Algebraic topology II
Algebraic Topology
Spring 2021
Notes by Patrick Lei
Lectures by Francesco Lin
Columbia University
Disclaimer
These notes were taken during lecture using the vimtex package of the editor neovim. Any errors are mine and not the instructor’s. In addition, my notes are picture-free (but will include commutative diagrams) and are a mix of my mathematical style and that of the instructor. If you find any errors, please contact me at plei@math.columbia.edu.
Acknowledgements
I would like to thank Nicolás Vilches and Alex Xu for pointing out errors and typos in these notes. I would also like to thank Alex for graciously offering to teach everyone about the Dirac operator. Finally, I would like to thank Google for Gemini 3 Pro allowing me to upscale the original hand-drawn figures to Tikz.
1 Spectral Sequences
Our goal in this chapter will be to compute the homotopy groups of spheres, but we are not algebraic topologists so we don’t actually care about that. What we know from the basic theory is that if and , and this isomorphism is given by the degree. We also know that and is generated by the Hopf fibration . Note that this map is the attaching map of the -cell of .
There is an analogous decomposition . Then we obtain a fibration , which is an -fibration. Using the long exact sequence of the fibration, we see that .
Next, we may consider , the octonionic projective plane. This has a cell decomposition and from this we obtain a fibration and compute that . We should note, however, that is not the set of lines in because the octonions do not form an associative algebra. There is enough associativity to define some sort of but not for . In fact, we will prove that no space with the expected cohomology ring exists.
More generally, we can study as follows: Consider the mapping cone . Then the cohomology is
Then choose generators . We then have for some integer , called the Hopf invariant.
Example 1.0.1.
The attaching maps of all have . This implies that .
We know that if are homotopic, then because the mapping cones are homotopy equivalent. Then we obtain a homomorphism . Note that when is odd, we have by graded commutativity, and so .
Example 1.0.2.
If is even, there exists with .
Now given a pointed space define . This is somewhere between the product and the smash product, and so is a CW complex with a single cell in dimensions .
Exercise 1.0.3.
The attaching map of the -cell has .
Corollary 1.0.4.
admits a -summand. This is because surjects onto either or , and so it splits.
We have the following results:
Theorem 1.0.5 (Serre).
is a finitely generated abelian group. It has rank if or and is even.
Theorem 1.0.6 (Adams).
If with , then .
Remark 1.0.7.
This is related to the following. Suppose admits the structure of a division algebra. This is a bilinear map that is invertible for . Then . We will prove this result using K-theory.
1.1 The Simplest Case
Here, we will consider for . By the Freudenthal suspension theorem, we have maps
Therefore the group we need to compute is . The key strategy is to exploit the interaction between homotopy and homology. Here, we will consider the Hurewicz map
Theorem 1.1.1 (Hurewicz).
Suppose and is -connected. Then and .
Remarks 1.1.2.
When , is the abelianization of . Also, there is a relative version of the Hurewicz theorem.
Sketch of Proof.
We can assume is a CW complex with a single -cell and no cells in dimension . Then we can replace with , so we can write
By homotopy excision, we have
and this is exactly . □
Our strategy to compute will be to construct a space whose homology is . Recall that by definition, and so consider be a generator. Then is an isomorphism. Now we can turn into a fibration . Considering the long exact sequence of the fibration, we have
and therefore . We also have for . In particular, we have .If we can understand in some reasonable way, then we will be done.
If we turn into a fibration, then the homotopy fiber is now a . Now we have a fibration and we know the homology of both and , so the question is to compute from the information we have.
Question 1.1.3.
Given a fibration, is there a relationship between ?
Recall that in the case when this relation is given by the Künneth formula. Here, we see that as chain complexes, and so we reduce the problem to homological algebra. This gives us
Theorem 1.1.4.
Let be a principal ideal domain. Then there is a natural short exact sequence
and this sequence splits (but not naturally).
Example 1.1.5.
We can compute that in dimensions .
However, we note that the Kunneth theorem does not hold for fibrations in general. To see this, consider the Hopf fibration.
Example 1.1.6.
Consider the fibration . Both the base and fiber have nontrivial homology, but clearly the total space is contractible, so it has trivial homology. In both examples, is smaller than what we would get from Künneth.
1.2 Spectral Sequences
Definition 1.2.1.
A spectral sequence is a sequence , where is an -module and is a differential. In addition, we require that .
Remark 1.2.2.
Note that determines , but not in general.
Now assume that for so that .
Definition 1.2.3.
We say that , or converges to , if admits a filtration
such that .
Remark 1.2.4.
This says that is recovered by up to extension problems.
Example 1.2.5.
Consider a short exact sequence of chain complexes. Then we will see that there is a spectral sequence with and .
First, consider the long exact sequence in homology. If we consider the boundary homomorphism , we obtain a long exact sequence
Now define with . Then we have
as desired. Then we let .
This tells us that if , then . Also, if , then with a shift.
Theorem 1.2.6 (Serre).
Let be a fibration and let . Then there is a spectral sequence such that
- •
;
- •
.
Note here that we need to define what it means to converge in this setting. Also, note that is smaller than , so this formalizes the notion that is smaller than what we naively expect.
Remark 1.2.7.
We actually do not know what is, so it is impossible to compute the homology in general.
Fortunately, there is more structure:
- •
is bigraded.
- •
has bidegree .
- •
.
- •
for depending on .
- •
has a filtration with associated graded .
Example 1.2.8.
Consider the example . Then we know that and therefore the spectral sequence is
This gives us , and so has degree and thus it has to be zero. By degree reasons, we see that and , so . Then, the total space is a point, and so writing the -page
we see that and that . This recovers the usual computation of the homology of .
Example 1.2.9.
We want to compute . Here, we will use the fiber sequence . Here, we know that , so the -page looks like
Then the -page looks like
and the differential has degree , so . This implies that , and so the associated graded pieces are . We see that and each is an isomorphism. This tells us that for all .
Example 1.2.10.
We can enhance this example to . In this case, the first possible nontrivial differential is , which has degree . This gives us for degree reasons, so we can compute
because is an isomorphism.
Recall that our goal was to study . However, we cannot compute this yet, so we will need to introduce even more structure. To do this, we will need to do some homological algebra.
Considered a filtered chain complex, which is an abelian group with a map such that . To this, we attach a filtration such that .
Remark 1.2.11.
is a subcomplex of .
Now if is a topological space with filtration . Then if , we can choose , and this is a filtration.
Given and a filtration, we can associate two objects:
- 1.
, the associated graded complex. Then we may consider the homology of this complex, which is a graded object.
- 2.
Given a subcomplex , we have a map . The image of this is , so we get a filtration on the homology . So then we obtain a graded
Now we want to consider therelationship between the two. Note that the first one is bigger, because is a cycle if , while is a cycle if .
Theorem 1.2.12 (Leray).
There exists a spectral sequence such that
- •
.
- •
The spectral sequence converges to , or more precisely, .
Proof.
Consider the group
Notice that
so all the groups are as expected. For the differential, we define and in fact and thus has degree . Choose and choose a representative. Then implies that and therefore . Therefore defines an element in , so we set .
The things that need to be checked are that this is well-defined, , and that canonically. All of this is painful homological algebra and is omitted. □
However, we will need something even more painful. Now we will considered a filter graded chain complex , where has degree . Now for a filtration, we can consider the bigraded
Also, we note that is naturally filtered with .
Theorem 1.2.13 (Leray).
There is a spectral sequence such that
- •
.
- •
The spectral sequence converges to , where
Proof.
We write
The differential has degree . Note that it decreases the filtration by and the total degree by . For , choose a representative with . Then and therefore , so we set . □
Returning to topology, consider a filtration , which givs a filtration on . This now defines a spectral sequence by setting . The page of this is simply
and is .
The -page of the spectral sequence is
and is the connecting homomorphism for the triple .
The -page of the spectral sequence is
which is the associated graded of for the filtration .
Example 1.2.14.
Consider the cellular homology of . Set , the -skeleton. Then we note that is the homology , and now if is the cellular complex, the -page is precisely the cellular homology of and is the same as the -page. This proves that cellular homology is the same as singular homology.
Remark 1.2.15.
This discussion works for infinite filtrations as long as and has the weak topology induced by the filtration.
If you are interested in working through the pain1 of spectral sequences, the book A User’s Guide to Spectral Sequences is a good resource. In this course, we will not open this Pandora’s box.
Proof of Serre Spectral Sequence.
Consider and assume . For simplicity, assume that is a CW complex and is a fiber bundle. Now has a filtration by skeleta
and therefore has a filtration induced by pulling back . Therefore there is a spectral sequence converging to with
We want to compute the -page, so by excision we have
By contractibility of , so
Unfortunately, this identification is not canonical, which is why we need the assumption that is simply-connected. In general, for a path , transport (homotopy lifting) gives us a map depending only on the relative homotopy class. For example, if we consider , then the two paths between give different identifications. Therefore, we have a canonical identification
Finally, we note that is exactly the cellular boundary map , so , as desired. □
Remark 1.2.16.
