๐Ÿงฎ Brain Teaser

Polynomial with No Integer Roots

Why it's beautiful: The statement sounds like it needs heavy machinery โ€” but the entire proof fits in two lines and uses nothing but parity (odd/even). The trick is so simple it's surprising: you split into cases based on whether the root is odd or even, and in each case evaluate pp at a known point to get a contradiction. No computation, no degree arguments, just parity.


Problem

Let p(x)p(x) be a polynomial with integer coefficients. Suppose p(0)p(0) and p(1)p(1) are both odd.

Prove that p(x)p(x) has no integer roots.


Hint: If nn is an integer root, consider whether nn is even or odd. In each case, try to relate p(n)p(n) to either p(0)p(0) or p(1)p(1) using a factoring trick.

PolynomialsParityInteger rootsModular arithmetic

Polynomial with No Integer Roots โ€” Answer


Key Fact: The Factoring Trick

For any polynomial p(x)p(x) with integer coefficients and any integers a,ba, b:

aโˆ’bโˆฃp(a)โˆ’p(b)a - b \mid p(a) - p(b)

Why? Because aโˆ’ba - b divides akโˆ’bka^k - b^k for every positive integer kk, so it divides every term of p(a)โˆ’p(b)p(a) - p(b).


Proof

Suppose for contradiction that nn is an integer root, so p(n)=0p(n) = 0.

Case 1: nn is even.

Apply the factoring trick with a=na = n, b=0b = 0:

nโˆ’0โˆฃp(n)โˆ’p(0)โ€…โ€ŠโŸนโ€…โ€Šnโˆฃ0โˆ’p(0)โ€…โ€ŠโŸนโ€…โ€Šnโˆฃp(0)n - 0 \mid p(n) - p(0) \implies n \mid 0 - p(0) \implies n \mid p(0)

Since nn is even, p(0)p(0) must be even. But p(0)p(0) is odd โ€” contradiction.

Case 2: nn is odd.

Apply the factoring trick with a=na = n, b=1b = 1:

nโˆ’1โˆฃp(n)โˆ’p(1)โ€…โ€ŠโŸนโ€…โ€Š(nโˆ’1)โˆฃ0โˆ’p(1)โ€…โ€ŠโŸนโ€…โ€Š(nโˆ’1)โˆฃp(1)n - 1 \mid p(n) - p(1) \implies (n-1) \mid 0 - p(1) \implies (n-1) \mid p(1)

Since nn is odd, nโˆ’1n - 1 is even, so p(1)p(1) must be even. But p(1)p(1) is odd โ€” contradiction.

In both cases we reach a contradiction, so pp has no integer roots. โ–ก\square


Why it works

The two conditions p(0)p(0) odd and p(1)p(1) odd serve as "parity guards" for even and odd candidates respectively. Any integer root must be one or the other, and each case kills itself via divisibility.

The key insight: you don't need to know anything about the degree of pp or the size of the root. The single fact aโˆ’bโˆฃp(a)โˆ’p(b)a - b \mid p(a) - p(b) does all the work.


Ray's Solution

Write p(x)=โˆ‘i=0naixip(x) = \sum_{i=0}^n a_i x^i.

Since p(0)=a0p(0) = a_0 is odd, a0a_0 is odd. Since p(1)=โˆ‘i=0naip(1) = \sum_{i=0}^n a_i is odd and a0a_0 is odd, we get โˆ‘i=1nai\sum_{i=1}^n a_i is even.

Suppose x0x_0 is an integer root, so p(x0)=a0+โˆ‘i=1naix0i=0p(x_0) = a_0 + \sum_{i=1}^n a_i x_0^i = 0.

Case 1: x0x_0 is even.

Each x0ix_0^i is even, so each term aix0ia_i x_0^i is even. Thus โˆ‘i=1naix0i\sum_{i=1}^n a_i x_0^i is even (sum of even numbers). So:

p(x0)=a0โŸodd+โˆ‘i=1naix0iโŸeven=oddโ‰ 0โ€”ย contradictionp(x_0) = \underbrace{a_0}_{\text{odd}} + \underbrace{\sum_{i=1}^n a_i x_0^i}_{\text{even}} = \text{odd} \neq 0 \quad \text{โ€” contradiction}

Case 2: x0x_0 is odd.

Each x0ix_0^i is odd, so x0iโ‰ก1(mod2)x_0^i \equiv 1 \pmod{2} for all iโ‰ฅ1i \geq 1. Therefore:

โˆ‘i=1naix0iโ‰กโˆ‘i=1naiโ‹…1=โˆ‘i=1naiโ‰ก0(mod2)\sum_{i=1}^n a_i x_0^i \equiv \sum_{i=1}^n a_i \cdot 1 = \sum_{i=1}^n a_i \equiv 0 \pmod{2}

So โˆ‘i=1naix0i\sum_{i=1}^n a_i x_0^i is even, and again:

p(x0)=a0โŸodd+โˆ‘i=1naix0iโŸeven=oddโ‰ 0โ€”ย contradictionp(x_0) = \underbrace{a_0}_{\text{odd}} + \underbrace{\sum_{i=1}^n a_i x_0^i}_{\text{even}} = \text{odd} \neq 0 \quad \text{โ€” contradiction}

In both cases p(x0)โ‰ 0p(x_0) \neq 0, so pp has no integer roots. โ–ก\square

This approach is more elementary than the divisibility trick โ€” it works directly from the coefficients, using that mod 2, all odd numbers are 1, so multiplying by an odd number preserves parity of the sum.