home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.physics
- Path: sparky!uunet!well!sarfatti
- From: sarfatti@well.sf.ca.us (Jack Sarfatti)
- Subject: re: Baez's unwitting proof that Lorentz invariance->advanced waves.
- Message-ID: <Bzp8y1.3t9@well.sf.ca.us>
- Sender: news@well.sf.ca.us
- Organization: Whole Earth 'Lectronic Link
- Date: Wed, 23 Dec 1992 06:34:48 GMT
- Lines: 19
-
-
- Baez writes:
- Since it's the holiday season I will be nicer than usual; to do the
- above one needs to know what something like delta(f(x)) really means,
- and they don't usually teach this well enough. Say f(0) = 0 and
- f'(0) > 0. What's the difference (near x = 0) between delta(f(x)) and
- the good old Dirac delta(x)? Well, delta(f(x)) only makes sense
- integrated against a smooth test function, and if one does the integral
- by a change of variables one sees that delta(f(x)) = delta(x)/f'(x).
- Makes sense: the faster f(x) changes with x, the skinnier the delta
- function delta(f(x)) is, so one must divide by f'(x). :-) (Yes, friends,
- this CAN be made rigorous. Read Gelfand's book Generalized Functions.)
- So one can show
-
- delta(t^2 - r^2) = delta(t - r)/2r + delta(t + r)/2r
-
- end of Baez's reamrk - I now add that this equation is central to Feynman's
- QED the delta(t-r) is a retarded propagation and the delta(t+r) is an
- advanced propagation backward in time - no way of avoiding it.
-