The theorem holds provided the action of on is trivial, which means that the fibration is homologically simple.
Example 1.2.17.
Consider a sphere bundle . This is homologically simple if and only if it is orientable.
Even more generally, we need to use homology with local coefficients. Here, we will take
and will have components given by composing with transport. An alternative interpretation is to consider the universal cover . Now acts on and on , so we obtain modules over the group ring . Therefore we have
Now we will use groups on the edges of each page to compute . Degree reasons tell us that .
Proposition 1.2.18.
is the image of .
Similarly, we can consider the groups on the bottom of the page. This tells us that , which is a quotient of .
Proposition 1.2.19.
is the image of .
Returning to algebra, suppose we have a chain map that preserves filtration. This gives a morphism of spectral sequences such that are the maps on the associated graded objects and is a chain map such that the induced map on homology is .
Example 1.2.20.
An example here is filtered spaces with such that .
Note that the propositions follow by looking at the maps on spectral sequences induced by
Now consider the map . Note that is a quotient of and is a subgroup of , so we have , which is the analogue of .
Definition 1.2.21.
We call this the transgression map.
Proposition 1.2.22.
The transgression is given by .
Example 1.2.23.
Let be a closed smooth oriented manifold. Then there is a fibration , where is the unit sphere bundle. This is homologically simple, and then we have , which is multiplication by .
Remark 1.2.24.
We can prove this result if we know the relationship between and zeroes of vector fields.
Example 1.2.25.
Consider the fibration . If is -connected, then is -connected. Then are the first possibly nonzero homology groups. Now we consider the spectral sequence
and thus the map is an isomorphism for .
Remark 1.2.26.
We can interpret this very explicitly as , and this is the inverse of
1.3 Spectral Sequences in Cohomology
Theorem 1.3.1.
Let and . Then there exists and and has degree . Furthermore,
- 1.
There is a multiplication
- 2.
is a derivation, which means
- 3.
The multiplication on is induced by the one on .
- 4.
The multiplication on is compatible with the one on . This means that is the associated graded of the cohomology.
Example 1.3.2.
Consider . We want to compute , which will compute . Now the -page of the spectral sequence for cohomology looks like
and now because . But then
This tells us that the -page is
and therefore and . By the universal coefficients theorem, we see that .
Corollary 1.3.3.
For , we have and the group is generated by the suspension of the Hopf map.
Theorem 1.3.4.
and .
Proof.
Let . We know that . Now for general , consider the fiber bundle . Now assume that . Then the -page of the spectral sequence is
and therefore we see that by grading reasons. In particular, we see that , and in general we see that . Therefore we see that .
To conclude, we know that , so for some , we can choose a lift in . Because is torsion free and for degree reasons, the lifts satisfy the desired identities. □
Remark 1.3.5.
The ring does not determine . There are and with . For example, we can consider and (the intersection forms are different).
Example 1.3.6.
Let . Then and . However, we have
as groups, but not as rings.
To study this, we will consider the Stiefel manifold of -orthonormal frames in . Note that . In the simplest case, and , the unit sphere bundle of . Now we have , so we have a spectral sequence
Observe that a finite CW complex might not have even finitely generated homotopy groups. For example, we have .
Theorem 1.3.7 (Serre).
Let be a -connected CW complex such that is finitely generated for all . Then is finitely generated for all .
The key ingredients to the proof are:
- 1.
Show that for with finitely generated (resp. finite), then is finitely generated (resp. finite).
- 2.
Inductively, kill homotopy groups using fibrations. This generalizes .
Consider the Whitehead tower of , which is a sequence , where is -connected and is an isomorphism in for . Also, we require to be a fibration with fiber . For example, we had and . We have .
To construct the Whitehead tower, we work inductively. Suppose we have . Then is obtained by attaching cells to , so we have a map that is an isomorphism on . Turning this into a fibration, we take to be the homotopy fiber.
Lemma 1.3.8.
Consider a fibration such that . Then if are finitely generated, then is finitely generated for all .
Proof.
Consider the Serre spectral sequence. Then is finitely generated for all , so because is Noetherian, then must also be finitely generated. This implies that has a filtration by finitely generated objects, so it must be finitely generated. □
We will apply this to .
Proposition 1.3.9.
is finitely generated for whenever is finitely generated.
Proof.
Write . Now up to homotopy, we have
By Künneth, we may assume that is cyclic, so consider the fibration . Note that and is an infinite lens space, so only the inductive step remains.
Consider and assume is not finitely generated but is finitely generated for . But then is not finitely generated (it is the kernel of a map to something that is finitely generated), so is not finitely generated, a contradiction. □
Remark 1.3.10.
This holds more generally for classes for classes of groups called Serre classes, for example finite abelian -groups.
1.4 Homotopy groups of spheres
We know that is a finitely generated abelian group.
Theorem 1.4.1 (Serre).
The rank of is is or for even and otherwise.
The key computation is the rational homology . This is given by
This is true for . We will do the case when is even, and consider . Then the Serre spectral sequence in cohomology is
and then .
Lemma 1.4.2.
Let be -connected and suppose is finitely generated for all . Also, suppose that for odd. Then
Proof.
Note that , so . Then there exists such that , where generates . This tells us that is an isomorphism. Now if is the homotopy fiber of , then for , so for all (by finite generation), and therefore is finite for all . Now fits in the long exact sequence with , so has finite kernel and cokernel, and thus . □
Proof of Serre.
We may assume that is even. Consider . This is the th stage of the Whitehead tower. Consider the Serre spectral sequence
and note that is an isomorphism because . But then we see that , so the desired result follows. □
An “easy” generalization is
Theorem 1.4.3 (Cartan-Serre).
Let be -connected with finitely generated homology groups such that
Then is the number of generators in degree .
Example 1.4.4.
Consider .
Remark 1.4.5.
In general, does not imply that .
Now the point of rational homotopy theory is to compute using and extra information (for example Massey products).
We have computed the ranks of the homotopy groups of spheres, so now we will consider the torsion. The tools we have developed are enough to show that
Theorem 1.4.6.
Let . For any prime , the group has no -torsion for and the -primary part of is .
Corollary 1.4.7.
In the stable range, is a -group and is the direct sum of a -group and .
The idea of the proof is induction on the Whitehead tower with the fibration and . We can study this group one prime at a time because vanishes if and is very interesting if . Then there is a large gap in cohomology depending on . If we study the -cohomology of , then we may consider the fibration . Then we know that where , and so the Serre spectral sequence of the fibration is
and then and thus and for . For more detail, see Fuchs-Fomenko.
Now we will compute the -primary part of these homotopy groups. Consider the fibration and the next step . We want to compute the -primary part of . This means we need to understand . We can understand it as cohomology operations
1.5 Cohomolgy Operations
Definition 1.5.1.
A cohomology operation between is a natural transformation between the two functors viewed as .
Example 1.5.2.
Let be a ring. Then the cup product given by is a cohomology operation. Note that this is not a homomorphism in general.
Now and so natural transformations are just homotopy classes of maps , which form the group .
Example 1.5.3 (Bockstein homomorphism).
Consider a short exact sequence . Then we get a short exact sequence
of chain complexes, and this induces a long exact sequence in cohomology. Then the map
is a cohomology operation, called the Bockstein homomorphism.
An interesting case of this is the exact sequence . Therefore we obtain an interesting map
so there is an interesting element in . This is actually useful in the homotopy classification of -dimensional lens spaces . In fact, we can show that if and only if for some . For example, .
To show this, consider the map
This is well-defined up to a choice of sign. There is a standard generator , which is the image of the arc . Then , where is the Poincaré dual of . Then if we have a homotopy equivalence , we see that and then by naturality.
Example 1.5.4 (Steenrod square).
For all , we will construct a cohomology operation . Some properties are
- •
Steenrod squares are additive.
- •
We have
- •
is the Bockstein homomorphism for
- •
(the Cartan relation).
We will take these to be axioms for the Steenrod squares.
Theorem 1.5.5.
There exist unique cohomology operations with these properties.
Remark 1.5.6.
Define the total squaring operation by
Here, the Cartan relation says that is multiplicative.
Example 1.5.7.
Let . Then . Then
This implies that . In particular, we see that .
Proposition 1.5.8.
The Steenrod squares commute with the suspension isomorphisms .
This is interesting because has trivial cup products.
Example 1.5.9.
Consider with . Then is nontrivial. Consider the mappinc cone where , where is the -cell and is the -cell. Then we see that , so by dimension reasons, all cup products are trivial. However, . Otherwise, if is trivial, then .
Proof of Proposition 1.5.8.
Note that is obtained by
This implies that
Now we will consider the following question. How many pointwise linearly independent vector fields can there be on . When are there ?
