2
$\begingroup$

I'm trying to numerically solve the differential equations of motion in a system with multiple springs of very high frequency. Because the solution is often a combination of rapidly-oscillating sine and cosine waves, it takes a very long time for a numerical solver to find the answer.

However, the fourier transform of the solution is probably fairly simple - it's nearly a delta function peaked at the frequency (or frequencies) at which the system oscillates.

Since Fourier transforms turn $\frac{df}{dx}$ into $2 \pi i \xi \hat{f}(\xi)$, it would seem that we could turn all of Newton's equations of the form $F_k=m\frac{d^2}{dt^2} x_k$ (representing, say, the $k$th particle or rigid body in a complicated system) into $\hat{F_k} = m (2 \pi i t)^2 \hat{x_k}$ (where the Fourier transform is taken with respect to time). Then, we could solve all of these simpler equations for the values of the $\hat{x_k}$, then take the Fourier transform (numerically).

So, do people ever apply Fourier transforms to the case of many-body systems with some harmonic oscillators?

$\endgroup$
  • $\begingroup$ Correction to the question (v1): The $t$ appearing in the last equation $\hat{F_k} = m (2 \pi i t)^2 \hat{x_k}$ should be replaced with the dual variable (=frequency). $\endgroup$ – Qmechanic Jun 4 '11 at 17:43
2
$\begingroup$

Maybe they do; but Fourier transforms have an inherent flaw which makes them less than useful for such cases. That flaw is that they are transforms of steady state conditions: The initial and final conditions of the system are assumed to be the same; and transients are not considered.

The transform for which you seek is the Laplace transform. Laplace is a very close cousin to Fourier, but takes into account initial conditions and allows you to inspect transients and the final state of the system.

$\endgroup$

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.