Hover over the menu bar to pick a physics animation.

This is the time-dependent analog of resonator1D page which found the stationary states of an electromagnetic wave (optical) resonator. Basically we release an optical pulse and observe its motion inside a 1D domain.

I recently discovered a very thorough paper on use of finite difference methods to solve the wave equation. In great detail it goes through boundary conditions, etc. Here I will use some of its methods to compute the propagation of a wave function.

For the wave equation, we have the second derivatives in both time and space:

`(d^2E)/(dt^2)=c^2(d^2E)/(dx^2)`

where c is the speed of propagation of light in vacuum. Translated to finite difference (FD) expressions this bocomes`1/(deltat)^2(E_i^(n+1)-2E_i^n+E_i^(n-1))=(c^2)/(deltax^2)(E_(i+1)i^(n)-2E_i^n+E_(i-1)^(n))`

where index n denotes the time increment and index i denotes the space increment. If we multiply both sides of this equation by `deltat^2` we have:`(E_i^(n+1)-2E_i^n+E_i^(n-1))=c^2(deltat^2)/(deltax^2)(E_(i+1)i^(n)-2E_i^n+E_(i-1)^(n))`

We can now solve for `E` at the next time increment, `E_i^(n+1)``E_i^(n+1)=2E_i^n-E_i^(n-1)+c^2(deltat^2)/(deltax^2)(E_(i+1)i^(n)-2E_i^n+E_(i-1)^(n))`

We can get the difference `deltaE_i^(n+1)=deltaE_i^(n+1)-E_i^n` from this equation:`E_i^(n+1)-E_i^n=E_i^n-E_i^(n-1)+c^2(deltat^2)/(deltax^2)(E_(i+1)i^(n)-2E_i^n+E_(i-1)^(n))`

We see that`deltaE_i^(n+1)=deltaE_i^n+c^2(deltat^2)/(deltax^2)(E_(i+1)i^(n)-2E_i^n+E_(i-1)^(n))`

Therefore, if the second derivative in space is not zero, the increment in `(dE)/dt`, which is `deltaE_i^(n+1)-deltaE_i^n`, which is really just `(d^2E)/(dt^2)` and will also not be zero.For the case of electromagnetic wave propagation the equation becomes

`deltaE_i^(n+1)=deltaE_i^n+c^2(epsilon_0)/epsilon(deltat^2)/(deltax^2)(E_(i+1)i^(n)-2E_i^n+E_(i-1)^(n))`

where `epsilon` is the dielectric constant of the slab and `epsilon_0` is the dielectric constant of vacuum. Then epsilon/epsilon_0` is the square of the index of refraction of the slabs.We start with a simple sine pulse with gaussian rolloff in the center of the domain (canvas). The index of refraction of the domain is 1.0 except for two fixed slabs to the left and right side of the sine pulse. Since the pulse is just a standing sine wave, it will split into two sub-pulses and propagate to the left and right until it reaches the edge of the domain. As you can see, the sub-pulses slow down, their amplitude is diminished, and their spatial frequency becomes much larger in the slabs. The edges of the slabs have swaged index rolloff which reduces reflections of the sub-pulses.

On the first cycle of the pulse propagation where the sub-pulses return to the center of the domain, the sub-pulses usually merge (reconstruct) into almost exactly the same pulse that was started. You can confirm this by use of the Single Step checkbox when the pulses are almost merged. However, on susbsequent cycles, depending on the slab index and its swaged rolloff width this reconstruction will get less and less perfect due to reflections from the slab boundaries.