gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (11/05/84)
One fairly simple way to see why c (speed of light) is an absolute speed limit is to consider the Lorentz equations relating two Cartesian co"ordinate systems that are in motion at a uniform velocity relative to each other; without loss of generality orient their X axes parallel to the direction of motion and keep their Y and Z axes parallel to the other system. Call the relative speed "v". The transformation from one system to the other is x' = (x - (v/c) ct) / sqrt(1 - (v/c)^2) y' = y z' = z ct' = (ct - (v/c) x) / sqrt(1 - (v/c)^2) These were obtained by Einstein from a conceptual analysis of motion coupled with the assumption that the speed of light is the same in all uniformly-moving reference frames (also assumes that similar units are used to measure time and distance). There are other ways to derive these equations; the important point is that they embody the MEANING of relative motion with uniform velocity. Now, as v approaches c, the Lorentz transformation equations develop a singularity (denominator approaches 0), so it is hopeless to try to reach superluminal speeds by accelerating from below the speed of light (if one calculates the energy required to do so, it would be infinite). However, these days we are well aware of the possibility of creating new particles in subatomic reactions, so that it is at least conceivable that a particle could be created that has v > c. If one considers v > c, then the primed co"ordinates x' and t' become pure imaginary quantities. This is a tip-off that we have pushed the model beyond the realm in which its quantities have a direct physical interpretation. One could stop here and say that this shows that such a speed has no meaning, but if we nonetheless persevere and try to wring some meaning out of the imaginary co"ordinates, here is what happens: In order to relate these quantities to intuitive co"ordinates, the simplest approach is to examine the Minkowski metric tensor, which for the above Lorentz transformations always takes the form ( -1 0 0 0 ) ( 0 -1 0 0 ) ( 0 0 -1 0 ) ( 0 0 0 1 ) (for simplicity I take (x,y,z,ct) as the four coordinates, in that order). It is the special nature of the [4,4]-component of this metric (which is already diagonalized, saving us some work) that indicates that the fourth co"ordinate has a different nature ("time-like") from the other three. If you don't know about the metric, is it one particularly simple way of describing invariant geometric structure on a manifold (there are others that are more fundamental, but they are beyond the scope of a special- relativistic discussion). Denoting the metric tensor as "g", it relates the co"ordinates (call them "w" collectively, which is not the usual notation but I want to avoid multiple use of symbols) to an invariant (co"ordinate-independent) measure of interval "s" by (ds)^2 = Sum over i (Sum over j (g[i,j] dw[i] dw[j])) (The d's are differentials.) Because ds is invariant and dw[] is a vector (rank (1+0) tensor), the symmetric part of g[,] is a rank (0+2) tensor. (The antisymmetric part is irrelevant to this discussion.) For more about metrics, consult a differential geometry text. There are several possibilities for mapping the imaginary co"ordinates obtained for v > c back into real measure-numbers so that we can puzzle out their meaning. All reasonable schemes amount to the same thing, so consider the simplest, which is to just multiply the relevant co"ordinates by i (the complex unit, sqrt(-1)): x'' = i x' y'' = y' z'' = z' ct'' = i ct' This transformation takes the components of the metric tensor to ( 1 0 0 0 ) ( 0 -1 0 0 ) ( 0 0 -1 0 ) ( 0 0 0 -1 ) The new (x'') co"ordinate is now the time-like one. To make comparison easier, one can optionally apply a further transformation x''' = ct'' y''' = y'' z''' = z'' ct''' = x'' in order to put the time-like coordinate back in its accustomed place. This final set of space-time coordinates (along with its "standard form" Minkowski metric) is suitable for direct interpretation as Cartesian spatial coordinates and time coordinate of a reference system moving "in the X direction with speed v > c" with respect to the original "stationary" coordinate system. But look what has happened in the process of straightening out the imaginary coordinates! The relation between the triple-prime set of coordinates and the original unprimed set, when you work it out, is precisely that of an equivalent Lorentz transformation with parameter u < c, where u = c^2 / v One way of looking at the above is to say immediately upon production of imaginary co"ordinates for superluminal v that that demonstrates the unreality of motion faster than light (for, in order to describe "motion", we must have a thing that "moves", and such a thing must be capable of having a real-valued space-time reference frame associated with it). If one wants to go farther and try to interpret such co"ordinates as meaningful, by the time one has coerced them into a form such that their meaning is clear, he has shown that such motion is entirely equivalent to (i.e., IS really) ordinary subluminal motion. This duality of velocity is interesting. Try playing with the dual interpretations in a variety of relativistic formulae, such as that for composition of relative speeds: v[AC]/c = (v[AB]/c + v[BC]/c) / (1 + (v[AB]/c)(v[BC]/c)) The practical significance, apart from keeping one from having to worry too much about superluminal speeds, is that if one DOES allow v > c in some formula, e.g. particle path-integral summations, then he will be overcounting by a factor of 2 (except possibly at v == c, which clearly calls for special treatment).
herbie@watdcsu.UUCP (Herb Chong, Computing Services) (11/07/84)
Weren't the Lorentz transformations derived after it was decided that the speed of light was the same in all inertial frames? I know that the mathematics yields negative energy and all kinds of things like that for beta greater than 1 in the energy/momentum equations. Herb Chong... I'm user-friendly -- I don't byte, I nybble.... UUCP: {decvax|utzoo|ihnp4|allegra|clyde}!watmath!watdcsu!herbie CSNET: herbie%watdcsu@waterloo.csnet ARPA: herbie%watdcsu%waterloo.csnet@csnet-relay.arpa NETNORTH, BITNET: herbie@watdcs, herbie@watdcsu