I have read that Parseval's theorem, relating the norm of a function $f$ and the norm of its Fourier transform $g(k)$:

\begin{equation} \int |f(x)|^2 dx=\int|g(k)|^2 dk \end{equation}

has the simple physical interpretation of "conservation of energy". I just don't see this, so can you suggest me a way to think about it?

  • $\begingroup$ it’s not quite right--- the conservation of energy assumes each Fourier mode is oscillating separately, so that the energy is either a sum over modes or a sum over positions, and this is a consequence of Parseval's theorem. Proving Parseval's theorem is best using the abstract idea that the integral is the "length" of the function considered as a vector, and the length doesn't depend on your choice of orthonormal basis. $\endgroup$ – Ron Maimon Apr 8 '12 at 23:20

Think of a vector $\mathbf{V}$. As seen in a coordinate system $S$ with basis vectors $\hat{e}_i$, it can be written $$\mathbf{V} = \sum_i V_i \hat{e}_i$$ where $V_i$ are the components of $\mathbf{V}$ in $S$. As seen from another coordinate system $S'$ with basis vectors $\hat{e}_i'$, it has a representation $$\mathbf{V} = \sum_i V_i' \hat{e}_i'.$$ Obviously the length of the vector is independent of the coordinate system used to represent it. In other words, we must have $$\sum_i V_i^2 = \sum_i (V_i')^2$$

Proceeding with this analogy, for a function $f(x)$ one can have a position space representation in $\delta$-function basis as $$f(x) = \int f(x') \delta(x-x') dx'$$ where the "component" of $f(x)$ along the "basis vector" $\delta(x-x')$ is $f(x')$ and we sum (integrate since $x$ is a continuous variable) over all the possible "axes".

One can look at the same function in Fourier-space representation as $$f(x) = \int g(k) e^{-i k x} dk$$ where $e^{-ikx}$ are the "basis vectors" and $g(k)$ are the "components" of $f(x)$ along these basis vectors.

You would then agree that $$ \int |f(x)|^2 dx = \int |g(k)|^2 dk$$

So Parseval's theorem is just the restatement of the invariance of the length of a "vector" independent of the representation used.

  • If $|f(x)|^2$ is proportional to the energy, then Parseval's theorem is a statement of the conservation of the energy as seen in the real-space domain or the Fourier-space domain

  • If $f(x)$ is a quantum-mechanical wavefunction, $|f(x)|^2$ is proportional to the probability density. Parseval's theorem is then a statement of the conservation of the probability as seen in the position-space representation or the momentum-space representation.

See also Parseval's identity

  • $\begingroup$ The question asked for an understanding of Parseval's theorem as a statement of conservation of energy, not just a proof of the theorem. $\endgroup$ – kleingordon Apr 8 '12 at 20:04

In short - Energy in time domain is equal to energy in frequency domain.

EDIT: Think of f(x) as a signal, then Parseval's theorem says that total energy of this signal over time equals to energy of this signal over frequencies.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.