Prove a CG module is irreducible

  • Thread starter Thread starter fishturtle1
  • Start date Start date
  • Tags Tags
    module
fishturtle1
Messages
393
Reaction score
82
Homework Statement
Suppose ##G = D_6 = \langle a, b \vert a^3 = b^2 = 1, b^{-1}ab = a^{-1} \rangle## and let ##\omega = e^{2\pi i/3}##. Prove that the ##2## dimensional subspace ##W## of the ##\mathbb{C}G## defined by
$$W = sp(1 + \omega^2a + \omega a^2, b + \omega^2ab + \omega a^2b)$$

is an irreducible ##\mathbb{C}G##-submodule of the regular ##\mathbb{C}G## module.
Relevant Equations
Definitions:

We define ##\mathbb{C}G = \lbrace \sum_{g \in G} \mu_g g : \mu_g \in \mathbb{C} \rbrace##. For ##r \in \mathbb{C}G## and ##h \in G##, we define ##rh = \sum_{g \in G} \mu_g (gh)##.

We say ##W## is a ##\mathbb{C}G## submodule of ##\mathbb{C}G## if it is a subspace of ##\mathbb{C}G## and is satisfies ##wh \in W## for all ##w \in W## and ##h \in G##.

We say a ##\mathbb{C}G## module is irreducible if its only submodules are itself and ##\lbrace 0 \rbrace##.
Proof: We first show ##W## is a ##\mathbb{C}G## module. Let ##u = 1 + \omega^2a + \omega a^2## and ##v = b + \omega^2ab + \omega a^2b##. It is enough to show ##ua, ub, va, vb \in W##. We have\begin{align}
ua &= a + \omega^2a^2 + \omega \\
ub &= b + \omega^2ab + \omega a^2b \\
va &= ba + \omega^2 aba + \omega a^2ba = a^3b + \omega^2 b + \omega^2 ab\\
vb &= 1 + \omega^2a + \omega a^2 \\
\end{align}

Then ##ub = v \in W## and ##vb = u \in W##. But I do not see how to show ##ua## and ##va## are in ##W##. If I could do this, then I would have to show ##W## has no non trivial submodule. Suppose ##U## is a submodule of ##W## such that ##\dim U = 1##. Then ##U = sp( \alpha u + \beta v)## for some ##\alpha, \beta \in \mathbb{C}##. Since ##U## is a ##\mathbb{C}G## module, we must have ##xg \in U## for all ##x \in U## and ##g \in G##. In particular, we have ##(\alpha u + \beta v )b = \alpha v + \beta u##. So, there exists ##\lambda \in \mathbb{C}## such that ##\lambda(\alpha v + \beta u) = \alpha u + \beta v##. We can rewrite this as

$$(\lambda\beta)u + (\lambda\alpha)v = \alpha u + \beta v$$

Since ##u, v## are linearly independent, we have ##\lambda \beta = \alpha## and ##\lambda \alpha = \beta##. And this implies ##\lambda\beta = \lambda^{-1}\beta##. So, ##\lambda = \lambda^{-1}## or ##\beta = 0##. Does this seem on the right track?
 
Physics news on Phys.org
fishturtle1 said:
Homework Statement:: Suppose ##G = D_6 = \langle a, b \vert a^3 = b^2 = 1, b^{-1}ab = a^{-1} \rangle## and let ##\omega = e^{2\pi i/3}##. Prove that the ##2## dimensional subspace ##W## of the ##\mathbb{C}G## defined by
$$W = sp(1 + \omega^2a + \omega a^2, b + \omega^2ab + \omega a^2b)$$
is an irreducible ##\mathbb{C}G##-submodule of the regular ##\mathbb{C}G## module.
Relevant Equations:: Definitions:

We define ##\mathbb{C}G = \lbrace \sum_{g \in G} \mu_g g : \mu_g \in \mathbb{C} \rbrace##. For ##r \in \mathbb{C}G## and ##h \in G##, we define ##rh = \sum_{g \in G} \mu_g (gh)##.

