Is the Discrete Fourier Transform a Unitary Transformation?

Emperor42
Messages
15
Reaction score
0
I'm trying to prove that the discrete form of the Fourier transform is a unitary transformation

So I used the equation for the discrete Fourier transform:
##y_k=\frac{1}{\sqrt{N}}\sum^{N-1}_{j=0}{x_je^{i2\pi\frac{jk}{N}}}##

and I put the Fourier transform into a N-1 by N-1 matrix form:
##U=\begin{pmatrix}
e^0 & e^0 & e^0 & ...\\
e^0 & e^{\frac{i2\pi}{N}} & e^{\frac{i4\pi}{N}} & ...\\
e^0 & e^{\frac{i4\pi}{N}} & e^{\frac{i8\pi}{N}} & ...\\
... & ... & ... & ...\\
\end{pmatrix}##

and then found the complex conjugate:
##U^*=\begin{pmatrix}
e^0 & e^0 & e^0 & ...\\
e^0 & e^{-\frac{i2\pi}{N}} & e^{-\frac{i4\pi}{N}} & ...\\
e^0 & e^{-\frac{i4\pi}{N}} & e^{-\frac{i8\pi}{N}} & ...\\
... & ... & ... & ...\\
\end{pmatrix}##

But if I multiply these matrices together I get nothing which even approaches the identity matrix. Anyone have any ideas? Is there something wrong with the matrix?
 
Physics news on Phys.org
You omitted the factor of ##1/\sqrt{N}##. Without it, the matrix won't be unitary. Also, the inverse of a unitary matrix is its adjoint, not the complex conjugate.
 
I understand that I haven't put in the ##\frac{1}{\sqrt{N}}##. But why do I need the inverse? I'm trying to calculate whether the matrix is unitary so I need to find the inner product of the matrix and its complex conjugate, wouldn't I?
 
Last edited:
Ok I think I can get the diagonal elements to go to if I add the ##\frac{1}{\sqrt{N}}## factor but I still don't understand how the off diagonal elements go to zero.
 
Emperor42 said:
I understand that I haven't put in the ##\frac{1}{\sqrt{N}}##. But why do I need the inverse? I'm trying to calculate whether the matrix is unitary so I need to find the inner product of the matrix and its complex conjugate, wouldn't I?
No. Unitary means that the inverse of the matrix is its adjoint. In other words, if you multiply a unitary matrix by its adjoint (not conjugate), you get the identity matrix, which is what you're trying to show.
 
Hello everyone, I’m considering a point charge q that oscillates harmonically about the origin along the z-axis, e.g. $$z_{q}(t)= A\sin(wt)$$ In a strongly simplified / quasi-instantaneous approximation I ignore retardation and take the electric field at the position ##r=(x,y,z)## simply to be the “Coulomb field at the charge’s instantaneous position”: $$E(r,t)=\frac{q}{4\pi\varepsilon_{0}}\frac{r-r_{q}(t)}{||r-r_{q}(t)||^{3}}$$ with $$r_{q}(t)=(0,0,z_{q}(t))$$ (I’m aware this isn’t...
Hi, I had an exam and I completely messed up a problem. Especially one part which was necessary for the rest of the problem. Basically, I have a wormhole metric: $$(ds)^2 = -(dt)^2 + (dr)^2 + (r^2 + b^2)( (d\theta)^2 + sin^2 \theta (d\phi)^2 )$$ Where ##b=1## with an orbit only in the equatorial plane. We also know from the question that the orbit must satisfy this relationship: $$\varepsilon = \frac{1}{2} (\frac{dr}{d\tau})^2 + V_{eff}(r)$$ Ultimately, I was tasked to find the initial...
Back
Top