Talk:Algebraic topology: Difference between revisions
Comment |
Saku~enwiki (talk | contribs) No edit summary |
||
| Line 4: | Line 4: | ||
[[User:Charles Matthews|Charles Matthews]] 10:45, 3 Dec 2003 (UTC) |
[[User:Charles Matthews|Charles Matthews]] 10:45, 3 Dec 2003 (UTC) |
||
In complement to Charles Matthews' comment: |
|||
1) The first homology group of a plane minus an infinite number of disjoint holes is definitely not finitely generated. I think it is safe to say that "for a complex of finitely generated chain groups, the homology groups are finitely generated." |
|||
2) The relation of torsion and orientability is indeed not very simple, and one should take some care: One can define the homology of topological spaces where the notion of orientability makes no sense. However, for e.g. simplicial complexes, one can define the notion of orienting cycle, but I don't know if there are simplicial complexes with such cycles which do not cover a topological space which is acceptable for a manifold. Moreover, the question is not simple even for manifolds: The Möbius band is an example of a non-orientable manifold with boundary with no torsion in its absolute homology groups, and the lens spaces (in Hatcher's book) seem to be orientable manifolds, even if their absolute homology groups have torsion. There is a theorem which states that "The absolute homology groups of compact, orientable manifolds with boundary which are embeddable into R^3 are torsion-free". If you ease any of the assumptions 1) "absolute" 2) "compact" 3) "orientable" 4) "embeddable into R^3", you can demonstrate a counterexample with torsion. |
|||
Revision as of 11:39, 30 July 2004
I'm sure the claim that homology groups are always finitely generated is a bit overenthusiastic. What do we need to assume, maybe connected manifolds? AxelBoldt
The talk about torsion and orientability: when a space isn't a smooth manifold, orientability isn't any longer a naive concept.
Charles Matthews 10:45, 3 Dec 2003 (UTC)
In complement to Charles Matthews' comment:
1) The first homology group of a plane minus an infinite number of disjoint holes is definitely not finitely generated. I think it is safe to say that "for a complex of finitely generated chain groups, the homology groups are finitely generated."
2) The relation of torsion and orientability is indeed not very simple, and one should take some care: One can define the homology of topological spaces where the notion of orientability makes no sense. However, for e.g. simplicial complexes, one can define the notion of orienting cycle, but I don't know if there are simplicial complexes with such cycles which do not cover a topological space which is acceptable for a manifold. Moreover, the question is not simple even for manifolds: The Möbius band is an example of a non-orientable manifold with boundary with no torsion in its absolute homology groups, and the lens spaces (in Hatcher's book) seem to be orientable manifolds, even if their absolute homology groups have torsion. There is a theorem which states that "The absolute homology groups of compact, orientable manifolds with boundary which are embeddable into R^3 are torsion-free". If you ease any of the assumptions 1) "absolute" 2) "compact" 3) "orientable" 4) "embeddable into R^3", you can demonstrate a counterexample with torsion.