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

This is an exercise from Werner's Funktionalanalysis. I have to show that the linear span of the functions $f_n(x)=x^ne^{-x^2/2}, n\geq0$ is dense in $L^2(\mathbb{R})$. The book gives the hint to first show that the Fourier transform of $\overline{f(x)}e^{-x^2/2}$ vanishes if $\langle f,f_n\rangle=0$ for all $n$.

Now I don't see how this knowledge about the Fourier transform would help with the original statement. The $f_n$ obviously don't contain some orthogonal set, so even if I could conclude that $f=0$ from this I couldn't use some maximal orthogonal system argument for Hilbert spaces.

Also, I am unable to actually show the hint because for $f \in L^2(\mathbb{R})$ the formula for the Fourier transform does not hold and most of the nice properties of the Fourier transform I know are only valid for Schwartzfunctions.

So any kind of hint as to how this fits together would be helpful.

share|cite|improve this question
Let $M=span\{f_n\}$. One way is to show that $M^{\perp}=\{0\}$ or equivalently, if $f\in L^2$ satisfies $(f,g)=0\ \forall\ g\in M$ then $f=0$. If you take $g=f_n$ in the last equality, you hae that $(f,f_n)=0\ \forall\ n$. Now try to use the hint. – Tomás Dec 7 '12 at 14:06
But I actually have to show $\overline{M}^{\perp}=\{0\}$ because it only says dense. It probably need not be true that $M^{\perp}=\{0\}$. – erlking Dec 7 '12 at 14:26
The Hilbert space $L^2(\mathbb{R})$ splits as $\overline{M} \oplus \overline{M}^{\perp}$. You need to show that $\overline{M}^{\perp} = \{0\}$. As $M^{\perp} = \overline{M}^{\perp}$, it is enough to show that $M^{\perp} = \{0\}$. – levap Dec 7 '12 at 14:31
up vote 2 down vote accepted

Let $t\in\Bbb R$. We have for $t\in\Bbb R$ that $$\sum_{n=0}^{+\infty}\int_{\Bbb R}f(x)e^{-x^2/2}\frac{(itx)^n}{n!}=0.$$ As for all integer $n$, $$\left|f(x)e^{-x^2/2}\frac{(itx)^n}{n!}\right|= |f(x)|e^{-t^2/2}\frac{|x|^n|t|^n}{n!},$$ and the RHS is integrable for the product of Lebesgue and counting measure, we can switch the integral and the sum, which gives that for all $t\in\Bbb R$, $$\int_{\Bbb R}f(x)e^{-x^2/2}e^{itx}dx=0.$$ As $x\mapsto f(x)e^{-x^2/2}$ is integrable, by uniqueness of Fourier transform $f(x)e^{-x^2/2}=0$ almost everywhere, hence $f\equiv 0$.

share|cite|improve this answer


If $f\in L_2(\mathbb{R})$, then $f(x)e^{-x^2/2}$ is Fourier transformable. You can prove this by Schwartz inequality.

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.