I was given the following:

Given that $P(x)$ is a second degree polynomial, and that $P(11)=151$, and that $$\forall x \in \mathbb{R}, \,\, x^2-2x+2 \le P(x) \le 2x^2-4x+3$$ determine $P(21)$.

Where do I even start from?

Is there a source (online or a book) where I can find similar testing questions on polynomials? (Testing not in the sense that Putnam-like questions, but in the sense that they are somewhat different than those found in textbooks.)


The graph helps conceptualise why this is sufficient information; the bounding quadratics given share a minimum at $(1,1)$. For the inequality to hold, $P(x)$ must also have its minimum at $(1,1)$, and there is only one parabola passing through $(11,151)$ with that minimum.

Since the minimum is at $(1,1)$, $P(x)=a(x-1)^2+1$.

$$P(11)=151$$ $$100a+1=151$$ $$a=1.5$$

Two parabolas and the point (11,151).

  • $\begingroup$ How does one show that if $P(x)$ have a minimum other than $(1,1)$, then the inequality is broken? $\endgroup$ – joeblack May 13 at 13:11
  • $\begingroup$ Call the bounding polynomials $A(x)$ and $B(x)$. We know that $A(1)\leq P(1) \leq B(1)$ so $P(1)=1$. $(1,1)$ is the minimum of $A(x)$ so $A(x)>1$ everywhere except $x=1$. But then $P(x)\geq A(x)>1$ everywhere except $x=1$, so $(1,1)$ is the minimum of $P(x)$. $\endgroup$ – dbmag9 May 13 at 13:17
  • $\begingroup$ @joeblack You can write any parabola with minimum at $(m,n)$ as $P(x) = a(x-m)^2 + n$ (with $a>0$). The term $(x-m)^2$ is never smaller than $0$, so $x=m$ must be the minimum, and $P(m) = a\cdot 0 + n = n$. But then, $P(x)$ only has one variable parameter left: $a$. So defining a single other point fixes the parabola. $\endgroup$ – Thern May 13 at 13:18
  • 3
    $\begingroup$ This is a valuable lesson: if you can draw it, it's usefull to do this. You immediately see that the function goes through (1,1). Nice answer! $\endgroup$ – Student May 13 at 13:42
  • 1
    $\begingroup$ @joeblack It needs to pass through $(1,1)$ because $A(x)$ and $B(x)$ go through that point (if $1\leq P(x) \leq 1$ then $P(x)=1$). Given that it passes through $(1,1)$, that must be its minimum because $A(x)$ has its minimum there. $\endgroup$ – dbmag9 May 13 at 14:39

We are given that $(1-x)^{2}+1 \leq P(x) \leq 2(1-x)^{2}+1$. Let $Q(x)=P(x)-1-(1-x)^{2}$. Then $0 \leq Q(x) \leq (1-x)^{2}$ and $Q$ is also a polynomial of degree $2$. This implies that $Q(x)=c(1-x)^{2}$ for some constant $c$. Can you finish from here?


Note that the given condition can be written as $$ (x-1)^2\leq P(x)-1\leq 2(x-1)^2 $$ So this automatically forces $P(x)-1$ to have a repeated root at $x=1$, thus $P(x)=\lambda (x-1)^2+1$ (since $\deg P=2$) for some $1\leq\lambda\leq 2$. Now use $P(11)$ to determine $\lambda$, hence $P(21)$.

  • $\begingroup$ I am sorry if this is a stupid question: Why does $(x-1)^2\leq P(x)-1\leq 2(x-1)^2$ forces $P(x)-1$ to have a repeated root at $x=1$? (Or, had it been $(x-2)^2\leq P(x)-1\leq 2(x-2)^2$, would $P(x)$ be forced to have a repeated root at $x=2$? $\endgroup$ – joeblack May 13 at 13:00
  • 1
    $\begingroup$ I do have a similar concern as @joeblack. $P(x)$ could be of the form $P(x) = \lambda(x-1)^2 + \kappa(x-1) + 1$ and still fulfill the inequalities. You still have to prove why $\kappa$ is zero. $\endgroup$ – Thern May 13 at 13:08
  • $\begingroup$ Yes. Dividing $(x-1)^2\leq P(x)-1\leq 2(x-1)^2$ by $\lvert x-1\rvert$ and take limit $x\to 1$ gives $0\leq\lvert P'(1)\rvert\leq 0$, so $P'(1)=0$ and $P(1)-1=0$. $P(x)-1$ has.a repeat root. Similar with your $(x-2)$ case. $\endgroup$ – user10354138 May 13 at 13:08

$$ P(x)=ax^2+bx+c$$ so $$151 = 121a +11b+c$$

$$(x-1)^2+1\leq P(x)\leq 2(x-1)^2+1\implies P(1) = 1$$

so $$ a+b+c=1$$

so $$120a+10b= 150\implies 12a+b=15$$ and $$P(x) = (x-1)(ax+a+b)+1$$

so $$x-1\leq ax+a+b\leq 2(x-1)$$ $$\implies x=1:\;\;\;2a+b=0$$

So $a={3\over 2}$, $b=-3$ and $c={5\over 2}$ so $P(21) = 601$.

  • 2
    $\begingroup$ You lost $200a$ between the left and right sides of $440a+20b+1=20\cdot15+1.$ Notice that $(1,1),$ $(11,151),$ and $(21,301)$ are collinear, so you've concluded that $P(x)=15x+1.$ This is incorrect. $\endgroup$ – David K May 13 at 10:18
  • $\begingroup$ Now it is corrected. $\endgroup$ – Tarzan May 14 at 8:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.