Some basic obervations are the following:
- •
If is even, then , so any vector field has a zero.
- •
By Gram-Schmidt, we can assume that is orthonormal at each point.
Theorem 1.5.10 (Steenrod-Whitehead).
If , then has at most linearly independent vector fields.
Example 1.5.11.
If is even, then , so there are no linearly independent vector fields.
Corollary 1.5.12.
If is parallelizable, then is a power of .
Remark 1.5.13.
Adams gave a complete solution. We will see using -theory that if is parallelizable, then .
Sketch of Proof.
Let be the Stiefel manifold of orthonormal -frames in . Then we have the natural map
Now if are orthonormal vector fields on , then we get a section of given by
Now the question is reduced to that of the existence of a section of . We will discuss obstructions using . Recall that . Now consider
This induces a map . Now if , there is a cell decomposition of for which is the -skeleton. Now suppose there is a section of . Then is an isomorphism. After homotopy, we can make a cellular map, so . Therefore factors through and induces an isomorphism in degree (using -coefficients). In degree , induces a map . But now we have a commutative diagram
and now we see that cannot be an isomorphism. Computing using naturality, we see that works. □
1.5.1 Construction of Steenrod Squares
Here, we are following Chapter 4.L in Hatcher. The idea is that the cup product is commutative on . However, this is not commutative on the chain level, so will measure this failure. We work at the level of spaces.
Let be a pointed space and consider with the swap map . Now we consider the homotopy quotient of . For example, we know that acts freely on with quotient . Then we define . This has a projection to with fiber . Then the inclusion of at the basepoint of is a section, so set (doing this kills many unnecessary cells in low dimension).
Now given , we can associate such that
Then induces a map
But now given such , this maps to a sum of the form
Remark 1.5.14.
The assignments are functorial.
Now because , we only need to construct for the nontrivial
Here, we will set , and therefore we need to construct the map step by step. Consider the map
Then if we consider , this must be homotopic to by some homotopy . Now we can use to define
Because this is basepoint preserving, it descends to . But now is obtained from by attaching cells of dimension stricly larger than , so we can extend and obtain a map .
Remark 1.5.15.
Note that is uniquely determined because the map
is injective.
There is an alternative easier way. Consider and . This gives us a classifying map , and again let be the swap map on . Now we see that with homotopy , so now we have a map such that map to the same point. This map extends to and therefore to . This now gives us a map which is -invariant, and precomposing with gives us a map .
Theorem 1.5.16 (Cartan).
For any , we have .
1.5.2 Stable Cohomology Operations
Definition 1.5.17.
A stable cohomology operation (SCO) of degree is a map
for all that commutes with suspension maps, i.e. the diagram
Example 1.5.18.
The Steenrod squares and compositions of Steenrod squares are all stable cohomology operations.
Lemma 1.5.19.
Let be a stable cohomology operation. Then the diagram
commutes.
Proof.
The left and right squares are clear because they come from maps of spaces. At , we consider
Because is a cofibration, then . Then we may collapse the to obtain , and then we see that is given by , and both arrows in the composition commute with . □
This gives us the following slogan:
Stable cohomology operations commute with transgression.
Proposition 1.5.20.
Let be a stable cohomology operation of degree . If is transgressive, then so is (which means that ). If is represented by , then .
Proof.
The transgression map is given by
Then commutes with and . □
Remark 1.5.21.
If , then we can compose stable cohomology operations. This is a noncommutative algebra!
Theorem 1.5.22.
The Steenrod algebra
is generated as an algebra by .
This is a very complicated algebra. For example, if , we have the Adem relations
For a proof of this, see Hatcher.
Example 1.5.23.
For any , we have . For example, . We say that is decomposable. This means it can be written in terms of elements of lower degree.
Example 1.5.24.
The elements of the form are indecomposable. Consider the action on . Then we see that
but for . The idea is that for all .
We want to interpret stable cohomology operations using . Then if , we see that
But then stability tells us that , and if we consider the nontrivial , then we obtain a map
Then stability is equivalent to the fact that . Alternatively, we may use the suspension-loop adjunction to see that
Proposition 1.5.25.
For any , we have
We will now compute . We will use the fibration and hope to proceed by induction.
Example 1.5.26.
Let . Then we know . Then the spectral sequence is given by
Then we see that . But then , so we see that . Inductively, we see that and . In the end, we will see that
Definition 1.5.27.
A set (possibly empty) is admissible if . Then we will write .
Note that if is not admissible, then can be simplified using Adem relations.
Definition 1.5.28.
Define the excess of an admissible to be
For example, if , then . If , then .
Theorem 1.5.29 (Serre).
We have
Example 1.5.30.
We have . When , the generators are
When increases, the description becomes more and more complicated.
Theorem 1.5.31 (Borel).
The set are generators of . Now .
Example 1.5.32.
When , we have and has excess . Similarly, we see that and has excess . We can also see that , and this also has excess .
Theorem 1.5.33 (Borel).
Let be a fibration with and . Assume that
- 1.
is generated as an algebra by elements which are transgressive.
- 2.
The set is an additive basis for .
Then is a polynomial algebra generated by .
Proof.
Let be the Serre spectral sequence. We will construct an algebraic model for it. Set
where . Then define to satisfy
To do this, we set
This tells us that for all and thus all differentials are well-defined on . Of course, we observe that . Now there is a natural map given by
This induces a map of spectral sequences, so is an isomorphism. But now if is not surjective, then cannot kill anything not in the image. On the other hand, if it is not injective, then cannot kill anything in the kernel. Therefore the map is an isomorphism. □
Now we will compute . We know that is a fibration, so consider the Serre spectral sequence
Then we know that , so we find generators . In addition, we note that . Also note that is not there!
Theorem 1.5.34 (Serre).
We have the identity
1.6 Computation of Homotopy Groups
We will compute for large (the stable homotopy group). If we consider the fibration and , we want to compute . First, we will consider cohomology with coefficients. The Serre spectral sequence is given by
If we write , we can compute to be
| generator |
|---|
This now gives us the Serre spectral sequence for the next fibration
But this implies that and . By the universal coefficients theorem, this implies that . To obtain , we will consider the Bockstein exact sequence
This implies that , so by the universal coefficients theorem.
In the next case, we want to compute . We can consider and we want . In the Serre spectral sequence, we need to compute , where has degree . But this will tell us that , so by the universal coefficients theorem, we have
To compute the order, we will have to redo all computations with coefficients in modulo odd torsion. We should be able to obtain
Theorem 1.6.1 (Rokhlin).
We have the identity .
For example, we can compute . We have
By Cartan, we can compute . By Bockstein, we see that .
Remark 1.6.2.
We used Whitehead towers, but there is a dual approach using Postnikov towers. The Postnikov tower looks like
and satisfies the properties
- •
for ;
- •
is an isomorphism in for ;
- •
We have a fibration .
Hatcher uses Postnikov towers to compute , and if you are particularly masochistic, Mosher-Tanpora compute the stable homotopy groups up to . Unfortunately, this approach will get stuck eventually, so we cannot compute all of the stable homotopy groups of spheres.
2 Characteristic Classes and -theory
2.1 Obstruction Theory
Here is a very basic question: Can we extend maps?
Question 2.1.1 (Extension Problem).
Consider and . Does there exist a map extending ?
Question 2.1.2 (Homotopy problem).
Let such that are homotopic on . Can we extend this homotopy to a homotopy on ?
Note that this is a special case of an extension problem
Remark 2.1.3.
This is different from the homotopy extension property, which is given by and is the definition of a cofibration.
Question 2.1.4 (Lifting problem).
Let be a fiber bundle. If we have can we lift to ?
For the extension problem, we will assume that the action of on is trivial. A special case of this is if we have , can we extend it to ? If we attach a single cell , then we may consider the map . This extends to if and only if in . This gives us a map , so we obtain the obstruction cocycle .
Lemma 2.1.5.
extends to if and only if .
Lemma 2.1.6.
We have is a cocycle, which means .
Proof.
Assume . By taking the universal cover, assume . We need to show that , where . But this is given by the commutative diagram
Definition 2.1.7.
Let . Then define the obstruction class .
Theorem 2.1.8 (Fundamental theorem of obstruction theory).
Let . Then if and only if can be extended to .
Here, we allow ourselves to change on . Now suppose agree on . Then given an -cell , we can consider , and performing on the top and on the bottom gives an element in . This gives us , the difference cochain.
Remark 2.1.9.
Observe that if and only if is homotopic to with a homotopy fixing .
Lemma 2.1.10.
Given and , there exists such that and .
Proof.
In the center of the disk, simply attach a new disk with the desired . □
Lemma 2.1.11.
Consider and the obstruction classes . Then
Proof.
For simplicity, suppose only differ on . Given , we want to show that
Up to homotopy, we may assume that in Figure 2.1, the shaded part is and is a homeomorphism on the shaded component. In particular, only differ on the shaded part, so the number of components of the shaded part is .
