FAQ Revised: Thursday 11 December 2003 14:03:43
When we say "fast" and "slow", we are referring to a quantity known as the group velocity of a pulse. Pulse propagation is very complicated and the group velocity is an approximation of how fast a pulse travels. For a simple description of the group velocity, see our tutorial on fast light.
Yes, there are many ways. We describe our technique (the gain doublet) and the absorption technique on our Making Fast Light page, but there are several others. These include frustrated total internal reflection, quantum tunneling, geometrical techniques, and more. Many of these can be modeled in similar ways, so some would argue that they are all equivalent effects.
Now that is an entirely different effect and our work says nothing about it. Quantum entanglement can be used to achieve apparently superluminal results, too. It is also accepted that entanglement cannot be used to transmit information faster than c, but for entirely different reasons.
Short answer? Yes! That's not as un-physical as you might think. In fact, requiring true finite bandwidth requires that the pulse be infinite in time! That is, the pulse must exist infinitely far into the future and past.
Most of the time when we say a signal has a small bandwidth, what we really mean is that its FWHM (full-width at half-maximum) is small. We usually don't worry about what's happening far from the center, where only a small fraction of the power is held. For most problems, that's a reasonable approximation, but it's still an approximation!
Filtering a signal will certainly change the spectrum of a signal, and it may change its spectral FWHM. However, if a filter truly eliminates all frequencies outside some range, then the filter is acausal. Such filters are unphysical because they (would) make the filtered signal exist for all times, including times before the input signal existed. Therefore, a filter can reduce the power outside some frequency range, but it cannot eliminate it.
The detection latency is the delay between when information starts arriving and when you're convinced that it's arriving. Lets take the very simple case of a smooth pulse arriving. In any real situation, there is always noise, and the front edge of the pulse is small. When are you sure that the pulse is really there? Only after the pulse is big enough to be clearly visible above the noise. At that point, though, the pulse has been arriving for a while. This delay is the detection latency.
The detection latency is always present. You may be able to make it small by using better equipment, pulse shapes, or detection schemes, but you can never eliminate it entirely.