We next need to build a connection between the concept of homotopy in topology, and the behaviour of chain complexes and homology in algebra.
Let and be chain complexes, and let be chain maps (so and ). A chain homotopy between and is a system of maps such that for all (or more briefly, ). In the case , we should interpret as , so the condition is . We say that and are chain homotopic if there is a chain homotopy between them. If so, we write .
In Section 2 we introduced a crude intuitive version of homology involving chains as subsets of a space . For such a chain we can define , which is a chain in ; we call this the thickening of . If is a filled triangle in , then is a triangular prism. Now is the boundary of this triangular prism, which consists of the top, the bottom and the sides. On the other hand, is what we get by thickening the boundary of , which is just the sides of the prism. After adjusting the -signs to account for orientations, we end up with the relation
This is a relation between chains in , but if we have two maps and a homotopy between them, then we can apply to get a relation between chains in . The main point of this section is to provide a rigorous and general version of this picture.
In Example 13.7 we introduced a chain complex with
The differential is given by and . Define by and . This is a chain map because . We claim that is chain homotopic to the identity. Indeed, we can define by , and we define to be zero for all . We then find that
The relation of being chain homotopic is an equivalence relation.
The zero map is a chain homotopy from to itself. If is a chain homotopy from to , then is a chain homotopy from to . If is also a chain homotopy from to , then is a chain homotopy from to . ∎
Suppose we have chain maps
Suppose that is chain homotopic to , and that is homotopic to . Then is chain homotopic to .
We are assuming that is chain homotopic to , which means that there is a chain homotopy with . Similarly, there is a chain homotopy with . Put . Using we get . Using we get . By adding these, we get
as required. ∎
We write for the set of chain homotopy classes of chain maps from to , or in other words equivalence classes under the equivalence relation defined above. Using Proposition 14.5, we see that these are the morphism sets of a well-defined category , whose objects are chain complexes. This is analogous to the category introduced in Definition 9.9.
Let and be chain complexes, and let be chain maps that are chain homotopic. Then the induced maps are the same.
Let be a chain homotopy from to , so . Consider an element , so with . Recall from Proposition 13.11 that so that the expressions and are meaningful and refer to elements of . By definition we have and , so we need to check that these are the same. We have but so . It follows that in , or in other words . This means that as claimed. ∎
Let and be topological spaces, and let be continuous maps that are homotopic to each other. Then the chain maps are chain homotopic to each other, so the induced maps are the same.
For the proof, we will first choose a homotopy from to , so is a continuous map with and for all . We need to use this to construct a chain homotopy between and , or equivalently a system of maps with . Before giving the general proof, we will discuss the cases and .
Consider a point . We can define a continuous map by . This can be regarded as an element of , and it satisfies
We define and extend linearly to get a homomorphism with for all .
Now consider instead an element , or in other words, a continuous map . We want to define , so should be a -linear combination of continuous maps from the triangle to . We have a map given by . Here is homeomorphic to so is a square, with corners , , and . We can divide this square into two triangles and restrict to these triangles, giving two different maps , which will be the terms in . In detail, we define maps by
The composites (for ) can be regarded as elements of . We define . This can be extended linearly to give a homomorphism . We claim that for all . Indeed, it will be enough to prove this when . We then have
(where as before). Here, for we have
This can be displayed as follows:
It follows that
Thus, in our formula for we see that the first and last terms give , the second and fifth terms cancel out, and the third and fourth terms give . Putting this together, we get as required.
We now extend the above discussion to cover .
For we define by
If it is necessary to specify , we will write instead of .
When the maps are given by
We saw above that is the union of the triangles and , which fit together nicely along one edge. In the same way, it can be shown that is the union of the images of the maps , and the intersection of any two of these images is another simplex of lower dimension. However, we will not need this so we omit the proof.
When we have
Given and as before, we put
We extend this linearly to define .
Suppose that and , so we can form the composite
If then is the same as the composite
or in other words .
On the other hand, if then .
For we also have .
Finally, we have and .
All the maps under discussion are affine maps from to . To check that two such maps agree, it suffices to check that they have the same effect on the vertices of . The map sends to (if ) or (if ). The map sends to (if ) or (if ).
When we find that both and have the following effect:
When we find that both and have the following effect:
When we find that both and have the following effect:
We also have and
∎
For all we have
It will be enough to prove this when , so . We can then define , so . This gives
We can divide this sum into four parts:
is the sum of the terms where
is the sum of the terms where
is the sum of the terms where with
is the sum of the terms where with
consists of the terms with or .
We thus have .
Now note that , and
This gives
We let be the sum of the terms where , and we let be the sum of the terms where .
Each term in can be rewritten (using part (a) of the lemma) as . This can then be written as , where and . Because for terms in , we see that , so the rewritten term is the negative of a term in . Similarly, each term in can be rewritten (using part (b) of the lemma) as . This can then be written as , where and . Because for terms in , we see that , so the rewritten term is the negative of a term in . Using this we see that and . A similar argument with part (c) of the lemma shows that . We now have
Also, part (d) of the lemma gives
The first of these terms has sign , and the second has sign . We therefore have as required. ∎
We can gain some insight into the above proof by considering a simple special case. Suppose that and . Suppose that is affine, i.e. for some matrix and vector . Suppose that is also affine, and that is just the linear homotopy . Consider a linear -simplex . As all the maps involved are affine, we see that
As in Example 10.17, we will use abbreviated notation, writing for or , and for , so the above equations become and . It is then not hard to check that
and similarly for the other terms in . The terms in , , and can now be laid out as follows:
Most terms cancel in the indicated groups, which correspond to the expressions in the proof of Proposition 14.13, leaving as expected. Here we have just displayed the case , but the pattern generalises in an obvious way to other values of . This presentation is only directly relevant for linear simplices and affine maps. However, in the general case, most of the work involves linear simplices in the space , and then we finish up by applying the map . Because of this, it is possible to deduce the general case from the linear case, although we will not spell out the details here.
If is a homotopy equivalence, then is an isomorphism.
Choose a map which is homotopy inverse to , so is homotopic to and is homotopic to . As homology is a functor, the composite is the same as . As is homotopic to , Proposition 14.8 tells us that . Using functoriality again, we have . Putting this together, we see that , and essentially the same argument shows that . Thus, and are mutually inverse isomorphisms. ∎
If is homotopic to a constant map, then the map is zero for all .
Suppose that is contractible. Then but for all . In particular, this applies if is a convex subset of for some .