□
Proof of fundamental theorem of obstruction theory.
Let . Then for some . But then there exists such that with . But then
so extends to .
In the other direction, assume with . Therefore , and
so . □
Now we will consider the relative case. Suppose is a CW-pair and we have . Can we extend this to ? Here, we obtain an obstruction class . We have a relative version of the fundamental theorem in this case.
Returning to the homotopy problem, consider such that . Can we extend this homotopy to ? We are considering the extension problem for
We are considering -cells of the form , where is an -cell in . Assume for simplicity that . Then the obstruction is exactly . Also, if , then .
Theorem 2.1.12.
Let agree on . Then if and only if are homotopic on , with the homotopy fixing .
Example 2.1.13 (Cohomology of ).
Consider by . Here, we can homotope such that is constant because for . Surjectivity is obvious from the previous discussion, so we will prove injectivity.
If , then is nullhomotopic. Because for , is nullhomotopic.
2.2 Primary obstruction class
We will change our notation. Denote the fiber bundle by . In this new notation, we want to extend a section to . If is a cell, then . We have and we want to extend this to . Now we have the map . We know that . Now we know extends to if and only if , , and the obstruction class is zero if and only if can be extended to .
Now we assume that is -connected. Therefore there exists a section with obstruction class .
Theorem 2.2.1.
Given , we have . In particular, this is an invariant of the bundle!
Definition 2.2.2.
Consider a fiber bundle with being -connected. Then the primary obstruction class is defined by be .
Theorem 2.2.3.
The fiber bundle always admits a section over and it admits a section over if and only if .
Proof of Theorem 2.2.1.
First we note that a homotopy of does not change . The homotopy extension property holds for sections where is a CW pair with and a homotopy of sections. Here, we can extend to .
Now suppose . Of course, on . We now show that if on , then on for . To see this, the homotopy extension property tells us that we may assume on after extending the homotopy. Then, we have . But we see that , so on . We now obtain such that on . We may assume that on . Then we note that , so and thus . □
The key property is that if we have and consider the pullback bundle , then . This tells us that is a characteristic class, or in other words, a natural cohomological invariant. Consider the basic example: Suppose is an oriented bundle. Then is -connected, so we have an obstruction class .
Definition 2.2.4.
This obstruction class is called the Euler class of . Then if and only if has a section over .
Here we have a special case. If is an oriented vector bundle and is the associated sphere bundle . Then is the Euler class of the associated sphere bundle. Therefore if and only if there exists a nonvanishing section of over .
Theorem 2.2.5.
Let be a smooth oriented compact manifold. Then is .
Corollary 2.2.6 (Poincare-Hopf).
admits a nowhere vanishing vector field if and only if .
Proof.
Note that (one for each top-cell). To compute we can use a vector field with only isolated zeroes. Now we may define the local degree at to be the degree of the map defined by . The local degree behaves like in Figure 2.2
and therefore by definition. Now we simply need to construct an explicit nice vector field. Because is smooth, it admits a smooth triangulation.1 The vector field is constructed by placing a zero at the center of every face and then for the center of a given positive dimensional face, the vector field points outward towards the vertices of that face. But then we have
Remark 2.2.7.
Let be a smooth vector bundle with compact. Then can be interpreted as follows. Choose a generic section . Then is a smooth manifold of dimension , and so we can write .
We will now relate the primary obstruction class and transgression. We will consider the Serre spectral sequence with coefficients in . We know we have the transgression map and the fundamental class . This is defined by for all .
Theorem 2.2.8.
The primary obstruction class is given by .
Proof of this is given in a series of exercises in Fuchs-Fomenko 23.5.
Now we will consider the Thom isomorphism. If is a rank vector bundle. Then we may associate the disk bundle and the sphere bundle . Now we define the Thom space to be (equivalently, the one-point compactification of ).
Theorem 2.2.9 (Thom isomorphism).
Suppose is an oriented vector bundle. Then there exists a unique such that is the generator of (given by orientation). We call the Thom class. Furthermore, we have
is an isomorphism for all . Finally, the map
sends the Thom class to the Euler class .
Proof.
Consider the Serre spectral sequence for . Then the Euler class is the Thom class by the prior discussion about transgression. Alternatively, we may construct the Thom class locally and patch, and this can be found in Chapter 10 of Milnor-Stasheff. □
2.3 Vector Bundles
We may consider the functor
We may also consider the functors of real vector bundles and oriented real vector bundles. This functor is representable!
Theorem 2.3.1 (Brown representability).
Let be a functor. If
- 1.
For with , then is an isomorphism;
- 2.
If is a union of subcomplexes and retrict to the same element in , then there exists restricting to ,
then there exists a CW complex and such that
is a bijection.
Note that for , the condition of being homotopy invariant is nontrivial to decide. We want such that . The idea is that for , we need .
Proposition 2.3.2.
Let be a CW complex (or in general is paracompact). Then if is a vector bundle over , we have .
Recall that is paracompact if is Hausdorff and any open cover admits a locally finite refinement . This is equivalent to every open cover admitting a partition of unity subordinate to it.
Remark 2.3.3.
We may also assume the indices are countable.
Example 2.3.4.
Any paracompact admits a Hermitian metric. To see this, simply trivialize over the charts and then glue using our partition of unity.
Now we note that if is trivial over and , then is trivial. This implies that any is trivial. Fix trivializations , . These do not necessarily agree on , but we simply consider and change by to get the desired trivialization.
The second observation is that if then there exists an open cover such that is trivial. To see this, we simply trivialize over and then glue the trivializations.
Now we return to the proof that is homotopy invariant. Given a partition of unity subordinate to (with trivial), we write . Then define to be the graph of . We know for some , so lifts to . We know is trivial on , so we have
Then we know is , so using local finiteness of the partition of unity, we obtain the desired result.
We now want to find a concrete space that represents this functor. Define the Grassmannian
of -planes in . This has a tautological bundle given by . Of course, if , we know , so we define . Because the tautological bundles are well behaved under this, we obtain a tautological bundle .
Theorem 2.3.5.
The bundle is a universal bundle. This means that , where .
Remark 2.3.6.
We can also write as the classifying space of .
Proof of Theorem.
Let . Then if and only if there exists which is linear injective on each fiber . Given , choose a countable cover with trivial and a partition of unity subordinate to . Then define
and consider the map . This defines .
Now we need to show that if that are injective on fibers, then there exist linear injective on fibers at each time. We define
This is injective at each and moves to the odd entries. Similarly, we can move to the even entries, so we can write . □
A characteristic class is a natural transformation . By the Yoneda lemma, characteristic classes are in bijection with the cohomology of . We now study basic facts of the Grassmannians:
- •
There is a transitive action of on . The stabilizer of is .
- •
Clearly we have .
- •
The Grassmannian has a nice cell decomposition generalizing .
Example 2.3.7.
Consider the Grassmannian of -planes in . Then the interior of cells is described by dimensions of intersections with . The cells are given by:
- -cell:
-
This is just corresponding to .
- -cell:
-
This is the set corresponding to .
- -cells:
-
These are corresponding to and corresponding to .
- -cell:
-
This is given by corresponding to .
- -cell:
-
This is everything else, which corresponds to .
For an alternative description, a plane corresponds to the row reduced echelon form of a matrix. Then we have the following correspondence:
- -cell:
-
This is the matrix .
- -cell:
-
This is the matrix .
- -cells:
-
These are the matrices .
- -cell:
-
This is the matrix .
- -cell:
-
This is the matrix .
These cells are called Schubert cells. These behave nicely with the inclusions and has cells. The number of -cells is given by the number of partitions of into at most integers that are each at most . The cells are of even dimension, so we can compute the group structure on the cohomology, but not the ring structure.
Theorem 2.3.8.
The ring struction on cohomology of the Grassmannian is given by , where .
Definition 2.3.9.
The characteristic classes of associated to the are the Chern classes . By convention, for .
Proof.
We need many fibrations. Consider the spaces of -planes in and of -flags in . It is clear that also parameterizes ordered -tuples of orthogonal complex lines. Of course there is a fibration . We can also understand inductively as well.
Now we will show that , where . Each is the pullback of the generator of under the maps . We will use the fibration . By induction, we have a fibration
and . Now we use the Leray-Hirsch theorem, but first we need to show that . This is because we have the map given by and this restricts to the generator of udner . Therefore, is a free -module with additive basis . Now we observe that
is a homotopy equivalence by Hurewicz.
Now we need to consider the fibration . This induces a surjective map in cohomology, so by Leray-Hirsch, is a free -module with basis and thus is injective. The map sends a -tuple of lines to their sum. We also know that is -invariant, so on cohomology, we have an injection
where is the th elementary symmetric polynomial. But then by a dimension count, this injection is surjective. □
Theorem 2.3.10 (Splitting principle).
