Let be a space. A singular -simplex in is a continuous map . We write for the set of singular -simplices in .
Recall that is the set with just one point. To give a function is the same as to give a point ; this, we can identify with .
As usual, we identify with , with the point corresponding to the point . Thus, a singular -simplex in is the same as a continuous map , or in other words a path in . This means that is the set of all possible paths in .
Suppose that . We can then define a map
(or in other words an element ) by
We call maps of this type linear simplices.
In the case we have and the map just corresponds to the point . In the case , the map corresponds to the straight line path from to . In the case , the image of the map is the triangle with vertices , and .
Now suppose that and that . It may or may not happen that the image of the map actually lies in ; this must be checked carefully in any context where we want to use this construction. If so, we can regard as an element of .
This picture shows a space , together with:
A linear -simplex , which is not an element of .
Another linear -simplex .
A nonlinear -simplex .
Let be a set. We write for the set of formal -linear combinations of elements of . Thus, if then , for example. We call the free abelian group generated by . (It is clearly an abelian group under addition.)
Suppose that is finite, say . We then have an isomorphism given by
However, we will most often be considering cases where is infinite.
A singular -chain in is a formal -linear combination of singular -simplices, or in other words, an element of . We write for the group of singular -chains. For convenience, we also define for .
Consider again the picture in Example 10.5:
The expression is a singular -chain.
The expression is a singular -chain.
The expression is a singular -chain.
No expression involving gives a singular chain in , because the straight line from to is not contained in .
Suppose we have paths and in . We can reverse to get a path , or we can join and to get a path .
We can regard and as elements of , but they are not the same. Similarly, we can regard and as elements of , but they are not the same. There is clearly an important relationship between and , and between and , but it will take a little work to formulate this mathematically.
We next need to define the algebraic boundary for a -chain . We start by considering the cases , and .
For we just define .
Now consider a singular -simplex . This is a path with endpoints and . These endpoints are elements of the set , which we identify with , so the difference can be regarded as an element of . We define . More generally, suppose we have a -chain , with and . We then put
This defines a homomorphism .
Note that for a linear -simplex , we just have . Thus, in the picture below we have a -chain
with
We now consider -chains. For the simplest case, suppose that and is a linear -simplex. In this case, we define
The rule for nonlinear singular -simplices is essentially a straightforward adaptation of the linear case, but it will rely on some auxiliary definitions given below. Once we have defined for all , we will then define for all by the rule
just as we did for singular -chains. This gives a homomorphism .
For a linear -simplex , we will have
For a general linear -simplex , we will have
For with we define by
Equivalently, we have
Thus, the coordinates of are the same as the coordinates of , except that we insert a zero in position .
In the case we have maps . These are given by and .
In the case , we have
Thus, the image of is the edge of opposite the vertex .
Similarly, in the case , we have a map from the triangle to the tetrahedron , and the image is the face of the tetrahedron that is opposite the vertex . The case is shown below.
Even more generally, we see that the map gives a homeomorphism from to .
Consider an element (with ), or equivalently a continuous map . For each with we have a map and we can compose this with to get a map , or in other words an element . We put
More generally, given an element , we define .
For a singular -simplex we have . Here sends the unique point of to , so the map corresponds to the point . Similarly, sends the unique point of to , so the map corresponds to the point . We therefore have , just as in Predefinition 10.11.
Now consider a linear -simplex , so
We find that
so and and . This gives
just as in Predefinition 10.11. It should be clear that the same pattern works for all , giving
The following result is crucial for the development of homology theory.
For all , we have in . Thus, the composite
is zero.
Recall that we defined for , and any homomorphism to the zero group is automatically the zero homomorphism. Thus, the proposition has no content for . For the first nontrivial case, suppose that , and consider a linear -simplex . We then have
We will often use abbreviated notation for this kind of calculation, writing for and for , for example. With this notation, the above calculation becomes
We now discuss where , using the same kind of notation. First, we have
We can write the terms of in a square array, with in the first column, in the second column, and so on. The result is as follows:
We find that the terms above the wavy line cancel in the indicated groups with the terms below the wavy line, leaving as claimed.
If then .
Consider a point . To form , we insert a zero in position . To form , we insert another zero in position . Because , inserting this second zero does not move the first zero, so we end up with zeros in positions and .
Similarly, to form , we insert a zero in position . To form , we insert another zero in position . As we see that the first zero is to the right of the point where we insert the second zero, so the first zero gets moved over by one space into position . Thus, we again end up with zeros in positions and . In the remaining positions, we have the numbers in order. Thus, we have as claimed. ∎
In the case where the claim is that . Explicitly, for we have
We will now prove Proposition 10.16 in the case . Consider a continuous map , or equivalently an element . We have
We can write the terms of in a square array, with in the first column, in the second column, and so on. The result is as follows:
Lemma 10.18 gives us the following identities:
Using this, we see that in the previous array, the terms above the wavy line cancel in the indicated groups with the terms below the wavy line, showing that as claimed. This generalises the argument for linear simplices given in Example 10.17.
