home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!think.com!news!columbus
- From: columbus@strident.think.com (Michael Weiss)
- Newsgroups: sci.physics
- Subject: Lowneheim-Skolem theorem (was: Continuos vs. discrete models)
- Date: 16 Nov 92 10:59:41
- Organization: Thinking Machines Corporation, Cambridge MA, USA
- Lines: 42
- Message-ID: <COLUMBUS.92Nov16105941@strident.think.com>
- References: <1992Nov7.214329.24552@galois.mit.edu>
- <1992Nov13.194334.20447@sun0.urz.uni-heidelberg.de> <350@mtnmath.UUCP>
- NNTP-Posting-Host: strident.think.com
- In-reply-to: paul@mtnmath.UUCP's message of 14 Nov 92 16:10:21 GMT
-
- In an earlier post, Paul Budnik writes:
-
- For example the real numbers definable in any consistent formal system
- are countable.
-
- and in his last post, he explains what he meant:
-
- This well known result is called the Lowenheim and Skolem theorem. The
- idea of the proof is that a formal system is a computer program for enumerating
- theorems. The names of all real numbers created by such a program are
- obviously countable. Of course the mapping of these real numbers to names is
- not definable within the the formal system and thus these reals cannot be
- shown to be countable within the system.
-
- This is not the correct statement of the Loewenheim-Skolem theorem. Here is
- one formulation (the so-called "downward" Loewenheim-Skolem theorem):
-
- Any model of a countable first-order theory has a countable elementary
- submodel. (By "countable theory", we mean one whose language is
- countable. "Elementary" means, roughly, that any assertion true of the
- model is true also of the submodel.)
-
- Budnik's statement about there being only a countable infinity of definable
- real numbers is correct, if by "definable" we mean "definable by a formula
- in a countable theory". This fact is essentially trivial, since there are
- only a countable number of formulas in such a theory.
-
- One has to be a bit cautious drawing philosophical conclusions from the
- Loewenheim-Skolem theorems. (Or rather, one doesn't have to be, but angels
- fear to tread.) Let me rush in anyway: say L is a countable first-order
- language rich enough to express "all of physics", as currently understood,
- and in particular includes enough of the theory of R (real numbers) for
- traditional QM, QED, GR, etc. (L could be the language of ZFC, for
- example.) Philosopher X objects to L, claiming that uncountable infinities
- do not really exist, and so demands that physics be reformulated on another
- basis. Philosopher Y retorts that a mere distaste for the uncountable is
- no grounds for rejecting L, since we can make the same assertions and just
- interpret them as applying to an elementary countable submodel.
-
- In short: there may be good reasons for radically changing the foundations
- of physics (and there may not-- pace, defenders of the status quo(ntum)!),
- but the Loewenheim-Skolem theorem is not one of them.
-