1 , P (x2 ) = P (x3 ) = · · · = P (xn ) = 0.
• P (xk ) = yk , P (x0 ) = P (x1 ) = · · · = P (xk−1 ) = P (xk+1 ) = · · · = P (xn ) = 0.
• P (x0 ) = y0 , P (x1 ) = y1 , . . . , P (xn ) = yn .

Still, it’s a massive pain. An alternate way to construct such polynomials bit by bit is via divided differences
and Newton’s interpolation formula:
Let p[x0 , x1 , . . . , xn ] be the leading coefficient of the Lagrange polynomial interpolating (x0 , y0 ), . . . , (xn , yn ).
Then p[x0 ] = y0 and p[x1 ] = xy11 −y −x0 .
0

Proposition. If x0 , x1 , . . . , xn , xn+1 are pairwise distinct, then


p[x1 , . . . , xn+1 ] − p[x0 , . . . , xn ]
p[x0 , . . . , xn+1 ] =
xn+1 − x0

Proof. Consider the polynomials q, r interpolating (x0 , y0 ), . . . , (xn , yn ) and (x1 , y1 ), . . . , (xn+1 , yn+1 ), and note
that
(x − x0 )r(x) − (x − xn+1 )q(x)
p(x) =
xn+1 − x0
interpolates (x0 , y0 ), . . . , (xn+1 , yn+1 ).

This makes calculating the p[x0 , . . . , xn ] relatively simple.


Define
n−1
Y
pn (x) = p[x0 ] + p[x0 , x1 ](x − x0 ) + · · · + p[x0 , x1 , . . . , xn ] (x − xi ).
i=0

Theorem. pn (x) satisfies pn (xi ) = yi for i = 0, 1, . . . , n.

Proof. Induct.

3 Descartes’ Rule of Signs


Here’s a niche result on the coefficients of a polynomial that has gotten surprising usage in recent years’ T-sets:
Proposition (Descartes’ Rule of Signs). Let f (x) = an xn + an−1 + xn−1 + · · · + a1 x + a0 be a real polynomial.
Let s be the number of sign changes in the sequence a0 , . . . , an . Then the number of positive real roots of f is
at most s, and also has the same parity as s.

The proof is a case-bashy induction on n (divide out by a positive root and see what happens). A corresponding
result exists for the number of negative real roots by replacing x with −x.

2
4 Cyclotomic polynomials
The cyclotomic polynomials are monic integer polynomials Φn (x), for n = 1, 2, . . . , defined by
Y  k

Φn (x) = x − e2iπ n .
1≤k≤n
gcd(k,n)=1

The definition there is not particularly helpful. Here’s a bunch of examples:


k−1
• Φp (x) = xp−1 + · · · + x + 1 for • Φ2k (x) = x2 + 1 for k ≥ 1 • Φ6 (x) = x2 − x + 1
p prime
• Φ1 (x) = x − 1 • Φ9 (x) = x6 + x3 + 1

Importantly, cyclotomics are irreducible over Z (see next section for what that means!) and satisfy
Y
Φd (x) = xn − 1.
d|n

Other properties:

• If n is odd then Φ2n (x) = Φn (−x).


• If p is a prime dividing n then Φnp (x) = Φn (X p ).
• If p and q are distinct primes then the nonzero coefficients of Φpq are alternately +1 and −1.
• If n is not divisible by at least three distinct odd primes then the coefficients of Φn are −1, 0 or 1.

There exist cyclotomics with coefficients other than −1, 0, 1 such as Φ105 .

5 Reducibility
An integer polynomial P is irreducible over Z (or Q or some other ring like Z/pZ = Zp !) if you can’t factorise
it non-trivially over Z – that is, whenever P = QR where Q and R are integer polynomials, either Q or R is
constant.
A useful trick is that if f (x) = g(x + 1), then f is irreducible if and only g is irreducible.

5.1 Gauss’ Lemma


Theorem (Gauss’ Lemma). If f is a polynomial that’s irreducible over Z, it’s irreducible over Q.

We won’t prove this theorem – we’ll leave it for when you learn about commutative rings and fraction fields at
uni. However, note that the converse is true almost by definition. So being irreducible over Z is the same thing
as being irreducible over Q.

5.2 Upstairs-downstairs and Eisenstein


Here’s an idea: If you can factorise P nontrivially over Z, then you certainly can mod n. Or, to rephrase it:

Proposition (Upstairs-downstairs). Let P be an integer polynomial. Let P̃ be the reduction of P mod n (so P
is the same polynomial but with coefficients reduced mod n) . If P̃ is irreducible mod n, then P is irreducible
over Z.

The ‘upstairs-downstairs’ name comes from working ‘downstairs’ mod n giving you a result about P ‘upstairs’
(over Z). Sometimes you’ll have to go back and forth upstairs and back downstairs.
Lemma (Eisenstein’s criterion). Let P (x) = an xn + · · · + a1 x1 + a0 be an integer polynomial. Let p be a prime
such that p | a0 , a1 , . . . , an−1 , p - an , p2 - a0 . Then P is irreducible over Z.

3
6 Warmups
1. Ross and Hadyn play the following game. Ross thinks of an integer polynomial p with non-negative
coefficients. Hadyn thinks of an integer a, and Ross tells him p(a). Then Hadyn thinks of another integer
b, and Ross tells him p(b). If Hadyn correctly guesses Ross’ polynomial, Ross agrees to play Keep Talking
and Nobody Explodes with the seniors in the evening. Hadyn thinks this would be hilarious. Can he devise
a winning strategy?

