3
$\begingroup$

I know, correct me if I am wrong, that the functions $H_n(x)\exp(-x^2/2)$ form a complete basis in $L^2(\mathbb{R},dx)$, where $H_n(x)$ is the $n$th Hermite polynomial. This must be true also for $x^n\exp(-x^2/2)$ with $n\in\mathbb{N}_0$. Does someone know a proof of the latter or can give me a reference? Since I am not a mathematician, I will really appreciate it if the proof contains all the details that might puzzle a non-mathematician.

I also have the problem to choose functions $f_k$ such that

\begin{equation} \int_{-\infty}^{+\infty}f_k(x)x^n\exp(-x^2)dx = \delta_{kn} \mbox{.} \end{equation}

Does anyone know a method to construct the $f_k$s?

$\endgroup$
  • $\begingroup$ "This must be true also for...": can you clarify this sentence? Are you asking whether the functions $x^n \exp(-x^2/2)$ also form a complete basis for $L^2(\mathbb{R}, dx)$? Note that $x^n$ (or any other polynomial) can be written as a linear combination of Hermite polynomials. $\endgroup$ – Nate Eldredge Apr 9 '13 at 14:00
  • $\begingroup$ Yes, $x^n$ can be written as superposition of Hermite polynomials. Therefore, I wrote "This must be true also for...". However, I don't know how the proof that this is a complete basis in $L^2$ looks like, neither for $x^n$ nor for $H_n$. That is why I asked for a detailed proof or a reference where such a proof can be found. $\endgroup$ – Physicist Apr 9 '13 at 15:12
5
$\begingroup$

Wikipedia has the following proof of completeness. Suppose $f\in L^2(\mathbb R,\exp(-x^2))$ is orthogonal to all $H_n$, that is $$\int_{\mathbb R} f(x)x^n \exp(-x^2)\,dx=0\quad \text{ for all }\ n\tag1$$ Consider the entire function $$ F(z)= \int_{\mathbb R} f(x)\exp(zx-x^2)\,dx \tag2 $$ where the integral converges because both $f$ and $e^{zx}$ are in $L^2(\mathbb R,\exp(-x^2))$. Calculating the $n$th derivative $F^{(n)}(0)$, we obtain
$$ F^{(n)}(0)= \int_{\mathbb R} f(x)x^n\exp( -x^2)\,dx =0 \tag3 $$ Thus $F\equiv 0$. In particular, $F(iy)\equiv 0$, which says precisely that the Fourier transform of $f(x)\exp(-x^2)$ is zero. By the Fourier inversion formula, $f(x)\exp(-x^2)\equiv 0$, completing the proof.

$\endgroup$
  • $\begingroup$ Another question removed from the unanswered queue.+1 $\endgroup$ – Shuhao Cao Jul 16 '13 at 4:34

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.