According to Wikipedia, a non-uniform Fourier transform can be calculated as follows:
where are sample points and
are frequencies.
Now, say I have samples $x_n$ taken at times $t_n$. To get my $p_n$ values, presumably I just scale the times of my samples between 0 and 1, like so:
$$p_n = \frac{t_n - \min(t_n)}{\max(t_n) - \min(t_n)}$$
All I need now for the calculation is $f_k$, which I'm a bit confused about.
- What value(s) should $f_k$ assume?
- Or does specifying $f_k$ somehow define the frequency bin associated with $X_k$? If so, how?
- If I set $f_k$ as a constant, say $\alpha$, then what value would the $k^\text{th}$ frequency bin assume?
Many thanks in advance.