We say ##W## is a ##\mathbb{C}G## submodule of ##\mathbb{C}G## if it is a subspace of ##\mathbb{C}G## and is satisfies ##wh \in W## for all ##w \in W## and ##h \in G##.

We say a ##\mathbb{C}G## module is irreducible if its only submodules are itself and ##\lbrace 0 \rbrace##.

Proof: We first show ##W## is a ##\mathbb{C}G## module. Let ##u = 1 + \omega^2a + \omega a^2## and ##v = b + \omega^2ab + \omega a^2b##. It is enough to show ##ua, ub, va, vb \in W##. We have
\begin{align}
ua &= a + \omega^2a^2 + \omega \\
ub &= b + \omega^2ab + \omega a^2b \\
va &= ba + \omega^2 aba + \omega a^2ba = a^3b + \omega^2 b + \omega^2 ab\\
vb &= 1 + \omega^2a + \omega a^2 \\
\end{align}

Then ##ub = v \in W## and ##vb = u \in W##. But I do not see how to show ##ua## and ##va## are in ##W##.

\begin{align*}
ua&=a+\omega^2a^2+\omega=\omega u\\
va&=uba=ua^{-1}b^{-1}=ua^2b=\omega uab=\omega^2ub=\omega^2v\\
\end{align*}

fishturtle1 said:
If I could do this, then I would have to show ##W## has no non trivial submodule. Suppose ##U## is a submodule of ##W## such that ##\dim U = 1##. Then ##U = sp( \alpha u + \beta v)## for some ##\alpha, \beta \in \mathbb{C}##. Since ##U## is a ##\mathbb{C}G## module, we must have ##xg \in U## for all ##x \in U## and ##g \in G##. In particular, we have ##(\alpha u + \beta v )b = \alpha v + \beta u##. So, there exists ##\lambda \in \mathbb{C}## such that ##\lambda(\alpha v + \beta u) = \alpha u + \beta v##. We can rewrite this as

$$(\lambda\beta)u + (\lambda\alpha)v = \alpha u + \beta v$$

Since ##u, v## are linearly independent, we have ##\lambda \beta = \alpha## and ##\lambda \alpha = \beta##. And this implies ##\lambda\beta = \lambda^{-1}\beta##. So, ##\lambda = \lambda^{-1}## or ##\beta = 0##. Does this seem on the right track?

Yes. This means you have ##\beta =0## or ##\lambda =\pm 1##. Now you need to find an argument why the latter cannot be, and do the same with ##\alpha .##
 
Last edited:
fresh_42 said:
\begin{align*}
ua&=a+\omega^2a^2+\omega=\omega u\\
va&=uba=ua^{-1}b^{-1}=ua^2b=\omega uab=\omega^2ub=\omega^2v\\
\end{align*}
Yes. This means you have ##\beta =0## or ##\lambda =\pm 1##. Now you need to find an argument why the latter cannot be, and do the same with ##\alpha .##
Thank you!

Continuing from above, we have ##\beta = 0## or ##\lambda = \pm 1##. We will show the latter is not possible. If ##\lambda = 1##, then ##\beta u + \alpha v = \alpha u + \beta v##. This implies ##(\beta - \alpha)u + (\alpha - \beta)v = 0##. Since ##u, v## are linearly independent, we have ##\alpha = \beta##. This implies ##U = sp(u + v)##. But this implies ##(u+v)a = ua + va = \omega u + \omega^2 v \in U##. Since ##\omega u + \omega^2 v## is not a scalar multiple of ##U##, we have reached a contradiction.

If ##\lambda = -1##, then ##-\beta u + -\alpha v = \alpha u + \beta v##. This implies ##(-\alpha - \beta)u + (-\alpha - \beta)v = 0##. Since ##u, v## are linearly independent, we have ##\alpha = -\beta##. This implies ##U = sp(-u + v)##. But this implies ##(-u+v)a = -ua + va = -\omega u + \omega^2v \in U##. Since ##-\omega u + \omega^2v## is not a scalar multiple of ##-u+v##, we have reached a contradiction.

