Exploration 3: Representation theory
January 6, 2024.Questions:
- TODO
Last exploration, we studied the properties of individual -matrices. This time we’re studying -matrices collectively.
First, we define as the set of all -module homomorphisms . When are free -modules, the elements of are -matrices. Generally this means is a noncommutative ring, given that matrix multiplication is not commutative in general.
Let’s start by representing groups using subsets of .
For instance, we can express the group (nonzero complex numbers under multiplication) as real matrices. We define a map :
so that and
Since it preserves product and inverse, these matrices ( homomorphisms) define a group, isomorphic to . This is just one of many encodings of complex numbers into matrices. The key is that , and that don’t interact with each other.
Here we found that was a suitable subgroup of -module homomorphisms. In general, to represent a group, we must select the subgroup of that consists of invertible -module homomorphisms. In practice, and are free modules and , and invertibility implies . Thus, when representing a group, we can choose among the invertible matrices in , also written as , the group of automorphisms on , whose elements we express as invertible -matrices.
So we assign to each group element an invertible -matrix. A matrix -representation of a group is a group homomorphism .
- Every representation assigns each to a specific permutation matrix in .
- But this implicitly defines an action of on . The action is given by matrix muliplication in , where each acts by multiplying by its corresponding matrix.
- Conversely, by Cayley’s theorem every group is isomorphic to a permutation group, and therefore can act on by permuting the basis -vectors. But permutations are automorphisms, so this essentially assigns each element an automorphism over , i.e. constructing a matrix -representation .
The representation (mentioned in the above proof) is known as the regular representation of a finite group . The main idea of this representation is to have permute the indices of according to the permutation group isomorphic to .
In this section, we show the module analog of group actions.
In the same manner that adjoining an indeterminate symbol to gives you a polynomial ring , we can adjoin a whole group to to get a group ring . Elements look like -linear combinations of elements of :
Since every element in is by definition uniquely represented as a linear combination of elements , we know that spans . Like in polynomial rings, the only linear combination equal to zero is the one where every coefficient is zero, so the elements of are linearly independent and therefore form a basis of . Having a basis means the elements of can be represented by -vectors of coefficients, and therefore the group ring is a free module by construction.
Thus a group ring is simultaneously a group and a free -module.
Since is a free -module, it has a basis whose elements are precisely the elements of the group . It is a property of groups that any element permutes the elements of by left-multiplication. Permuting the elements is an automorphism on and therefore an automorphism on , thus is exactly the regular representation .
A representation assigns an automorphism on the -module to every element .
Since the automorphisms on -modules are exactly the linear transformations by definition, every is assigned a linear transformation on . In other words, the action of on must be linear in .
But is exactly every linear combinations of with respect to , and therefore contains every possible action of on .
This means that every action defined for on some module can be linearly extended to an action of on . Specifically, if the group action is defined for all , then we define as . Such a module is known as a -module, a module over a group ring.
is a free -module, and has a natural group action on by left-multiplication in the ring. Thus it is a -module.
To preserve the -module structure, -submodules of -modules must be -invariant: just as how -modules are closed under the action of , -submodules must be closed under the action of in the sense that for all , .
Similarly, homomorphisms between -modules must be -equivariant: they must commute with the action of in the sense that for all , .
The trick to get -equivariance is to take a sum over all actions of : which we can do because is finite. Then we can show that for : and thus is -equivariant.
- Since is free, it has a basis, and so does the submodule . We can obtain a projection by mapping the basis vectors not in to basis vectors in .
- Recall that endomorphisms like projections can be made -equivariant via an averaging trick. To ensure that the resulting endomorphism is still a projection, we need to ensure that it is idempotent and that its image is . It is enough to construct as a map that fixes elements (thus ensuring idempotence and that the image is at least ) and maps other elements in to an element in (thus ensuring that the image is at most ).
- Given that is a unit in , the map obtained by the averaging trick can be modified to ensure the above properties:
- This fixes : and maps all to an element in , since
- Thus as defined is a -equivariant projection onto .
In this section, we consider representations as algebraic structures in their own right.
Note that for our representation for in the beginning, each element of is assigned a distinct element of , i.e. the representation is injective. When the representation is injective (i.e. the domain is isomorphic to its image), we call it a faithful representation, and they are the typical ones where each element of is represented by a distinct matrix in .