Let be a line bundle. Then there exists such that is a direct sum of line bundles and is injective.
This gives us the following slogan:
When working with Chern classes, we can assume is a direct sum of line bundles .
Proof.
Set . Then the fiber over is the space of -flags in . Therefore we have a fibration . Now has a tautological splitting, where over the point we have the vector space . This pulls back as a direct sum of line bundles. □
Now here are some properties of Chern classes:
- 1.
For any , we have .
- 2.
If is the tautological bundle over , then and thus .
- 3.
For vector bundles , we have . Equivalently, if we define the Chern class , then .
Proof.
We prove this for the universal bundles. We have the bundle over , and this is given by a classifying map . Now under the map , we know pulls back to a direct sum of tautological bundles, and the same is true for pulled back to . Therefore, we have
In fact, the three properties determine the uniquely.
Remark 2.3.11.
- •
If is the trivial bundle, then .
- •
If is the complex conjugate bundle of , then . Note that this is also the dual bundle.
Example 2.3.12.
We have . To describe this, we consider the tautological bundle . Clearly , where the perp is taken in . Therefore, and . This gives us
and therefore
We will now consider Chern classes as obstructions. Let be a rank bundle. Then we can take orthonormal frames to obtain . Then sections of are -tuples of sections of which are orthonomal at each point.
Lemma 2.3.13.
We have
Therefore, the first obstruction to finding a section of over the skeleton lives in . In fact, this is the Chern class. The Chern class is the obstruction to finding linearly independent sections over the -skeleton. Now observe that . Therefore, if , we have .
2.4 Real Vector Bundles
Now consider a vector bundle .
Lemma 2.4.1.
For , we have . In addition,
Therefore we obtain obstruction classes or . This is an obstruction to finding linearly independent sections on the -skeleton. Reducing everything modulo , we have
Definition 2.4.2.
The Stiefel-Whitney classes of are defined by .
Here are some facts about the Stiefel-Whitney classes:
- •
If is the tautological class, then .
- •
These correspond to , where . Note that it is harder to compute the cohomology in this case than in the complex case.
- •
We have if and only if is orientable.
- •
If and is orientable, then .
- •
They are related to on (and in fact this is the definition given in Milnor-Stasheff).
Now we will assume that our real vector bundles are oriented. Recall that is the space of -planes in . Then we set to be the space of oriented -planes in . There is a double cover . Recall that every oriented bundle has an Euler class!
Remark 2.4.3.
Suppose is an oriented odd-dimensional vector undle. Then the map reverses orientation on the sphere bundle, so and thus .
Recall that varies greatly depending on the parity of . We should obtain the same thing for the classifying spaces.
Theorem 2.4.4.
Over , we have , where . In the even case, we have .
We are mostly interested in even-dimensional manifolds, so we will mostly work in the second case.
Definition 2.4.5.
The characteristic classes corresponding to are called the Pontryagin classes of .
Concretely, we can define the Pontryagin classes in terms of the Chern classes. If is a real vector bundle, we may take the corresponding complex vector bundle. Then we have
Remark 2.4.6.
Recall that is always isomorphic to its dual. Complex conjugation gives an antilinear involution, so is -torsion.
Example 2.4.7.
We will compute the Pontryagin classes of .
Lemma 2.4.8.
If is already a complex vector bundle, then as complex vector bundles.
Proof.
Define by the complex structure on . Then the map is a -linear isomorphism. □
In particular, if is complex, then the total Pontryagin class can be computed using
For , we have
so we obtain
Now we will consider Pontryagin roots. The splitting principle for says that can always be thought of as a direct sum of oriented -plane bundles analogously to Chern classes. Note that an oriented -plane bundle is the same as a complex line bundle (because rotates by counterclockwise). In the case where we actually have , we obtain
These have Chern roots , so we have
This tells us that the are the Pontryagin roots. Also, , so .
Remark 2.4.9.
This is related to the fact that has an -invariant square root, called the Pfaffian.
2.4.1 Relation to Steenrod problem and oriented bordism
Let be a closed smooth manifold and .
Question 2.4.10.
Is there an oriented smooth manifold and continuous such that ?
Definition 2.4.11.
A pair is singular manifold representing .
Now define
with addition given by disjoint union. In addition, we set . This is well-defined, and in fact there is a map
The group is called the -th bordism group. Clearly is functorial by postcomposition.
Remark 2.4.12.
The functors give a generalized homology theory. Note that is very complicated. These are given by
Theorem 2.4.13.
If , then the signature of is .
Example 2.4.14.
In , .
Lemma 2.4.15.
If is symmetric and nondegenerate and is isotropic and , then the signature of is .
Remark 2.4.16.
Actually, it is easy to see that .
Lemma 2.4.17.
Let be the inclusion onto the boundary. Then for all .
Proof.
This is given by because in the exact sequence
we have , so . □
Proof of Theorem.
Consider the commutative diagram
We know that is isotropic in , so we need to show that it is half-dimensional. But then under identification. But then is dual to , so and thus is half-dimensional. □
Recall that a spectrum is a sequence . Then we have the associated homology theory
where the colimit is over the sequence of maps given by suspension followed by .
Example 2.4.18.
Consider and let be adjoint to . Then the associated homology theory is .
Thom described the spectrum representing . Nowadays we call this the Thom spectrum . Here, we write , where is the oriented Grassmannian. Then there is a natural map
In particular, we have . Then we have a map , but in the homework we prove that .
Theorem 2.4.19 (Thom 1954).
The spectrum represents the homology theory .
As a consequence, we have . In particular, this is finitely generated. Tensoring with , we may use techniques of Serre to prove that
Therefore is finitely generated with rank when and the number of partitions of when .
Remark 2.4.20.
The generators of are , where is a partition of . Alternatively, we may define
This is called the -Pontryagin number. Then the map
is an isomorphism after tensoring with .
Recall that the signature gives a homomorphism . Therefore it is a linear combination of Pontryagin numbers. We may compute exactly which combination using the Hirzebruch signature theorem. Later, we will prove this as a consequence of the Atiyah-Singer index theorem. We have
- 1.
.
- 2.
.
- 3.
.
- 4.
.
An application of this formula is in Milnor’s 1956 paper On manifolds homeomorphic to the -sphere.
Theorem 2.4.21 (Milnor 1956).
There exist which are homeomorphic but not diffeomorphic to .
The idea to distinguish smooth structures is this. We know . Then because there exists such that . Then we have an intersection form on . In particular, the signature of is an integer and have a number .
Lemma 2.4.22.
The number is is independent of the choice of . Therefore is an invariant of .
Proof.
Consider two bounding manifolds . Then we can glue to obtain a closed manifold . Then and . Then the Hirzebruch formla gives us
and now basic arithmetic gives us the desired result. □
Remark 2.4.23.
This is a diffeomorphism invariant of . Then because bounds and gluing them together gives .
Examples 2.4.24.
We will give examples of homeomorphic to but not diffeomorphic to it. We will construct these as sphere bundles of oriented rank bundles over . Consider . These can be described using clutching functions. If we take , and we know . Therefore it suffices to describe the gluing along the equator, which gives us a map . Now oriented vector bundles are described by .
Very explicitly, we can identify with the unit quaternions and . Then is defined by
For example, the trivial map is given by . The unit sphere is . Similarly, gives and gives us the unit sphere . Now for odd, we will choose such that . Then define to be the unit sphere bundle for .
We will see that is homeomorphic to but . Thus is homeomorphic but not diffeomorphic to . First, we will write an explicit with exactly one maximum and minimum. Then using gradient flow, we have via some . Because extends radially to , we have a homeomorphism . To compute , note that bounds the unit disk bundle of .
2.5 Topological K-theory
Let be a finite CW complex (more generally a compact Hausdorff space). Then let denote the set of isomorphism classes of (complex) vector bundles on .
Remark 2.5.1.
The rank may vary on each component of .
Then is a unital semiring with operations and unit the trivial line bundle .
Remark 2.5.2.
Unlike , is not cancellative. This means that if , then in general.
Example 2.5.3.
Consider . However, .
We may define . This only works because is cancellative, but if we modify the condition to be if for some , then this defines the Grothendieck ring.
Definition 2.5.4.
Define the K-theory ring to be the Grothendieck ring of .
This defines a functor from .
Example 2.5.5.
We know that , so . This is because vector bundles on a point are the same as vector spaces.
Thus if we consider the inclusion , we obtain the (virtual) rank map . This is locally constant. Note that if is finite, then given , there is such that for some . Here are some consequences:
- 1.
If , then we can write for some . Here, if , then we choose , and thus we get .
- 2.
If and , then if and only if they have the same rank everywhere and are stably equivalent (which means that for some ).
To see this, we have for some . Then we simply add such that . Now we have , so we are done.
