home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.physics
- Path: sparky!uunet!psinntp!scylla!daryl
- From: daryl@oracorp.com (Daryl McCullough)
- Subject: Re: TIME HAS INERTIA. Att: PRATT FUND. THM. OF ALGEBRA
- Message-ID: <1992Nov22.214312.13337@oracorp.com>
- Organization: ORA Corporation
- Date: Sun, 22 Nov 1992 21:43:12 GMT
- Lines: 61
-
- Abian writes:
- >(3) 1/(kz^n + ...+ bz + a) = (1/k) z^(-n) + ........
- >
- >But (3) shows that 0 is an essential isolated singularity of
- > 1/(kz^n + ...+ bz + a)
- >contradicting (2). Hence our assumption is false and the Theorem is
- >proved.
-
- Vaughan Pratt writes:
-
- >The gist of Abian's proof is that expansion (3) indicates a singularity
- >where, as both you and Abian point out, there can't be one.
- >
- >I was hoping someone would pronounce on the validity of (3), and if it
- >is valid, whether proving it requires essentially the full force of the
- >FTA. If (3) is not valid then Abian owes $500 to anyone taking him up
- >on his bet. If (3) is valid but proving it is as much work as proving
- >the FTA, then Abian would technically win his bet, but his proof would
- >be about as sound as proving the FTA by assuming it (so a bush lawyer
- >might try to collect anyway :-).
-
- The validity of (3) is trivial to prove: multiply top and bottom of
- 1/(kz^n + ...+ bz + a) by (1/z)^n to get:
-
- (1/z)^n/(k + ... + b (1/z)^{n-1} + a (1/z)^n)
-
- Changing variables to let w = 1/z gives
-
- w^n/(k + ... + b w^{n-1} + a w^n)
-
- which is an analytic function of w, provided that the denominator is not
- zero, which it never is (by assumption). An analytic function can always
- be expanded into a power series that converges inside a radius r, where
- r is the largest number such that there is no singularity of the function
- inside r. Since the function above is total, the radius of convergence
- is infinite, so the above function is expandable into a power series in
- w which converges everywhere.
-
- k w^n + ...
-
- Going back to z = 1/w gives us (3) above.
-
- What I think is the tricky part of Abian's proof is the assumption
- that just because a function has an essential singularity at z = 0,
- then it can't be well-defined in the limit z -> 0. Just because the
- expansion of 1/(kz^n + ...+ bz + a) has an infinite number of negative
- powers of z, why does it follow that the limit of the function as
- z -> 0 cannot be 1/a?
-
- It actually is true that if a function f(z) has an expansion with an
- infinite number of negative powers of z that is valid near z = 0, then
- the limit of f(z) as z -> 0 is undefined (it doesn't have a value
- which is independent of the way the limit is taken). However, the
- proof of this fact may very well require something as powerful as the
- Fundamental Theorem of Algebra. Professor Abian, could you tell us a
- simple proof that a power series with an infinite number of negative
- powers of z cannot have a limit as z -> 0?
-
- Daryl McCullough
- ORA Corp.
- Ithaca, NY
-