4
$\begingroup$

Let $f\colon[-1,1]\to\mathbb{R}$ be a continuous function. Prove that $$ 2\int_{-1}^{1}f(x)^2\: dx - \left(\int_{-1}^{1}f(x)\: dx\right)^2 \ge 3\,\left(\,\int\limits_{-1}^{1}x\,f(x)\: dx\right)^2 $$


I found something similar:

Integral Inequality $\int\limits_0^1f^2(x)dx\geq12\left( \int\limits_0^1xf(x)dx\right)^2.$

but don't know if estimation of that sort can be applied here.

$\endgroup$
4
$\begingroup$

Let $\displaystyle\;\bar{f} = \frac12\int_{-1}^1 f(x) dx\;$ and $g(x) = f(x) - \bar{f}$. It is easy to check $$\int_{-1}^1 g(x) dx = 0 \quad\text{ and }\quad \int_{-1}^1 x g(x) dx = \int_{-1}^1 xf(x) dx $$ The LHS of inequality at hand can be rewritten as $$\verb/LHS/ = 2\int_{-1}^1 (g(x) + \bar{f})^2 dx - 4\bar{f}^2 = 2\int_{-1}^1 g(x)^2 dx = 3\left(\int_{-1}^1 x^2 dx\right)\left(\int_{-1}^1 g(x)^2 dx\right) $$ Apply Cauchy Schwarz to $x$ and $g(x)$ over $[-1,1]$, we find

$$\verb/LHS/ \ge 3 \left(\int_{-1}^1 xg(x)\right)^2 = 3 \left(\int_{-1}^1 xf(x) dx\right)^2 = \verb/RHS/ $$

$\endgroup$
4
$\begingroup$

It is useful to exploit the fact that Legendre polynomials give a complete orthogonal base of $L^2(-1,1)$ with the usual inner product. Given: $$ f(x) = \sum_{n\geq 0} a_n P_n(x) \tag{1}$$ we have: $$ \begin{array}{llcl}\text{(Parseval's identity)}&\displaystyle\int_{-1}^{1}f(x)^2\,dx &=& \displaystyle2\sum_{n\geq 0}\frac{a_n^2}{2n+1},\\ (P_0(x)=1)&\displaystyle\int_{-1}^{1} f(x)\,dx &=& \displaystyle2a_0,\tag{2}\\(P_1(x)=x)& \displaystyle\int_{-1}^{1} x\,f(x)\,dx &=& \displaystyle\frac{2}{3}a_1\end{array}$$

hence the original inequality is equivalent to: $$ 4\sum_{n\geq 0}\frac{a_n^2}{2n+1}-4a_0^2\geq \frac{4}{3}a_1^2 \tag{3} $$ or to:

$$ \sum_{n\geq 2}\frac{a_n^2}{2n+1}\geq 0 \tag{4}$$

that is trivial. We also have that equality is attained only by linear functions, $f(x)=a+bx$.

$\endgroup$
  • $\begingroup$ A very good answer, +1, but very likely an elementary solution is expected. $\endgroup$ – user261263 May 27 '16 at 13:27
  • $\begingroup$ Very interesting approach! one question - there is an infinite series to integrate, what makes us sure we can integrate it function by function? $\endgroup$ – user340508 May 27 '16 at 13:34
  • $\begingroup$ @user340508: a continuous function is for sure an $L^2(-1,1)$ function, Legendre polynomials give a complete base of $L^2(-1,1)$ and Parseval's identity holds. $\endgroup$ – Jack D'Aurizio May 27 '16 at 13:39

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.