leichter (05/11/82)
Following all he articles recently about QM got me curious enought to go back and dig up my copy of "The Quantum Theory and Reality" - which is in the November 1979 Scientific American, BTW - and well worth reading. The author (whose name I would certainly misspell, so I won't try) makes one comment at the end of the article to the effect that we might have to give up Einstein seperability - space-like events canot be correlated - but that this does not necessarily mean that FTL COMMUNICATION is possible. A variation on the Einstein-Rosen[?]-? thought experiment that a friend and I came up with many years back, though, seems to provide just such communication! We never did succeed in finding any problems with it, although our QM was too weak for this to mean much (this was while taking a first QM course, all the details of which have long faded). Here is the setup: S, the Sender, will send to R, the Receiver, a stream of bits. To set up their communication, EXACTLY half-way between them they place a source of paired photons; every second, say, the source emits one photon toward S and one toward R, and the photons are created by a process such that a perfect anti-correlation exists between measurements by S and those by R of polarization. (This isn't hard to do.) Between S and the source, we place a polarizer that passes only photons with an Up polarization; between R and the source, we place one that passes only "Down" photons. (Actually, this is badly phrased - if you like, use spins.) S and R have synchronized clocks, so they can observe photons that were emitted from a given event - i.e. photons that are part of a pair. R watches for photons and measures their Up-Down polarization. To send 0, S does nothing. R then sees all Down photons, since they all came through a Down polarizer and nothing has changed them. To send 1, S measures CIRCULAR polarization of the incoming photons. This knocks the pair out of its Up-Down eigenstate; hence, R sees random up-down polarizations. Obviously, this is a noisy channel, so S must measure circular polarization for a while; but eventually he gets through. Let's look at some of the assumptions. We need accurate, synchronized clocks for S and R; standard relativistic arguments show you how to build them (by moving apart slowly.) Given such, placing S and R and precisely the same distance from the source is easy. (In fact, it's not even relevant - it's only necessary that S be no further from the source than R.) The speed of the photons from the source seems to be a limiting factor, but a little thought shows that it only influences how long it takes to set up the experiment; once it's set up, you've got a pipeline going with a much higher effective information transfer rate. I know of no simple way out of this. My guess, which I can't justify, is that when you add in all the probabilities you somehow find that noise is inherently being added at least as fast as you can extract information, so you can't actually get any information through - but it's hard to see where all the additional noise comes in if, say, I double my photon emission rate and keep my "bit time" unchanged, apparently halving my probability of error. Any ideas? (Story idea: Maybe pulsars are sources set up for just this sort of communication. Too bad we have good theories to account for them.) -- Jerry decvax!yale-comix!leichter leichter @ yale