2

Prove that the Legendre polynomials $P_n$ have $n$ zeros in $(-1,1)$

I already proved that the Legendre polynomials $P_n$ have degree $n$ and any zero belong to $(-1,1)$, but Im stuck trying to prove that these polynomials have $n$ zeros.

I tried different induction hypothesis to build a proof but no one works.

The definition of the Legendre polynomials is

$$P_n(x):=\frac1{2^nn!}\partial^n[(x^2-1)^n],\quad n\in\Bbb N_{\ge 0}$$

I derived the alternative forms

$$\begin{align}P_n(x)&=2^{-n}\sum_{k=0}^n\binom{n}{k}^2(x+1)^{n-k}(x-1)^k\\&=\frac1{2^nn!}\sum_{k=\lceil n/2\rceil}^n\binom{n}{k}(2k)^{\underline n}(-1)^{n-k}x^{2k-n}\end{align}$$

where $(2k)^{\underline n}=\prod_{j=0}^{n-1}(2k-j)$ is a falling factorial. I have some useful identities to play as

$$\partial^n (fg)=\sum_{k=0}^n(\partial^k f)\cdot(\partial^{n-k}g)$$

$$\partial^{n+1}(x^2-1)^{n+1}=\partial^n[\partial (x^2-1)^n]=2n\partial^nx(x^2-1)^{n-1}$$

$$\partial^nx(x^2-1)^n=x\partial^n(x^2-1)^n+n\partial^{n-1}(x^2-1)^n$$

but as I said I cant get something useful from here. Some help will be very welcome, thank you.

P.D.: for this proof I cant use any integration theorem or so, this is a proof on the context of differentiability.

Masacroso
  • 30,417
  • 2
    Rolles theorem will help. [If noone gives a full answer I will write one tomorrow.] –  Dec 14 '16 at 21:31
  • 1
    Check this answer http://math.stackexchange.com/questions/527140/proof-the-legendre-polynomial-p-n-has-n-distinct-real-zeros?rq=1 – Daniel Dec 14 '16 at 21:49
  • 1
    This theorem: https://en.wikipedia.org/wiki/Gauss%E2%80%93Lucas_theorem and Rodrigues' formula provide a simple proof. – Jack D'Aurizio Dec 14 '16 at 22:35

0 Answers0