2

The title basically. I was wondering if the inner automorphism group "distributed" over direct product. My intuition says yes but i am really not sure. My reasoning is that for any $\sigma_{(g,h)} \in Inn(G \times H)$ you can map it to $(\sigma_g,\sigma_h) \in Inn(G)\times Inn(H)$ and that would be an isomorphism.

Thanks in advance

2 Answers2

7

Recall that $\operatorname{Inn}(G)\cong G/Z(G)$, where $Z(G)$ is the center of $G$. By fundamental properties of the center, we have that $Z(G\times H)=Z(G)\times Z(H)$ (see here). Also, by properties of the quotient group, if $N\trianglelefteq G$ and $K\trianglelefteq H$, then $(G\times H)/(N\times K)\cong G/N\times H/K$ (see here). Gathering everything together, we obtain $$\operatorname{Inn}(G\times H)\cong(G\times H)/(Z(G\times H))=(G\times H)/(Z(G)\times Z(H))\cong G/Z(G)\times H/Z(H)\cong\operatorname{Inn}(G)\times \operatorname{Inn}(H),$$ as desired. $\square$

Ikeroy
  • 682
  • Wonderful. Thank you very much for the detailed answer. – eagle I Feb 09 '25 at 09:22
  • This question seems not to meet the standards for the site. Instead of answering it, it would be better to look for a good duplicate target, or help the user by posting comments suggesting improvements. Please also read the meta announcement regarding quality standards. – Martin Brandenburg Feb 10 '25 at 13:28
0

While this follows from standard properties of the center and quotients, it's also easy to argue directly that your suggested map is an isomorphism. A benefit is that the implicit isomorphism becomes entirely explicit.

Via projections: One way to organize the argument is to first verify the following, each of which is simple:

  1. $\sigma_{g_1} \circ \sigma_{g_2} = \sigma_{g_1 g_2}$
  2. $\sigma_{(g,h)} = \sigma_g \times \sigma_h$
  3. $\alpha \times \beta = \alpha' \times \beta'$ iff $\alpha = \alpha', \beta = \beta'$ (when the domains are non-empty)

Your suggested map is $\sigma_{(g, h)} \mapsto (\sigma_g, \sigma_h)$. Verify it's a well-defined bijection as follows: \begin{align*} \sigma_{(g, h)} = \sigma_{(g', h')} &\Leftrightarrow \sigma_g \times \sigma_h = \sigma_{g'} \times \sigma_{h'} \\ &\Leftrightarrow \sigma_g = \sigma_{g'}, \sigma_h = \sigma_{h'} \\ &\Leftrightarrow (\sigma_g, \sigma_h) = (\sigma_{g'}, \sigma_{h'}). \end{align*}

It's also a homomorphism, hence isomorphism: \begin{align*} \sigma_{(g_1, h_1)} \circ \sigma_{(g_2, h_2)} &= \sigma_{(g_1 g_2, h_1 h_2)} \\ &\mapsto (\sigma_{g_1 g_2}, \sigma_{h_1 h_2}) \\ &= (\sigma_{g_1} \circ \sigma_{g_2}, \sigma_{h_1} \circ \sigma_{h_2}) \\ &= (\sigma_{g_1}, \sigma_{h_1}) \circ (\sigma_{g_2}, \sigma_{h_2}). \end{align*}

Via centers: If you don't like products of maps, you can use centers instead:

  1. $\sigma_g = \sigma_{g'}$ iff $g^{-1}g' \in Z(G)$
  2. $(g, h) \in Z(G \times H)$ iff $g \in Z(G), h \in Z(H)$.

Well-defined bijection: \begin{align*} \sigma_{(g, h)} = \sigma_{(g', h')} &\Leftrightarrow (g, h)(g, h)^{-1} = (gg^{-1}, hh^{-1}) \in Z(G \times H) \\ &\Leftrightarrow gg^{-1} \in Z(G), hh^{-1} \in Z(H) \\ &\Leftrightarrow \sigma_g = \sigma_{g'}, \sigma_h = \sigma_{h'} \\ &\Leftrightarrow (\sigma_g, \sigma_h) = (\sigma_{g'}, \sigma_{h'}). \end{align*}

(The homomorphism verification is the same.)

  • This question seems not to meet the standards for the site. Instead of answering it, it would be better to look for a good duplicate target, or help the user by posting comments suggesting improvements. Please also read the meta announcement regarding quality standards. – Martin Brandenburg Feb 10 '25 at 13:28
  • 1
    I'm sorry, but you're simply incorrect that the linked post is a duplicate of this one. The truth of this one is taken as a starting point over there, and user Lee Mosher has independently linked to this one from there in the meantime. It's true that you can unwind various statements over there, appropriately specialized, and also expand some unwritten arguments to piece together a proof here. But it's totally unreasonable to expect somebody asking this question in the first place to do so--it's just assumed to be trivial. I've consequently voted to reopen. – Joshua P. Swanson Feb 11 '25 at 05:47