From a Masters Qual. Practice Exam:

Let $G \leq GL(3, \mathbb{F}_3)$, be the group of invertible $3 × 3$ upper triangular matrices over the field with $3$ elements (i.e. entries below the diagonal are zero). Find the order of $G$ and show that $G$ is not a simple group i.e. show that $G$ has a proper non-trivial normal subgroup.

If we didn't have "invertible" condition, this would give us $3^6$ matrices, but I'm not sure how to quickly weed out which are and which aren't invertible.

Intuitively, I think the normal subgroup should be the diagonal matrices (I could be wrong), but I'm not sure how to prove that without some really gross matrix multiplication.

  • $\begingroup$ Since the matrices are upper triangular, the determinant is the product of the $3$ elements in the main diagonal. Thus, if you avoid zeros there, the matrix will be invertible. $\endgroup$
    – math
    Jan 6 '17 at 2:26
  • $\begingroup$ @MathChat Of course! It hadn't occurred to me to think of it that way. $\endgroup$ Jan 6 '17 at 2:35
  • $\begingroup$ Also, if you remember the general fact that the center of any general linear group is the subgroup of scalar matrices (so a subset of the diagonal matrices), that would be a good normal subgroup here. $\endgroup$
    – pjs36
    Jan 6 '17 at 2:43

In order that an upper triangular matrix

$$\begin{pmatrix} a & b & c \\ 0 & d& e \\ 0 &0 & f\end{pmatrix}$$

be invertible, it is necessarily and sufficient that the determinant $adf$ be nonzero. This means that $b,c,e$ can be anything you want, while $a, d, f$ cannot be zero. This means there are

$$3 \cdot 3 \cdot 3 \cdot 2 \cdot 2 \cdot 2 = 216 $$

such matrices.

The subgroup $D$ of diagonal invertible matrices is as far from being normal in $G$ as you can get (if $x \in G$, but not in $D$, then $xDx^{-1} \neq D$).

However, if you multiply two upper triangular matrices $x, y \in G$, notice that the entries on the diagonal of $xy$ are obtained by multiplying the corresponding entries on the diagonal of $x$ and $y$. This implies that

$$N = \{ \begin{pmatrix} 1 & b & c \\ 0 & 1& e \\ 0 &0 & 1\end{pmatrix} : b, c, e \in \mathbb{F}_3 \}$$ is a normal subgroup of $G$.

Actually, $G$ is the semidirect product of $N$ and $D$.

  • $\begingroup$ Thanks! Looks great. One question: is there a mathematical definition for "as far from being normal as you can get"? $\endgroup$ Jan 6 '17 at 2:37
  • $\begingroup$ Yes. If $H$ is a subgroup of $G$, the normalizer $N_G(H) = \{ g \in G : gHg^{-1} = H\}$ contains $H$, and is the unique maximal subgroup of $G$ in which $H$ is normal. So $H$ is normal in $G$ if and only if $N_G(H) = G$, and the other extreme is when $N_G(H) = H$, as we have here. $\endgroup$
    – D_S
    Jan 6 '17 at 2:44
  • 1
    $\begingroup$ Wow. I was actually joking, but that's really cool! $\endgroup$ Jan 6 '17 at 2:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.