2. Find all polynomials P with real coefficients such that P ((x + 1)3 ) = (P (x) + 1)3 and P (0) = 0.
3. The polynomial P has integer coefficients and P (x) = 5 for five different integers x. Show that there is
no integer x such that −6 ≤ P (x) ≤ 4 or 6 ≤ P (x) ≤ 16.
4. (USAMO 1974) Let a, b, c be three distinct integers, and let P be a polynomial with integer coefficients.
Show that P (a) = b, P (b) = c, P (c) = a cannot be satisfied simultaneously.
5. (Vietnam MO 1986) The polynomial p(x) has degree n and p(1) = 2, p(2) = 4, p(3) = 8, . . . , p(n + 1) =
2n+1 . Find p(n + 2).
6. Suppose f is a non-zero polynomial with exactly r non-zero real roots, counting multiplicity. Show that
f has at least 2r + 1 non-zero coefficients.
7. Show that Φp (x) = xp−1 + · · · + x + 1 is irreducible over Q. (No using facts about cyclotomic polys!)
8. Find all polynomials P (x) that satisfy P (x2 ) = P (x)2 .
9. If a1 , a2 , a3 , . . . , an are distinct integers, prove that the polynomial P (x) = (x − a1 )(x − a2 ) . . . (x − an ) − 1
is irreducible.
10. (A useful lemma!) Let n > 0 and p(x) a polynomial of degree d < n. Show that
n  
i n
X
(−1) p(i) = 0.
i=0
i

7 Problems
k
11. (USAMO 1975) If P (x) denotes a polynomial of degree n such that P (k) = k+1 for k = 0, 1, 2, . . . , n,
determine P (n + 1).
12. (ISL 1997) Let p be a prime number and f an integer polynomial of degree d such that f (0) = 0, f (1) = 1
and f (n) is congruent to 0 or 1 modulo p for every integer n. Prove that d ≥ p − 1.

13. A function f : R2 → R is such that fixing any coordinate gives a polynomial – that is, for each a ∈ R,
the functions ga (x) := f (x, a) and ha (y) := f (a, y) are polynomials. Is it necessarily the case that f is a
2-variable polynomial?
14. (BaMO SL 2020) Let P (x), Q(x) be distinct polynomials of degree 2020 with non-zero coefficients. Suppose
that they have r common real roots counting multiplicity and s common coefficients. Find the maximum
value of r + s.
15. Determine all polynomials p(x) with real coefficients such that

p(x)p(x + 1) = p(x2 )

for all real x.


16. (MUMC 2008 Senior) Let f (x) be a polynomial of positive degree with all of its coefficients being positive
integers. Let m ≥ 1. Prove that f (f (m) + 1) is divisible by f (m) if and only if m = 1.
17. (IMO 1993 Q1) Let n > 1 be an integer and let f (x) = xn + 5xn−1 + 3. Show that f is irreducible over Z.

18. Suppose polynomials p,q satisfy p(q(x)) = q(p(x)) for all x. Prove that if the equation p(x) = q(x) has no
real roots, then neither will p(p(x)) = q(q(x)).

4
19. (ISL 2017 A1) Let a1 , a2 , . . . an , k, and M be positive integers such that
1 1 1
+ + ··· + =k and a1 a2 · · · an = M.
a1 a2 an
If M > 1, prove that the polynomial

P (x) = M (x + 1)k − (x + a1 )(x + a2 ) · · · (x + an )

has no positive roots.


20. If q is a rational number and cos qπ is also rational, show that cos qπ ∈ {0, ± 21 , ±1}.
21. (USAMO 1997) Show that for each integer n there exists a unique integer polynomial Q with coefficients
in {0, 1, 2 . . . , 9} such that Q(−2) = Q(−5) = n.
22. Let P and Q be monic polynomials with real coefficients such that P (P (x)) = Q(Q(x)). Prove that
P (x) = Q(x).
23. (IMO 2016 Q5) The equation

(x − 1)(x − 2) . . . (x − 2016) = (x − 1)(x − 2) . . . (x − 2016)

is written on the board, with 2016 linear factors on each side. What is the least possible value of k for
which it is possible to erase exactly k of these 4032 linear factors so that at least one factor remains on
each side and the resulting equation has no real solutions?

24. (ISL 2019 A5) Let x1 , x2 , . . . , xn be different real numbers. Prove that
X Y 1 − xi xj 
0, if n is even;
=
xi − xj 1, if n is odd.
16i6n j6=i

25. Let P (x) be a given non-constant polynomial with complex coefficients, and let Q(x, y) = P (x) − P (y).
Assume that Q(x, y) has exactly k linear factors. Let R(x, y) be a nonconstant factor of Q(x, y) whose
degree is less than k. Prove that R(x, y) is a product of linear polynomials with complex coefficients.

26. (ISL 2019 A6) A polynomial P (x, y, z) in three variables with real coefficients satisfies the identities

P (x, y, z) = P (x, y, xy − z) = P (x, zx − y, z) = P (yz − x, y, z).


Prove that there exists a polynomial F (t) in one variable such that

P (x, y, z) = F (x2 + y 2 + z 2 − xyz).

27. (IMO 2007 Q6) Let n be a positive integer. Consider

S = {(x, y, z) : x, y, z ∈ {0, 1, 2, . . . , n}, x + y + z > 0}

as a set of (n + 1)3 − 1 points in three-dimensional space.


Determine the smallest number of planes, the union of which contains S but does not include (0, 0, 0).
28. (ISL 2018 A6) Let m, n ≥ 2 be integers. Let f (x1 , . . . , xn ) be a polynomial with real coefficients such that
 
x1 + · · · + xn
f (x1 , . . . , xn ) = for every x1 , . . . , xn ∈ {0, 1, . . . , m − 1}.
m

Prove that the total degree of f is at least n.

You might also like