Is this still a valid way to write the Fourier transform?

47 Views Asked by At

I was trying to separate the periodicity aspect of the unit circle in the Fourier transform, by creating some patches:

$$f(x) = \sum_{n=-\infty}^{\infty} \mathbb{1}_{[0,\lambda]} (x-n\lambda) f(x), $$

where $\mathbb{1}_{[0,\lambda]}(x)$ is the indicator function. Now let's define the patches as,

$$f^n(\lambda,x) = \mathbb{1}_{[0,1]}(x) f(\lambda x + \lambda n),$$

which limits the domain to $[0,1]$. Lets then define:

$$\hat{f}(k)=\sum_n \int_0^{1} w(x) f^n \left(2\pi k^{-1},x\right) dx.$$

If $w(x)=e^{2\pi i x}$, is $\hat{f}(k)$ the same as the usual Fourier transform?