MathOverflow is a question and answer site for professional mathematicians. 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 know coefficients of some function in basis $p_j,j=1...K$ where

$p_{j}(x)=\sum_{s\in Z}a_{s,j}\exp\left(-2\pi i(j+sK)x\right)$

With respect to inner product $(f,g)=\int_{0}^{1}f(x)\overline{g(x)}dx$ this basis is orthogonal. What kind of basis is it? Can I somehow use Fourier analysis framework for this basis?

One more question about this basis. How to solve this integral equation? $$a_{j}=\int_{-\infty}^{\infty}g(x)p_{j}(x)dx$$ $g-$unknown function. Could you give me some hint, or some links?

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

If $a_{s,j}$ is in $\ell_2$ for any fixed $j$, $p_j$ forms an orthogonal basis for some finite dimensional sub-space of $L^2(\mathbb{T})$. If $f(x) = \sum_{j=0}^K b_j p_j(x)$, then its fourier coefficients are just given by $\hat{f}(\xi) = a_{s,j} b_j$ where $j = \xi \mod K$ and $s = \lfloor \xi / K \rfloor$, so surely normal Fourier analytic techniques apply.

I don't quite understand your question about "What kind of basis is it?" Can you clarify?

share|cite|improve this answer
I am wondering if it is possible to use basis $p_j$ instead of$\exp(2\pi ij)$ to define Fourier transform, inverse fourier transform and so on. – vilvarin May 25 '10 at 14:31
It is a basis of a finite dimensional sub-space, so not all functions will be in the span of $p_j$. Recall that the Fourier coefficients can be defined as $\hat{f}(\xi) = \langle f(\cdot), \exp(2\pi i \xi \cdot)\rangle$. A general representation formula is that, for some Hilbert space $H$ with orthonormal bases $e_1,e_2, \ldots$ you can write $f = \sum \lambda_j e_j$ where $\lambda_j = \langle f,e_j\rangle$ (this essentially captures the Fourier transform and its inverse, completely analogous to the finite dimension inner product space case)... – Willie Wong May 25 '10 at 16:20
... so consider $f(x) = 1$, the constant function on $[0,1]$. Compute $\langle f, p_j\rangle = \delta_{jK} a_{-1,0}$ (in your notation where numbering of $j = 1... K$. But obviously $a_{-1,0} p_0 \neq f$ unless all other $a_{s,0} = 0$. But if you have some a priori knowledge that all the functions you care about lies in the span of the $p_j$s, then you don't even need Fourier analysis: you are just working with finite dimensional linear algebra on some inner product space. – Willie Wong May 25 '10 at 16:24
.. I meant to write that $\langle f,p_j\rangle = \delta_{jk}a_{-1,K}$ and $a_{-1,K}p_K \neq f$ unless all other $a_{s,K} = 0$. Sorry for the typo. – Willie Wong May 25 '10 at 16:25
thank you very much. The last question I have is if Fourier inversion theorem is valid for this basis. It is not true in general case, isn't it? Probably i need to check this – vilvarin May 25 '10 at 19:14

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.