Weak homotopy equivalence of topological spaces: Difference between revisions

From Topospaces
(inserted twisted homology criterion)
 
(One intermediate revision by one other user not shown)
Line 11: Line 11:
''Basepoint choice disclaimer for homotopy group isomorphism'': To concretely define the map <math>\pi_n(f)</math>, we need to choose basepoints for <math>A</math> and <math>B</math>. Change of basepoint, however, only results in pre-composition and post-composition by isomorphisms and does not affect whether or not the map is an isomorphism.
''Basepoint choice disclaimer for homotopy group isomorphism'': To concretely define the map <math>\pi_n(f)</math>, we need to choose basepoints for <math>A</math> and <math>B</math>. Change of basepoint, however, only results in pre-composition and post-composition by isomorphisms and does not affect whether or not the map is an isomorphism.


===Definition for path-connected spaces in terms of fundamental groups, and homology groups===
===Equivalent definition for path-connected spaces in terms of homology groups===


Let <math>A</math> and <math>B</math> be [[path-connected space]]s. A '''weak homotopy equivalence''' from <math>A</math> to <math>B</math> is a continuous map <math>f:A \to B</math> such that the following two conditions are true:
Let <math>A</math> and <math>B</math> be [[path-connected space]]s. Then a continuous map <math>f:A \to B</math> is a weak homotopy equivalence iff both of these conditions hold:


* The functorially induced map <math>\pi_1(f): \pi_1(A) \to \pi_1(B)</math> is an isomorphism of [[fundamental group]]s.
* The induced map <math>\pi_1(f):\pi_1(A) \to \pi_1(B)</math> is an isomorphism of groups.
* The functorially induced maps <math>H_n(f): H_n(A) \to H_n(B)</math> are group isomorphisms for all <math>n \ge 1</math>.
* For every bundle of abelian groups <math>\mathcal A</math> over <math>B</math>, the induced map of twisted homology groups <math>f_*:H_n(A;f^* \mathcal A) \to H_n(B;\mathcal A)</math> is an isomorphism of groups for all <math>n \ge 0</math>.


Note that since the maps are homomorphisms anyway, it is enough to require them to be bijective.
As above, all of these maps are homomorphisms anyway, so it is enough to require them to be bijective. The above basepoint disclaimer for <math>\pi_1</math> also applies here.


''Basepoint choice disclaimer for fundamental group isomorphism'': To concretely define the map <math>\pi_1(f)</math>, we need to choose basepoints for <math>A</math> and <math>B</math>. Change of basepoint, however, only results in pre-composition and post-composition by isomorphisms and does not affect whether or not the map is an isomorphism.
If both <math>A</math> and <math>B</math> are [[Simply connected space|simply connected]] then the criterion is simpler: a continuous map <math>f:A \to B</math> is a weak homotopy equivalence iff the induced map on homology with <math>\mathbb Z</math> coefficients <math>f_*:H_n(A; \mathbb Z) \to H_n(B; \mathbb Z)</math> is an isomorphism of groups for all <math>n \ge 0</math>.


===Definition for spaces that are not path-connected===
===Definition for spaces that are not path-connected===
Line 28: Line 28:
* The functorially induced map <math>\pi_0(f): \pi_0(A) \to \pi_0(B)</math> is a bijection between the [[set of path components]] <math>\pi_0(A)</math> and the set of path components <math>\pi_0(B)</math>.
* The functorially induced map <math>\pi_0(f): \pi_0(A) \to \pi_0(B)</math> is a bijection between the [[set of path components]] <math>\pi_0(A)</math> and the set of path components <math>\pi_0(B)</math>.
* For every path component of <math>A</math>, the restriction of <math>f</math> to a continuous map from that to its image path component of <math>B</math> is a weak homotopy equivalence of path-connected spaces.
* For every path component of <math>A</math>, the restriction of <math>f</math> to a continuous map from that to its image path component of <math>B</math> is a weak homotopy equivalence of path-connected spaces.


==Facts==
==Facts==

Latest revision as of 04:38, 19 November 2013

This article defines a property of continuous maps between topological spaces

Definition

Definition for path-connected spaces in terms of homotopy groups

Let A and B be path-connected spaces. A weak homotopy equivalence from A to B is a continuous map f:AB such that the functorially induced maps πn(f):πn(A)πn(B) are group isomorphisms for all n1.

Note that since the maps are homomorphisms anyway, it is enough to require them to be bijective.

Basepoint choice disclaimer for homotopy group isomorphism: To concretely define the map πn(f), we need to choose basepoints for A and B. Change of basepoint, however, only results in pre-composition and post-composition by isomorphisms and does not affect whether or not the map is an isomorphism.

Equivalent definition for path-connected spaces in terms of homology groups

Let A and B be path-connected spaces. Then a continuous map f:AB is a weak homotopy equivalence iff both of these conditions hold:

  • The induced map π1(f):π1(A)π1(B) is an isomorphism of groups.
  • For every bundle of abelian groups A over B, the induced map of twisted homology groups f*:Hn(A;f*A)Hn(B;A) is an isomorphism of groups for all n0.

As above, all of these maps are homomorphisms anyway, so it is enough to require them to be bijective. The above basepoint disclaimer for π1 also applies here.

If both A and B are simply connected then the criterion is simpler: a continuous map f:AB is a weak homotopy equivalence iff the induced map on homology with Z coefficients f*:Hn(A;Z)Hn(B;Z) is an isomorphism of groups for all n0.

Definition for spaces that are not path-connected

Let A and B be topological spaces. A weak homotopy equivalence from A to B is a continuous map f:AB such that:

  • The functorially induced map π0(f):π0(A)π0(B) is a bijection between the set of path components π0(A) and the set of path components π0(B).
  • For every path component of A, the restriction of f to a continuous map from that to its image path component of B is a weak homotopy equivalence of path-connected spaces.

Facts

  • The existence of a weak homotopy equivalence from A to B does not imply the existence of a weak homotopy equivalence from B to A. Thus, to get an equivalence relation on topological spaces, we need to take a symmetric transitive closure. We say that spaces are weak homotopy-equivalent topological spaces if they are in the same equivalence class under the equivalence relation thus obtained.
  • The mere fact that πn(A)πn(B) as abstract groups is not enough to guarantee that A and B are weak homotopy-equivalent, even when A and B are manifolds or CW-spaces (see isomorphic homotopy groups not implies weak homotopy-equivalent). Rather, it is specifically important that the map must induce those isomorphisms.
  • The exception to the above is in the case that both πn(A) and πn(B) are the trivial group/one-point set for all n. In this case, any map must induce isomorphisms since that's the only possible map between trivial groups/one-point sets. In this case, the spaces A and B are both weakly contractible spaces.
  • Similarly, the mere fact that π1(A)π1(B) as abstract groups and Hn(A)Hn(B) as abstract groups does not imply that A and B are weak homotopy-equivalent. See isomorphic homology groups and isomorphic fundamental groups not implies weak homotopy-equivalent. Rather, it is specifically important that the map must induce those isomorphisms.
  • The exception to the above is, once again, where the fundamental group and all the homology groups Hn,n1, are trivial.

Relation with other properties

Stronger properties

Weaker properties