Now we would like to throw away the information about stably equivalent bundles. If is a pointed space, we define the reduced K-theory . This depends only on the component of and is a non-unital ring.
Proposition 2.5.6.
We can identify with vector bundles on up to stable equivalence.
Now stabilization corresponds to the inclusion , so a bundle up to stabilization has classifying space . Therefore, we have
Corollary 2.5.7.
.
Remark 2.5.8.
This is false for noncompact spaces in general.
We may also take the maps
Now we define . Then we have a fibration , where . We also have fibrations
Because for , we see that is contractible.
Example 2.5.9.
It is easy to see that
Similarly, we have
If is the tautological class (here, take ), then this is generated by . This is because . Therefore is generated by . The map is given by . Now that . In particular, . Thus we have .
Now we would like to extend this to a generalized cohomology theory.
Lemma 2.5.10.
Let be a finite CW pair. Then is exact (because is a cofibration).
Because is a cofibration, we know that . Now the exact sequence extends to
by our discussion of cofiber sequences from last semester. This gives us a long exact sequence
Now we can manually define a cohomology theory as follows:
Definition 2.5.11.
For , define . Analogously there is an unreduced theory . This gives .
Example 2.5.12.
We have .
The K-groups satisfy a miraculous identity:
Theorem 2.5.13 (Bott periodicity, version B).
For all spaces , .
Therefore is -periodic and we will consider as the interesting cases. The original statement of Bott periodicity is
Theorem 2.5.14 (Bott periodicity, version C).
The homotopy groups are -periodic. Therefore is is odd and if is even.
The most useful form of this result is
Theorem 2.5.15 (Bott periodicity, version A).
For all spaces , we have . Here, is the pullback of the tautological bundle .
Concretely, we have a map
Francesco says that proving this result is difficult online, so we only give a sketch.
Sketch of proof.
We will use generalized clutching functions. Choose a bundle . This gives a bundle . But of course is contractible, so we can glue them together using some . Thus , so we obtain a class .
The key point is that stably, any bundle on has the form . □
Example 2.5.16.
We can represent as .
Proof of version B.
Consider the long exact sequence
Therefore is surjective because . This argument also works for , and thus
By version A, we see that . If if , , then its restrictions to both are trivial. Restricting to , we require that . Restricting to , we see that , so . Therefore the map given by is an isomorphism. □
Corollary 2.5.17.
The group is generated by and .
Now we want to consider in the noncompact case. The options are
- 1.
The Grothendieck ring of ;
- 2.
The set of maps ;
- 3.
The limit for compact .
Unfortunately these are not the same. We are interested in the case where is Hausdorff and locally compact. For example, if is a CW complex such that every point is contained in only finitely many cells or if is a vector bundle to a compact Hausdorff space, then is Hausdorff and locally compact. In our case, the one-point-compactification is Hausdorff.
Definition 2.5.18.
If is locally compact and Hausdorff, define .
This is a (unital) ring and is functorial under proper maps. Also, if is open and relatively compact with , we obtain a map . In particular, we obtain a map . Then we have
where ranges over the relatively compact subsets of . Of course, if is compact, we recover the original -theory.
There is an alternative point of view in terms of complexes. Let
be a complex of bundles over . We define the support to be the locus where the sequence is not exact.
Example 2.5.19.
If we consider the bundle , then the support of this complex is .
Define to be the homotopy classes of complexes with compact support, where if there exists with compact support and . Now this contains the set of complexes with empty support.
Proposition 2.5.20 (Segal).
If is Hausdorff and locally compact, then as rings.
The basic idea is that if , then the dimensions of are equal at , so we can consider the complex where is a chosen isomorphism near and extended arbitrarily with compact support to the entire .
Remark 2.5.21.
We have the inclusion . Therefore, the multiplication is well-defined.
Remark 2.5.22.
The sequence in corresponds to .
Example 2.5.23.
We have and this is generated by . The idea is to tensor multiple copies of , and the generator is the Koszul complex
Now recall the Thom isomorphism in cohomology. If is oriented, then is an isomorphism. The Thom class restricts to the generator of for each fiber. In K-theory, if we have , then we can use define the Koszul complex of . Restricting to a fiber , we obtain the Koszul complex of . In particular, the support is the zero section of , which is . Thus if is a compactly supported complex on , is a compactly supported complex on . Therefore we obtain a map .
Theorem 2.5.24.
This is the -theoretic Thom isomorphism.
Proof.
We can prove this locally using Bott periodicity and then patch to obtain the global result. □
Remark 2.5.25.
The Thom isomorphism holds under weaker assumptions. This is when admits a structure by Atiyah-Bott-Shapiro.
Now suppose is a finite CW complex with only even-dimensional cells. Now we have an exact sequence
By induction, we can obtain . Here, we note that and is just , so this is very easy. In particular, we have . This can also be done using the Leray-Hirsch theorem, and in fact the splitting principle holds. For all , there exists such that is injective and is a direct sum of line bundles.
2.6 K-theory and cohomology
Now we will discuss the relationship between K-theory and cohomology. For a space , we have the two rings . Can we find a morphism between them? The natural idea is to consider characteristic classes. We can attempt to use the total Chern class, but this is not a homomorphism!
Definition 2.6.1.
Given a vector bundle , where is a finite CW complex, define the Chern character by
Here are the Chern roots of . We can actually rewrite this as
If we write , where is the Newton polynomial and are the elementary symmetric polynomials, we obtain
Proposition 2.6.2.
The Chern character defines a homomorphism .
Proof.
This is obvious for line bundles because , and now we are done by splitting. □
Theorem 2.6.3.
The Chern character is an isomorphism for any finite CW complex .
Note that this result is false over the integers. In addition, the map
is an isomorphism.
Proof.
We induct on the skeleta. This is obvious for the case of a point. So now consider the long exact sequences
Now we need to show that the maps on the relative groups are isomorphisms by the five lemma. Because , we need to show this for spheres. In general, this is because
commutes. Returning to the case of a sphere, we have
and also
is the generator. Now we simply need to show that are isomorphisms for . □
Remark 2.6.4.
The torsion of and can be very different. For example, we have , while .
In general, K-theory is very hard to compute. One can compute it using equivariant K-theory, or we may use the Atiyah-Hirzebruch spectral sequence. Here, if is a finite CW complex, then we have a spectral sequence with
In fact, this is true for any generalized cohomology theory.
Corollary 2.6.5.
The Chern character actually restricts to which is an isomorphism onto .
An application of this2 is the following: If is given by , then . This is because , so .
Theorem 2.6.6 (Borel-Serre).
If admits a complex structure , then .
Proof.
Suppose we have a complex structure. Then , so and thus . We know has no complex structure by Wu’s theorem. □
Remark 2.6.7.
When , then is a complex manifold, so we have holomorphic charts. When , we can construct using the octonions (viewing in the purely imaginary octonions), and the problem of whether admits a complex structure is open.
Here is an important point. For a vector bundle on a finite CW complex, we have a diagram which does not commute:
On one hand, we have and on the other hand we have . In general, ! Assume there exists such that . Formally, we obtain . If has Chern roots , then we have . On the other hand, . Applying the Chern character, we have , and thus
Definition 2.6.8.
The Todd class is defined by
Later, when we do index theory, this will be very important.
Remark 2.6.9.
We have the identity , so this is the exponential generating function for the Bernoulli numbers.
Theorem 2.6.10 (Adams).
If there exists with , then .
We will interpret using K-theory. Consider the mapping cone . then we have an exact sequence
and here we have . Then because , we can write .
The tool we will use to study this are Adams operations. These are natural transformations satisfying the following properties:
- 1.
For a line bundle, .
- 2.
For any , .
- 3.
If is prime, then . This means that for some .
The way we will define these is by guessing a formula using splitting and then write it in an invariant way. We know that if , we want
Introducing a formal variable , this becomes . Then we obtain
so we have
and thus we may define
The properties are easily checked using the splitting principle! For example, the third property is Fermat’s little theorem. Now this induces .
Proposition 2.6.11.
The map is multiplication by .
Proof.
Observe that is given by . Now commutes with multiplication, so we simply need to compute this for . But now we have
where . Now assume the result is true for . Then is the Bott periodicity map, so because commutes with multiplication, we obtain the desired result. □
Proof of Adams’ theorem.
Let satisfy . Write . Now we have an exact sequence
and this is given by . Then we know . For every , we have and for . Because multiplication of Adams operations is commutative, we have . Therefore, for all , we have
Because , we know
and thus is odd. Setting , we have
and thus , which implies . □
Corollary 2.6.12.
If is parallelizable, then . This is equivalent to being a division algebra. This means that there exists that is continuous and linear in each variable such that implies that one of . Thus .
Sketch of proof.