There are also non-faithful representations. An example of a very not faithful representation is the trivial representation, which represents every element as the zero element in the zero module. Another is the sign representation: , which simply takes the sign ( or ) of every permutation in . Another example is , where is some matrix representation. In general, a representation is a group homomorphism (where is some -module).
This result comes naturally from the fact that a group ring encodes all possible linear group actions and therefore its action on a module models every possible automorphism on the -modules , since automorphisms on -modules are linear by definition.
A faithful representation is injective, i.e. its domain is isomorphic to its image . This means that only one element maps to the identity automorphism, so only one element acts as identity on .
If two elements have the same action on , then must be the zero action (sending all elements of to zero). But implies .
The regular representation (or equivalently, ) represents elements of as automorphisms on the free -module (whose elements are -vectors). It is faithful, since it is composed of distinct permutations of , so there is always a faithful matrix representation of any group . Can we do better? Ideally, we want to represent elements of a group using -vectors that are perhaps smaller than , without affecting the faithfulness of .
By using the above fact that any representation is equivalent to some -module , we can take subrepresentations of as -submodules of . To begin, given a representation , we can think about quotienting the corresponding -module by one of its -submodules. However, quotienting might create a representation that isn’t faithful. When does quotienting the underlying module of a faithful representation preserve faithfulness?
Recall that preserving the group action means that every element of the group is identified with a distinct action. In particular, that means there is only one element that behaves like the identity action: the identity element . But if the representation maps only to the identity in , then it is injective, i.e. faithful.
Therefore, quotienting preserves faithfulness of exactly when the quotient preserves the group action on . In other words, the submodule must be -invariant: it must remain unchanged under the group action of . But since -submodules are -invariant by definition, quotienting by an -submodule always preserves faithfulness.
As it turns out, by quotienting repeatedly by different -submodules, we can “factor” into a direct sum of submodules known as irreducible representations, or irreps.
- Recall that if a projection exists on
,
then
is isomorphic to
.
So it is enough to define a projection
where
and
.
Let’s see what those conditions imply.
- To ensure that , we need to ensure that every element in differs by some element in .
- The requirement implies that maps to the -invariant submodule of . Thus we need to be -equivariant () so that it preserves -invariance.
- Finally, to be a projection, must be idempotent.
- We’ll start with the requirement that is -equivariant. In order to get -equivariance, one trick is to take the sum of all products with : Then we can show that for : where because the act of multiplying every element of the sum by just permutes the order of the sum. Therefore, commutes with the action of on , and is therefore -equivariant.
- Next up is ensuring that the elements of differ by elements of . One way is to send every to via some projection . We redefine : Then is a linear combination of elements of . This means that is a subset of , and we can show that two elements of differ by :
- Finally, must be a projection. Therefore, it must be idempotent, and every element of must be mapped to. Without losing the previous properties, we can have for every by taking the inverse action of , and dividing by (which works since ): because Therefore for all , meaning . By definition, remains a linear combination of elements of , and therefore . Therefore .
- Finally, since is an idempotent endomorphism, we know that decomposes into . Since and by construction, we have .
- If the -module corresponding to the representation doesn’t reduce into irreps, there is a proper -invariant submodule of .
- By the above theorem, which we can use since , there is some complementary -invariant submodule such that .
- By recursively decomposing proper -invariant submodules of each of the factors, you get smaller and smaller submodules. Since is finite, this process eventually stops when you are left with a direct sum of irreps that is isomorphic to the given representation .
A module that can be written as a direct sum of simple modules is semisimple. Hence:
Corollary: A -module is semisimple if is finite and .
In this section, we try to find irreps of any matrix representation.
Now that we know by Maschke’s theorem that we can factor any finite group representation into irreps (assuming ), let’s do so for matrix representations (which are always finite).
A matrix representation is special compared to more general representations because it implies that the module is a free module. This simplifies finding irreps considerably.
In this section, we find an easier way to discover the eigenvalues in an algebraically closed field.
Find the eigenvalues of a given -matrix, where is a PID.
I know that the characteristic equation is used to find eigenvalues in vector spaces, which are defined over fields. What happens if we move to the world of modules defined over integral domains?
< Back to category Exploration 3: Representation theory (permalink)Exploration 2: Module actions