home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!snorkelwacker.mit.edu!galois!riesz!jbaez
- From: jbaez@riesz.mit.edu (John C. Baez)
- Subject: What is a knot?
- Message-ID: <1992Nov7.212557.24399@galois.mit.edu>
- Sender: news@galois.mit.edu
- Nntp-Posting-Host: riesz
- Organization: MIT Department of Mathematics, Cambridge, LA
- Date: Sat, 7 Nov 92 21:25:57 GMT
- Lines: 52
-
- In article <COLUMBUS.92Nov6105242@strident.think.com>
- columbus@strident.think.com (Michael Weiss) writes:
-
- >How would one define a (tame) knot, intrinsically? Definitions I am
- >familiar with either involve modding out by ambient isotopy (in fact
- there
- >are subtle points here, I believe-- perhaps someone more knowledgeable
- >would like to post), or by Reidemeister moves.
-
- Well, a (tame) knot is just a circle embedded in R^3, or, if you prefer
- something less intuitive, S^3. But you seem to be speaking of the
- ambient isotopy class of the knot, i.e., where one forgets about
- everything except its topology. As you note this can be defined as
- either
- 1) equivalence class of knots under diffeomorphisms of R^3 that are
- connected to the identity in the group of diffeomorphisms (i.e.,
- reflection is a no-no).
- 2) equivalence class of knots under the following: two knots
- (embeddings) are equivalent if they can be connected by a continuous
- 1-parameter family of embeddings.
- 3) equivalence class of knot diagrams (2d pictures of knots, which I
- won't define) under the Reidemeister moves and diffeomorphisms of the
- plane that are connected to the identity in the group of
- diffeomorphisms.
-
- These are all equivalent.
-
- >Is the field just too young to have a suitably slick and (on first
- >encounter) unintuitive definition?
-
- If you want some definitions that are less intuitive, you could try the
- following. These are actually definitions of isotopy classes of *links*
- - it soon becomes clear that there's no point in studying knots without
- studying links too.
-
- 4) equivalence classes of braids under the Markov moves
- 5) Hom(0,0) in the category of unframed unoriented tangles.
-
-
- Actually, it turns out to be very good to work with framed oriented links.
- Then we have the marvelously erudite
-
- 5') Hom(0,0) in the free tortile tensor category generated by a single
- object.
-
- This is the best definition for getting link invariants from quantum
- group representations.
-
- All the above definitions involve equivalence relations, although in
- some cases this is hidden.
- ~
-
-