If is a division algebra, then it is a division algebra with unit. Next, it is clear that is a unital -space. But now this implies and then there exists with Hopf invariant . □
To go from being a divison algebra to being an -space, let be the multiplication on . Then the multiplication on is simply . To go from being parallelizable to being an -space, choose an orthonormal frame of . Then for all , is an orthonormal basis of . We may assume that at , this is , so we obtain a matrix with . Now the -space structure is and this is clearly unital with unit .
Remark 2.6.13.
Adams used similar techniques to provide sharp upper bounds on the number of linearly independent vector fields on . To construct the vector fields, we need to use the theory of modules over Clifford algebras. The Clifford algebras are all associative, and the first few examples are . The category of modules over these are periodic, and this is some sort of algebraic Bott periodicity.
3 The Atiyah-Singer Index Theorem
This is a result that relates analysis and topology. First, we will discuss the relation between analysis and -theory.
3.1 Fredholm Operators
Let be a linear map. Then define the index . In finite dimensions, this is just
by rank-nullity. In infinite dimensions, things are more interesting.
Definition 3.1.1.
Let be Hilbert spaces and be a bounded operator. Then is Fredholm if , is closed, and . Now we may define the index .
Remark 3.1.2.
For Hilbert spaces, the condition that is closed is automatic. This is a consequence of the open mapping theorem.
Example 3.1.3.
Consider be the left shift . Then . Similarly, we may consider the right shift given by , and .
Theorem 3.1.4.
The space is open and the index is locally constant.
Remark 3.1.5.
If is a family of operators, then the index is constant, but of course the dimensions of the kernel and cokernel can vary wildly. Thus, index should be a topological quantity in some sense.
Lemma 3.1.6.
Let be invertible. Then if has small enough norm, is invertible.
Thus being invertible is an open condition!
Proof.
Write . Then has small norm, so we can write
and this sum converges. Therefore we have
Proof of Theorem 3.1.4.
Let be Fredholm. Then we know and , where and . Now we can write in the block matrix form
Now suppose has small norm. Then we have
and is invertible. Then by Gaussian elimination using
we see that
so we have
Theorem 3.1.7 (Atiyah-Jänich).
Let be a separable complex Hilbert space. Then there is a natural bijection
for all finite CW complexes .
Remark 3.1.8.
Let be Fredholm. Then is also Fredholm. In particular, is naturally a semigroup and the map is a map of semigroups.
Corollary 3.1.9.
There is a weak homotopy equivalence .
Remark 3.1.10.
The map corresponds to the index.
To construct the map , note that an element of is a family , where . Then because is compact, there exists a finite dimensional such that for all . Therefore , so is a vector bundle. Now we set
Note that the vector bundle has dimension
The key point in the proof of Atiyah-Jänich is the fact that the space of bounded invertible operators on is weakly contractible. This is called Kuiper’s theorem. Note that this is in contrast to , which is not contractible.
Definition 3.1.11.
A bounded operator between Hilbert spaaces is compact if it sends bounded sets to precompact sets.
Example 3.1.12.
Any finite rank is compact. Also, the operator
is compact.
Theorem 3.1.13.
Let . Then if is Fredholm and is compact, then is Fredholm and .
Sketch of proof.
We want to show that , and this is the same as the unit sphere being compact. Here, if and , then the sequence has a convergent subsequence, and thus the unit sphere is compact.
To show that the cokernel is finite dimensional, we use Schauder’s theorem. Here, if is compact, then is also compact. □
3.2 Index and topological invariants
The goal of this section is to prove that if is a closed oriented smooth manifold, then are naturally interpreted as indices of differential operators. Recall the de Rham complex of -forms
This has cohomology . Note from this point of view that a cohomology class does not have a preferred representative.
A goal of Hodge theory is to find a preferred representative when is equipped with a Riemannian metric . Intuitively, each has an inner product
so we obtain an norm. Now a good candidate for the preferred representative of a cohomology class is the shortest element! It is not obvious that this even exists.
Now let be an oriented real Euclidean vector space of dimension . Then we define the Hodge star
to satisfy the property that for any oriented orthonormal basis , we have
In a more invariant manner, we have . Now if , then is the inner product . On a manifold, if , we obtain
Returning to the de Rham complex, we would like to write an adjoint . We set
Proposition 3.2.1.
The morphism is the adjoint of , so if , then
Proof.
Consider . Then we have
Then the resired result follows from Stokes, which gives us . □
Now suppose that minimizes in . Then any other element differs by an exact form , so we have
so we must impose . In particular, we must have .
Definition 3.2.2.
The Hodge Laplacian is the operator . Note that .
This satisfies the following properties:
- 1.
is the negative of the usual Laplacian .
- 2.
The operator is self-adjoint. Here, note that is self-adjoint.
- 3.
is nonnegative. Here, note that .
- 4.
We have on .
Theorem 3.2.3 (Hodge).
There is an orthogonal decomposition
Here, is the space of Harmonic forms. In addition, we have , and in particular, each cohomology class admits a unique harmonic representative.
Remark 3.2.4.
Note that preserves the natural grading on . Therefore, we may consdier , so we may consider its index. Therefore we have
Assuming the Hodge theorem, we therefore obtain . In general, the Atiyah-Singer index theorem tells us that this is .
Now let be oriented. We have an intersection form . For de Rham cohomology, we may write
Now we will realize the signature as the index of with respect to a different splitting of . Set . Then it is easy to see that , so we write . Also on and anticommutes with . Therefore, we can consider
This has kernel containing for , where as well as such that . On the other hand, the cokernel consists of forms of the form for with and such that . Therefore, we may consider only the forms in , and we see that , where is the space of self-dual forms and is the space of anti-self-dual forms. But now we see that
when , and similarly we see that is negative definite on . Because , we see that . The Atiyah-Singer index theorem will express this index in terms of Pontryagin classes, and this will give us the Hirzebruch signature formula.
3.3 Elliptic Operators
Let be a smooth manifold and be vector bundles over .
Definition 3.3.1.
A differential operator (with coefficients) of order is an operator
that looks in charts , where , like an operator of the form
where is a multi-index and is a matrix smooth in . Here, we write
and is the -norm.
Definition 3.3.2.
A differential operator of order is elliptic if for all and and we substitute , the polynomial
is an isomorphism.
Example 3.3.3.
The Laplacian of order is elliptic. Here, we have
and thus is elliptic.
Example 3.3.4.
The wave operator is not elliptic. Here, note that
and this vanishes for .
Example 3.3.5.
Consider acting on -valued functions. Then we have
and this is an isomorphism for .
More intrisically, consider . Then is elliptic at if and only if for all with , and with such that , we have . Here, we can compute .
The key point is that if is elliptic, then solutions to are very nice. For example, if and , then . Similarly, if and , then . On the other hand, this is false for the wave equation. The correct functional analysis setup for this is Sobolev spaces. Now consider a Euclidean vector bundle on a Riemannian manifold. Then fix a compatible connection on and . For , we will define
Definition 3.3.6.
The -norm of is defined by
The completion of with respect to this norm is the Hilbert space .
Example 3.3.7.
The space contains functions that are continuous and are well-behaved. On the other hand, contains discontinuous functions. Also, functions in are not necessarily continuous.
Remark 3.3.8.
If is compact, then the space is independent of the choices made as a topological vector space.
Lemma 3.3.9.
Let be commpact and be a linear differential operator of order . Then for all there exists such that
In particular, this means we can take , and this is bounded. There is another estimate, which is only true for elliptic operators.
Theorem 3.3.10 (Elliptic estimate).
If is elliptic, then there exists such that
Remark 3.3.11.
The is there to take care of . Also, this estimate is false for norms.
We will simply give the idea of the proof in the simplest case . Recall the Fourier transform
This is an isometry up to factors of . Under this correspondence, the differential operator corresponds to . Now the Sobolev norm corresponds to the norm
Finally, corresponds to . Therefore, we see that
and thus if , we see that . This gives us the result
In general, is a degree polynomial in that only vanishes at . This gives us an estimate of the form
Here are some nice results about Sobolev spaces:
- Sobolev embedding theorem:
-
If , we have .
- Rellich compactness:
-
If , then is a compact operator if is compact. This is an analog of Arzela-Ascoli for Sobolev spaces.
Theorem 3.3.12.
If is an elliptic operator of order for a compact manifold , then for all ,
is Fredholm. In addition, and consist of smooth configurations. Therefore is independent of .
Corollary 3.3.13.
Let be a family of elliptic operators of order . Then is constant.
Therefore if an operator depends on some choice from a connected space (for example metrics), then its index is independent of the choice. Therefore, one hopes that the index depends only on topology.
Lemma 3.3.14.
Consider and such that is compact and . Then and is closed.
Now Fredholmness of elliptic operators follows from taking to be the embedding of inside . The second part of Theorem 3.3.12 follows from elliptic regularity. Here, we know that if , then . Now if , then . In particular, implies that for all , and therefore by Sobolev embedding.
