Hi, sure I can provide more detail. a can be any factor, but regarding the implementation I used a=1.
As to what I have tried, I wanted to offrain from using too much detail because it will streamline the discussion into a certain direction, eg using fftw, which implements the discrete fourrier transform and I am not sure about the relationship between discrete fourrier transform and the actual fourrier transform.
I tried the following:
I tried using julia for the computation and doing some plots
x = -5:dx:5
p = fftfreq(length(x),2pi/dx)
scatter(p, real.(fft(step.(x))))
scatter!(p, imag.(fft(step.(x))))
however the real and imaginary parts dont really seem to match the expected result at all.
However the absolute value only seems to be off by a constant
Also when looking at the formula, the discrete fourrier transform is given by
X_k=\sum_{n=0}^{N-1} x_n \cdot e^{-\frac{i 2 \pi}{N} k n}
while the fourrier transform is given by
\hat{f}(\xi)=\int_{-\infty}^{\infty} f(x) e^{-i 2 \pi \xi x} d x
I am however unsure how they are related.
I do see some relationship between the two, in particular if I would to approximate the integral by a Riemann sum
\int_{-\infty}^{\infty} f(x) \, dx \approx \frac{1}{N}\sum_{N}f(x[N])
this reletionship would suggest divideing by N to obtain the correct fourrier transform but that does’t lead to a correct result eather, also it doesn’t help with the feact that the real and complex components don’t match.
I also looked at [this post][2] from maths stackexchange, but it did not really prove to be useful, because the answers dont answer my question.