Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have the following:

Suppose $p(x) =(x^{2}-1)^{m}$ for some $m \geq2$.

Applying Rolle's Theorem on each of the intervals $[-1,c]$ and $[c,1]$, show that $p''(x)$ has at least two roots in $(-1,1)$.

By induction, prove that for $k=1,2,\ldots,m$ the $k$-th derivative $p^{(k)}(x)$ has at least $k$ distict real roots lying in the interval $(-1,1)$.

I have first of all shown that for $p'(x)$ that there is a root in $(-1,1)$, but struggling with the other two parts, any help would be most appreciated.

share|cite|improve this question
up vote 1 down vote accepted

We have that $$p(x) = (x^2-1)^m$$ Note we need $m >2$.

Taking derivatives we have



Rolle's Theorem asserts that if $f(x)$ is continuous in $I=[a,b]$ and $f(a)=f(b)$, then there is a $c \in I$ such that


We have that $f(-1) = f(1)$, so there is a $c$ such that $f'(c)=0$. We can immediatelly see such $c$ is $c=0$.

Moreover, we see that $f'(-1) = f'(1)=0$, but since $f'(0)=0$, we can assert that there exists $c_1 \in (-1,0)$ and $c_2 \in (0,1)$ such that $f''(c_1)=f''(c_2)=0$. Note that if $m=1,2$ the test would fail, since the first (third) derivative is a line.

You want to prove now that for $k=1,2,3,\dots ,m$, it is the case $p^{(k)}(c_i)=0$ for $i=1,2,\dots, m \in (-1,1) $. The induction hypothesis would then be:

$\mathcal H.$ It is true that $p^{(k)}(c_i)=0$ for $i=1,2,\dots,k \in (-1,1) $

This means that $p^{(k)}(c_1) = p^{(k)}(c_2) = \cdots = p^{(k)}(c_i)=0$. Since we also have two more roots, namely $1$ and $-1$, then we can assert by Rolle's Theorem, that $p^{(k+1)}(x)$ will have $n=k+2-1=k+1$ roots. Since $k=2$ is true, this completes the induction phase.

share|cite|improve this answer

You know that $p'(x)$ has a root somewhere in $(-1,1)$. Call that "somewhere" $c$.

Now, $p'(x) = m(x^2-1)^{m-1}(2x)$. Since $m-1\gt 0$, it is still the case that $p'(-1)=p'(1)=0$. But you also know that $p'(c)=0$. So set $f(x)=p'(x)$, and note that we have $f(-1)=f(c)=0$, and you can use Rolle's Theorem to conclude that $f'(x) = p''(x)$ has a root in $(-1,c)$; likewise, since $f(c)=f(1)=0$, it follows that $f'(x)=p''(x)$ has a root in $(c,1)$.

Now assume that you know that $p^{(r)}(x)$ has $r$ roots on $(-1,1)$; call them $-1\lt c_1\lt c_2\lt\cdots \lt c_r\lt 1$, and that $p^{(r+1)}$ still is zero at $-1$ and at $1$ (this requires $r\lt m$, as is not hard to check). Now you know that $p^{(r)}(-1)=p^{(r)}(c_1)=\cdots=p^{(r)}(c_r) = p^{(r)}(1) = 0$. You can apply Rolle's Theorem to $p^{(r)}$ on $[-1,c_1]$, on $[c_1,c_2]$, on $[c_2,c_3]$, and so on.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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