Consider a continuous map , or equivalently an element . We have
We can write this as , where
Here and are just dummy variables, so we can rewrite as
We now reindex again, taking and . The condition becomes or equivalently . The condition becomes or equivalently . The sign becomes . This gives
However, Lemma 10.18 tells us that here, so , so as claimed.
This proves that whenever is a singular -simplex. More generally, and singular -chain has the form for some integers and singular -simplices . We then have for all and so . ∎
We say that an element is a -cycle if . We write for the abelian group of -cycles, so .
We say that an element is a -boundary if there exists with . We write for the abelian group of -boundaries, so .
We note that if then for some , so by Proposition 10.16, so . This means that , so we can form the quotient abelian group . We call this the ’th homology group of .
The elements of are cosets with , so with . We will often write for . Before writing notation like one must check that ; it is an error to use that notation in other cases. Note that iff iff there exists with .
There is essentially only one example that we can calculate directly from the definition.
If consists of a single point, then and for .
There is only one possible map from to , sending all possible points in to the unique point of . We call this map , so and for all (whereas for by definition). For we have . Here is a map from to so it can only be equal to . This gives
and so on. In general, we have and . It follows that and . In particular, for all we have so the quotient group is trivial. On the other hand, and so . All this can be tabulated as follows:
∎
We leave the following slight generalisation to the reader. Suppose that is a finite, discrete set of points, so that every continuous map is constant. Then , and for .
We can also calculate for all .
There is a canonical isomorphism for all topological spaces . Thus, if then .
This should not be a surprise. Both and are ways of constructing an abelian group from , in such a way that points connected by a path give the same element of the group. We just need to check that the technical differences between these two constructions do not affect the final answer.
First note that is zero by definition, so the map sends everything to zero, so . This means that the quotient group is the same as .
Next, let be the usual quotient map, which sends every point to the corresponding path component . We can extend this linearly to give a homomorphism , by the rule
We will show that is surjective, with kernel . Assuming this, the First Isomorphism Theorem will give us an isomorphism from to , as required.
Next, for each path component , we choose a point , so . This means that the composite
is the identity. We can also extend linearly to give a homomorphism by the rule . In this context, we see that the composite
is again the identity. In particular, any element is the same as , so it is in the image of ; this proves that is surjective.
Now suppose we have a path . We then have , so . However, we have a path joining to , so the corresponding path components are the same, so . As everything is extended linearly, the rule remains valid for all . The image of is , so this means that , or equivalently .
Next, consider a point and the corresponding path component . The points and both lie in the same path component , so there must exist a path from to in . We choose such a path and call it . This defines a function from to the set of paths in , which we extend linearly to get a hoomorphism . For any point we know that runs from to , so . As everything is extended linearly, the rule is valid for all . In particular, if then so this simplifies to , proving that is in the image of , or in other words .
We can now conclude that is surjective with kernel . By the First Isomorphism Theorem, there is a well-defined homomorphism given by for all , and this is in fact an isomorphism. ∎
The above proof can be illustrated by the following diagram. It shows a space with three path components , and , so and
We have chosen points and and , so and and . To say the same thing in different notation, we have and and , so . The points and also lie in , so , or equivalently . The path runs from to . Similarly, we have , and we have labelled a path running from to .
A typical example of an element of could be the element . This has , so
so . This illustrates the fact that , which is a key step in our proof of Proposition 10.25.
We next discuss homology classes of paths, revisiting Remark 10.10.
Let be a topological space.
For any the constant path actually lies in , so in the quotient group .
For any path in with reversed path , we have so in .
For any paths and we have in .
Video (Path homotopy, loop homotopy and homology)
Let be a topological space. A loop in is a path with , so that , so we have a coset . If , we say that is a loop based at .
Let be a path connected space, and let be a point in . Then for every there exists a loop based at with . Moreover, if and are loops based at then so are , and , and we have and and in .
Let be the subset of consisting of classes that can be expressed as for some loop based at . We must show that this is all of .
It is clear that if and are loops based at , then so are , and . By specialising Lemma 10.27, we see that and and in . It follows from this that is a subgroup of .
Now let be a loop based at a point which may be different from . As is path connected, we can choose a path from to . The path is then a loop based at , and using Lemma 10.27 again we see that
or in other word in . This proves that contains all loops, irrespective of the base point.
Now let be an arbitrary element of . We can write as , where is a -linear combination of paths in . Any term with negative coefficient like can be replaced by without affecting the coset, so we can assume that all coefficients are positive. Then we can replace any term like by repeated times; this gives an expression like
for some list of paths . As this is a homology class, the representing chain must be a cycle, so we must have in . As , this means that
As this is happening in the free abelian group , the terms on the left hand side must just be a permutation of those on the right hand side, so we have a permutation of with for all . We can now write as a product of disjoint cycles. If one of these cycles is , for example, then the paths , , and meet end-to-end and so can be joined together to form a loop which is congruent to modulo . By doing this for all cycles, we see that can be expressed as a sum of loops (probably with different basepoints). Our earlier discussion shows that each of these loops lies in and then that the sum lies in , so as claimed. ∎
Let be a loop based at . A filling in of is a map with and .
If can be filled in, then in .
Let be a filling in of . Then
so , so . ∎