We can conclude ##\beta = 0##. By a similar argument, we can show ##\alpha = 0##. We can conclude ##U = \lbrace 0 \rbrace##, which contradicts our assumption that ##\dim U = 1##. This shows ##W## is irreducible.
 
fishturtle1 said:
Thank you!

Continuing from above, we have ##\beta = 0## or ##\lambda = \pm 1##. We will show the latter is not possible. If ##\lambda = 1##, then ##\beta u + \alpha v = \alpha u + \beta v##. This implies ##(\beta - \alpha)u + (\alpha - \beta)v = 0##. Since ##u, v## are linearly independent, we have ##\alpha = \beta##. This implies ##U = sp(u + v)##. But this implies ##(u+v)a = ua + va = \omega u + \omega^2 v \in U##. Since ##\omega u + \omega^2 v## is not a scalar multiple of ##U##, we have reached a contradiction.

If ##\lambda = -1##, then ##-\beta u + -\alpha v = \alpha u + \beta v##. This implies ##(-\alpha - \beta)u + (-\alpha - \beta)v = 0##. Since ##u, v## are linearly independent, we have ##\alpha = -\beta##. This implies ##U = sp(-u + v)##. But this implies ##(-u+v)a = -ua + va = -\omega u + \omega^2v \in U##. Since ##-\omega u + \omega^2v## is not a scalar multiple of ##-u+v##, we have reached a contradiction.

We can conclude ##\beta = 0##. By a similar argument, we can show ##\alpha = 0##. We can conclude ##U = \lbrace 0 \rbrace##, which contradicts our assumption that ##\dim U = 1##. This shows ##W## is irreducible.
Looks fine. Just one final question: Why are ##u,v## linear independent?
 
  • Like
Likes fishturtle1
fresh_42 said:
Looks fine. Just one final question: Why are ##u,v## linear independent?
I think that is because ##W = sp(u, v)## and we are given ##\dim W = 2##. But i think that is circular?

Suppose there are scalars ##\lambda_1, \lambda_2## such that ##\lambda_1 u + \lambda_2 v = 0## where ##\lambda_1 \neq 0## or ##\lambda_2 \neq 0##. Multiplying by ##b## on both sides gives ##\lambda_1 v + \lambda_2 u = 0##. WLOG, suppose ##\lambda_1 \neq 0##. Then ##u + \lambda_1^{-1}\lambda_2v = 0## and ##v + \lambda_1^{-1}\lambda_2u = 0##. This implies ##u## and ##v## are scalar multiples of each other.

But any scalar multiple of ##u## looks like ##\lambda u = \lambda + \lambda\omega^2a + \lambda \omega a^2## i.e. it has a constant term. And ##v## does not have a constant term. We can conclude ##u## and ##v## are not scalar multiples of each other, which implies ##\lambda_1 = 0 = \lambda_2##?

I am not sure.
 
fishturtle1 said:
But any scalar multiple of ##u## looks like ##\lambda u = \lambda + \lambda\omega^2a + \lambda \omega a^2## i.e. it has a constant term. And ##v## does not have a constant term.
Yes. That is the argument.

We obviously have ##u,v\neq 0## so ##0=\mu u+\nu v## can be written as ##u=\lambda v## (or the other way around, or ##\mu=\nu=0## in which case we are done). ##\lambda \in \mathbb{C}## so comparing the scalar component yields ##\lambda \cdot 1 =0## which implies ##u=0.## But ##u\neq 0.##
 
  • Informative
Likes fishturtle1
There are two things I don't understand about this problem. First, when finding the nth root of a number, there should in theory be n solutions. However, the formula produces n+1 roots. Here is how. The first root is simply ##\left(r\right)^{\left(\frac{1}{n}\right)}##. Then you multiply this first root by n additional expressions given by the formula, as you go through k=0,1,...n-1. So you end up with n+1 roots, which cannot be correct. Let me illustrate what I mean. For this...
Back
Top