Maths Olympiad Prep

Library / /18 of 34

Algebra Difficulty 6.8 National Olympiad Prove it United States

As usual, let Z[x]\mathbb{Z}[x] denote the set of single-variable polynomials in xx with integer coefficients.
Find all functions θ:Z[x]Z\theta : \mathbb{Z}[x] \to \mathbb{Z} such that for any polynomials p,qZ[x]p, q \in \mathbb{Z}[x],
* θ(p+1)=θ(p)+1\theta(p + 1) = \theta(p) + 1, and
* if θ(p)0\theta(p) \neq 0 then θ(p)\theta(p) divides θ(pq)\theta(p \cdot q).

Solution

The answer is θ:pp(c)\theta : p \mapsto p(c), for each choice of cZc \in \mathbb{Z}. Obviously these work, so we prove these are the only ones. In what follows, xZ[x]x \in \mathbb{Z}[x] is the identity polynomial, and c=θ(x)c = \theta(x).

First solution (Merlijn Staps) Consider an integer ncn \neq c. Because xnp(x)p(n)x - n \mid p(x) - p(n), we have
θ(xn)θ(p(x)p(n))    cnθ(p(x))p(n). \theta(x-n) \mid \theta(p(x)-p(n)) \implies c-n \mid \theta(p(x))-p(n).
On the other hand, cnp(c)p(n)c-n \mid p(c)-p(n). Combining the previous two gives cnθ(p(x))p(c)c-n \mid \theta(p(x))-p(c), and by letting nn large we conclude θ(p(x))p(c)=0\theta(p(x))-p(c) = 0, so θ(p(x))=p(c)\theta(p(x)) = p(c).

Second solution First, we settle the case degp=0\deg p = 0. In that case, from the second property, θ(m)=m+θ(0)\theta(m) = m + \theta(0) for every integer mZm \in \mathbb{Z} (viewed as a constant polynomial). Thus m+θ(0)2m+θ(0)m + \theta(0) \mid 2m + \theta(0), hence m+θ(0)θ(0)m + \theta(0) \mid -\theta(0), so θ(0)=0\theta(0) = 0 by taking mm large. Thus θ(m)=m\theta(m) = m for mZm \in \mathbb{Z}.
Next, we address the case of degp=1\deg p = 1. We know θ(x+b)=c+b\theta(x+b) = c+b for bZb \in \mathbb{Z}. Now for each particular aZa \in \mathbb{Z}, we have
c+kθ(x+k)θ(ax+ak)=θ(ax)+ak    c+kθ(ax)ac. c+k \mid \theta(x+k) \mid \theta(ax+ak) = \theta(ax) + ak \implies c+k \mid \theta(ax) - ac.
for any kck \neq -c. Since this is true for large enough kk, we conclude θ(ax)=ac\theta(ax) = ac. Thus θ(ax+b)=ac+b\theta(ax+b) = ac+b.
We now proceed by induction on degp\deg p. Fix a polynomial pp and assume it's true for all pp of smaller degree. Choose a large integer nn (to be determined later) for which p(n)p(c)p(n) \neq p(c). We then have
p(c)p(n)cn=θ(pp(n)xn)θ(pp(n))=θ(p)p(n). \frac{p(c)-p(n)}{c-n} = \theta\left(\frac{p-p(n)}{x-n}\right) \mid \theta(p-p(n)) = \theta(p)-p(n).
Subtracting off cnc-n times the left-hand side gives
p(c)p(n)cnθ(p)p(c). \frac{p(c)-p(n)}{c-n} \mid \theta(p)-p(c).
The left-hand side can be made arbitrarily large by letting nn \to \infty, since degp2\deg p \geq 2. Thus θ(p)=p(c)\theta(p) = p(c), concluding the proof.

Want a route through all this instead of an archive? The track puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement reproduced verbatim; metadata (topic, difficulty) added by this project.