We observe that depends only on the family of principal symbols . Thus all lower order terms do not matter. Here, the lower order terms give an operator , and this is compact. Now we see that the index only depends on . This leads us to the following problem.
Question 3.3.15 (Gelfand).
Write a formula for in terms of topological data (for example characteristic classes).
Example 3.3.16.
The operators , are elliptic. We simply need to show that the symbols are injective, so let , and . Then if , we have
Similarly, we obtain , the contraction. Therefore we have
and this is nonzero whenever are nonzero. Therefore is elliptic.
Remark 3.3.17.
The Hodge theorem requires some extra work.
- 1.
First one needs to show that a generalized solution in exists (this is by general nonsense such as Hahn-Banach and Riesz representation).
- 2.
Now use elliptic regularity to get that the solution is smooth.
3.4 The index theorem
First, we will discuss symbols and K-theory. We will always denote our elliptic differential operator by . Now for all and , we know is an isomorphism. Now we can put everything together to obtain a bundle map , where is the projection. Now we know is an isomorphism outside of the zero section of , which is compact. This gives us a well-defined element .
Remark 3.4.1.
The inclusion of as the zero section of gives us a map , and we have .
Remark 3.4.2.
An orientation of gives an orientation of . Here, locally we have , and if is an oriented basis of , then we obtain an oriented basis .
We are finally ready to state
Theorem 3.4.3 (Atiyah-Singer index theorem).
Let be as above. Then we have
Here, and . Then we can multiply them together to obtain an element of , and then evaluate it on .
Note that this form of the theorem is somewhat inconvenient because may be very complicated. In some special cases, we may be able to simplify. First, let be the Thom isomorphism. Then for , we have
Here, the sign comes from our weird orientation convention. Note that if , then , where is the Thom class. Thus . Formally, we may write . Assuming that this is actually possible, we obtain , so we obtain
Remark 3.4.4.
This does not depend on . In fact, it depends only on .
For example, this makes sense when is not a zero-divisor. This also makes sense if we can interpret everything in terms of classifying spaces (for example if comes from where is even). Thus if is the pullback of a class on , then we are still fine, and here we call a universal operator.
Example 3.4.5.
Consider on the complexified tangent bundles. Also assume is even. We need to compute and . Now has Chern roots . Therefore, we have
On the other hand, we have
Finishing the computation, we obtain
Corollary 3.4.6.
This proves that in even dimension. In fact, this is the most complicated proof of this fact that Francesco knows.
Remark 3.4.7.
Our machinery is also good to work directly with complexes of differential operators
For example, we want to consider the de Rham complex. Here, we need that the associated symbol complex is exact for all . Here, we need to replace the index of with the alternating sum of dimensions of cohomologies and the Chern character of the alternating sum of the bundles.
Example 3.4.8.
Consider . If , where is oriented of dimension , then . Here, we have and . This implies that behaves nicely under direct sums. Now we use the splitting principle, so we may assume is a sum of oriented plane bundles. If with Chern root , then and . Therefore
Now we see that
Finally, we see that
Now note that is a power series in the even powers of , so we obtain
If we actually compute the , then we obtain the Hirzebruch signature formula.
Corollary 3.4.9 (Hirzebruch signature formula).
Let be an oriented manifold. Then .
Now let be a complex compact manifold. Now we know , so we have . Therefore, we have . Now if we set , we obtain the Dolbeaut complex
where because we have a complex manifold. Now it is easy to see that is elliptic. This will imply that the complex has finite-dimensional cohomology, but in fact we have
Theorem 3.4.10 (Dolbeaut).
We have and has dimension .
Therefore the index is . Using the index theorem, we can compute the index as
If has roots , we obtain that , so it has the same Chern roots. Therefore, . Next, it is easy to see that
Finally, we have . In particular, we obtain
As a corollary, we obtain
Theorem 3.4.11 (Hirzebruch-Riemann-Roch).
Let be a compact complex manifold. Then
More generally, if is a vector bundle, then .
Remark 3.4.12.
The original proof of Hirzebruch is for projective varieties.
Example 3.4.13.
If has a complex structure with structure sheaf , then we obtain
so some of .
Remark 3.4.14.
The remaining interesting case of a “universal operator” is the Dirac operator. If you would like to know more, contact Alex Xu at axu@math.columbia.edu.
3.5 Sketch of proof of the index theorem
First, we will give a short history of the proof. The original proof announcement and sketch was in 1963 using cobordism theory. This was inspired by Hirzebruch’s proof of the Hirzebruch-Riemann-Roch theorem in 1956. The drawback is that this does not generalize to the equivariant setting. The published proof in 1968 was heavily inspired by Grothendieck’s 1957 proof of the Grothendieck-Riemann-Roch theorem.
First, if is a point, then ellipticity is a vacuous assumption, so the index theorem is always true. Now the difficult part is to build the machinery to deal with functoriality. The first thing is to define the topological index of (in contrast with the analytic index ). Here, we have a commutative diagram
First, suppose that is an embedded compact submanifold with normal bundle . We can identify this with an open neighborhood of in . If is a complex vector bundle, then by the Thom isomorphism. Now we will call the map
where the second arrow is induced by the open inclusion. Now we have functoriality for embeddings with complex normal bundle.
Now suppose that is a real vector bundle. Then . If , then , and this is naturally a complex vector bundle . Therefore any inclusion induces an inclusion . Now the normal bundle to is complex. As before, we obtain a map .
One needs to check that
- •
This construction is well-defined.
- •
induces the identity.
- •
If we have , then (functoriality).
Example 3.5.1.
Consider . In this case, , so . Then is the Thom isomorphism.
Definition 3.5.2.
The map is the composition
for any inclusion .
Remark 3.5.3.
This does not depend on .
Here are some obvious properties.
- 1.
If is a point, then .
- 2.
If is an inclusion of compact smooth manifolds, then for all .
Proposition 3.5.4.
- •
There exists a unique homomorphism satisfying the above properties.
- •
Furthermore, for all , .
Given this, all that is left is to prove that the diagram
commutes. This is the hard part of the proof. The basic strategy is to find which is the inverse of to index, where . Then we need to check that satisfies the two axioms for . The problem with this strategy is that such a does not exist because is not surjective.
Proof of proposition.
Suppose is another map satisfying the properties. Consider Then the diagram
commutes. Then the desired result follows from staring at the diagram.
For the second part, consider . Here, is a rank -bundle. Thus is a rank complex bundle. Now we have a chain of morphisms
Now we have , and thus . If we take the Chern character of everything, we obtain the commutative diagram
Now on the right, we have , and therefore
by the commutativity of the middle square. Now
where is the Thom isomorphism in cohomology. Notice that , so is trivial. Therefore , and thus we have
Substituting this in , we obtain the desired result. □
Now we will see why is not surjective and how to fix this. There are topological obstructions. For example, consider . Then is a cylinder. We will choose coordinates on and on . Now choose . We obtain with compact. Now if we fixe trivializations , then for , the map . Then . This only depends (up to homotopy) on the sign of . Now has a well-defined degree, so we have
Now if , then , and thus , so . On the other hand, we note that if
then , so it cannot come from a differential operator. This tells us that elliptic differential operators are not the right framework to implement our strategy because they are too rigid. Recall that a differential operator on of order is a linear map
Here we will set
To define a pseudodifferential operator, we will create a much more complicated definition. Recall the Fourier transform
Then recall that , so if we set , then we have
Using Fourier inversion, we obtain
Definition 3.5.5.
An operator is called a pseudodifferential operator of order if there exists some such that uniformly in compact subsets in such that
Also, for all and all , the principal symbol
exists.
Definition 3.5.6.
A pseudodifferential operator is elliptic if is an isomorphism for all .
Warning 3.5.7.
Pseudodifferential operators are not local. This means that is not determined by in a neighborhood of . Thus it takes some work to define pseudodifferential operators on manifolds.
However, once we have such a definition, we have the following properties:
- •
If is an elliptic pseudodifferential operator of order , then is Fredholm.
- •
The index of only depends on restricted to the unit tangent bundle. This means the index does not depend on the order.
Example 3.5.8.
Consider . Now consider the operator given by
is a pseudodifferential operator of order . Here, if and if . For the order version of this, choose
Example 3.5.9.
Set
This defines an order elliptic pseudodifferential operator. The order version of this is given by
Here, .
Now if we consider the diagram
we have now defined . Now we want to show that . We need to show that satisfies the two axioms. For , clearly gives the dimension, but the hard part is functoriality. The complication is that we need to work in an equivariant setting. Historically, what Atiyah and Singer did was prove an equivariant version of the index theorem directly.
Some interesting consequences of the equivariant version of the index theorem are fixed point formulas. For example, we obtain the following statement:
Proposition 3.5.10.
Suppose is a compact complex manifold and acts on holomorphically. Then gives an operator on , and consider the quantity
called the holomorphic Lefschetz number. Then if all fixed points of are isolated,
where ranges over the fixed points of .