There is a group representation of $\Bbb Z/p\Bbb Z$ that is not decomposable over the field $\Bbb F_p$, where $p$ is prime

62 Views Asked by At

let p be prime. prove there is a group representation of $\Bbb Z/p\Bbb Z$ that is not decomposable over a field $\Bbb F_p$

for similar and simpler questions i showed homomorphism $\mu$ such that

$$\mu(x)(v) = \begin{pmatrix} 1 & x \\ 0 & 1 \\ \end{pmatrix} \begin{pmatrix} a \\ b \\ \end{pmatrix}$$

and then showed invariant field

$L = \begin{pmatrix} a \\ 0 \\ \end{pmatrix}$

that does not have an invariant complement.

I'm not sure if thats the correct way and if so then what is the explanation for that.

Any help will be appreciated.

2

There are 2 best solutions below

4
On BEST ANSWER

You have the right idea. Let $\mu : \Bbb Z/p\Bbb Z \to GL_2(\Bbb F_p)$ such that $$\mu(x)=\begin{pmatrix} 1 & x \\ 0 & 1 \\ \end{pmatrix}$$

A non-trivial invariant subspace $L$ has dimension $1$, generated by some $(a,b)$.

You can show that $L$ must be generated by $(1,0)$ [see below]. So the only non-trivial invariant subspace is $L = \{(a,0) \mid a \in \Bbb F_p\}$

But then your vector space $V=\Bbb F_p^2$ is not the direct sum of $L$ with itself. More precisely, if $V$ was a direct sum of non-trivial invariant subspaces $$V= \bigoplus\limits_{i=1}^r W_i$$ then $r>1$ and $W_i=L$ for all $i$. But this can't happen because we have for instance $W_1 \cap W_2 = L \cap L = L \neq \{0\}$, so the sum can't be a direct sum.


Here is how you can show that $L = \langle (a,b) \rangle$ must be generated by $(1,0)$.

It is necessary for $L$ to be invariant under $\mu$ that $$\mu(1)(a,b)= \begin{pmatrix} 1 & 1 \\ 0 & 1 \\ \end{pmatrix} \begin{pmatrix} a \\ b \\ \end{pmatrix} = \begin{pmatrix} a+b \\ b \\ \end{pmatrix} = (a+b,b)= k(a,b)=(ka, kb) \in L$$ for some $k \in \Bbb F_p$. Then you have $a+b=ka$ and $b=kb$.

  • If $a=0$ then $0+b=b=k \cdot 0=0$ and then $L$ is just $\{(0,0)\}$ which is not of dimension $1$. So we must have $a≠0$.
  • If $b≠0$ then $k=1$ from the second equation. Then $a+b=a$ from the first one, so that $b=0$, contradiction. Therefore $b=0$.

Conclusion: $L$ is generated by $(a,0)$ with $a≠0$, so it is also generated by $(1,0)$.

0
On

Here is another way to think about this. Denote the representation you give as $V=\mathbb F_p^2$.

Recall that representations of $\mathbb Z/p$ over $\mathbb F_p$ are the same as $\mathbb F_p[\mathbb Z/p]\cong\mathbb F_p[\xi]/(\xi^p-1)=\mathbb F_p[\xi]/(\xi-1)^p=\mathbb F_p[\mu]/\mu^p$-modules (here, $\mu=\xi-1$).

Now, the $\xi\in\mathbb F_p[\xi]/(\xi^p-1)$ acts on $V$ as $(u,v)\mapsto (u+v,v)$, so $\mu=\xi-1$ acts as $(u,v)\mapsto (v,0)$. Moreover, $1\in\mathbb F_p[\xi]/(\xi^p-1)$ acts as the identity.

Thus, there is an isomorphism of modules $V\to\mathbb F_p[\mu]/\mu^2:(u,v)\mapsto v+u\mu$.

Now, the fact that $V$ does not decompose into irreducible representations corresponds to the fact that $\mathbb F_p[\mu]/\mu^2$ cannot be written as a direct sum of $1$-dimensional $\mathbb F_p[\mu]/\mu^p$-modules. There is only one $1$-dimensional $\mathbb F_p[\mu]/\mu^p$-module, $\mathbb F_p$, and $\mathbb F_p[\mu]/\mu^2\ncong\mathbb F_p\oplus\mathbb F_p$ (look at how $\mu$ acts on them).