Carroll lists the important parts of the diagram
i^+= future timelike infinity (T=\pi,R=0)
i^0= spatial infinity (T=0,R=\pi)
i^-= past timelike infinity (T=-\pi,R=0)
J^+= future null infinity (T=\pi-R,0<R<\pi)
J^-= past null infinity (T=-\pi+R,0<R<\pi)
Carroll use a symbol like \mathcal {J} not J which he calls "scri". It is hard to reproduce.
Conformal diagrams are spacetime diagrams with coordinates such that the whole of spacetime fits on a piece of paper and moreover light cones are at 45° everywhere. The latter makes it easy to visualize causality. Since Minkowski spacetime has 45° light cones, if coordinates can be found which have a metric which is a conformal transformation of the Minkowski metric, the job is done. The first part of appendix H is devoted to finding conformal coordinates for flat Minkowski spacetime, expressed in polar coordinates - presumably to ease our work later in spherically symmetrical manifolds such as Schwarzschild. So we start from that metric: {ds}^2=-{dt}^2+{dr}^2+r^2\left({d\theta}^2+\sin^2{\theta}{d\phi}^2\right)
On the way to finding conformal coordinates we tried coordinates \bar{t}=\arctan{t}\ \ ,\ \ \bar{r}=\arctan{r}
which certainly pack spacetime into the range -\frac{\pi}{2}<\bar{t}<\frac{\pi}{2}\ ,\ 0\le\bar{r}<\frac{\pi}{2}
as you will see below if you press the button. Carroll says it might be fun to draw the light cones on that, so I made a movie:
Light cone at various \bar{r}
Carroll's Figure H.2 is also quite confusing. It does not show the u,v axes and I naturally assumed that the u axis pointed down and to the right. It does not. It does the opposite.
Read all the details at Commentary App H Conformal Diagrams.pdf (12 pages)
I emailed Claude Semay about this to thank him. He replied very quickly
ReplyDeleteDear colleague,
Thank you very much for your interest in my work (the only one with a weak link to general relativity). The incorrect diagram is frequently seen in popular articles or books, even on Wikipedia. I'm glad your calculations confirm mine.
Best regards, CS