4
$\begingroup$

I am trying to answer to this question: What can be said about adjunctions between groups (regarded as one-object categories)?

Here is what I have arrived at till now, but I can't conclude in some way. I mean, it seams to me that one last step is missing.

Let's consider $G_1$ and $G_2$ groups and let's view them as one-object categories $\mathcal{G}_1$ and $\mathcal{G}_2$, whose only objects are $G_1$ and $G_2$ respectively.\ Let's then take $F:\mathcal{G}_1\longrightarrow\mathcal{G}_2$, $H:\mathcal{G}_2\longrightarrow\mathcal{G}_1$ functors and suppose $F\dashv H$. This is equivalent to state that the map $\eta_{G_1}:G_1\longrightarrow HF(G_1)$ is initial in $(G_1\Rightarrow H)$ (it suffices to state it for this only map $\eta_{G_1}$ since $G_1$ is the only object of $\mathcal{G}_1$). However, being $\mathcal{G}_1$ and $\mathcal{G}_2$ one-object categories, we already know how the functors $F$ and $H$ are defined on objects, namely $F(G_1)=G_2$ and $H(G_2)=G_1$. So the map $\eta_{G_1}$ is nothing but a function from $G_1$ to itself.\ Let's now look at the comma category $(G_1\Rightarrow H)$. Its objects are the maps of the form $g_1:G_1\longrightarrow H(G_2)$, so they are simply the functions from $G_1$ to itself. A map between $g_1$ and $g_1'$ in $(G_1\Rightarrow H)$ is a function $g_2:G_2\longrightarrow G_2$ such that $g_1'=H(g_2)\circ g_1$, so we can write it as ``$g_1\rightarrow H(g_2)$''.\ Now, we want to impose that $\eta_{G_1}$ is an initial object in $(G_1\Rightarrow H)$. So let $g_1$ be an arbitrary object in $Ob(\mathcal{G}_1)$, and we force the map $g_2\in (G_1\Rightarrow H)(\eta_{G_1},g_1)$ to be unique. Thus we must have $g_2$ to be the unique map such that $g_1=H(g_2)\circ \eta_{G_1}$.

So now, what can we say really interesting about this $g_2$???

Thanks in advance for any help!

$\endgroup$
  • $\begingroup$ $\eta$ is not a function from $G_1$ to itself. It's a morphism in the category $\mathcal G_1$ from the unique object to itself. What are the morphisms in $\mathcal G_1$, again? $\endgroup$ – Kevin Carlson Jan 13 '17 at 19:14
  • $\begingroup$ You're right! I don't really know what they are in reality, indeed I have always thought of as the elements of the group $G_1$ . What are they effectively? $\endgroup$ – any_one Jan 13 '17 at 19:18
  • $\begingroup$ No, they are indeed elements of the group-not functions. $\endgroup$ – Kevin Carlson Jan 13 '17 at 22:33
  • $\begingroup$ Mmm ok...thank you :) $\endgroup$ – any_one Jan 14 '17 at 13:04
5
$\begingroup$

Since groups as one-object categories are a special case of groupoids, that is, categories in which every morphism is an isomorphism, for any adjunction $F\dashv H\colon\mathcal G_1\to\mathcal G_2$ between two group(oid)s, the unit and counit natural transformations $\mathrm{id}_{\mathcal G_1}\overset{\eta}\Rightarrow HF$ and $FH\overset\epsilon\Rightarrow\mathrm{id}_{\mathcal G_2}$ will be natural isomorphisms. In particular, all adjoint pairs of functors between two group(oid)s are adjoint equivalences.

In the special case of the adjunction $F\dashv H\colon\mathcal G_1\to\mathcal G_2$ of one-object categories (i.e. monoids), we can identify $\eta$ and $\epsilon$ with elements of $\mathcal G_1$ and $\mathcal G_2$ respectively, which must satisfy the naturality conditions

  1. $HF(g_1)\eta=\eta g_1$ for all $g_1\in\mathcal G_1$
  2. $g_2\epsilon=\epsilon FH(g_2)$ for all $g_2\in\mathcal G_2$

Furthermore, the zig-zag identities which makes the unit and counit into an adjunction between the two functors (i.e. monoid homomorphisms) are

  1. $\epsilon F(\eta)=\mathrm{id}_{\mathcal G_2}$
  2. $H(\epsilon)\eta=\mathrm{id}_{\mathcal G_1}$

By multiplying the naturality conditions with the inverses given by the zig-zag identity, we end up reducing all four conditions to the following two

  1. $H(\epsilon F(g_1))\eta=g_1$ for all $g_1\in\mathcal G_1$
  2. $g_2=\epsilon F(H(g_2)\eta))$ for all $g_2\in\mathcal G_2$

Indeed, the zig-zag identities are the special case where $g_1$ and $g_2$ are the identity elements of the groups. In particular, in the case of a group (where left and right inverses coincide), we have $HF(g_1)=\eta^{-1}g_1\eta$ and $FH(g_2)=\epsilon^{-1}g_2\epsilon$ where $\eta^{-1}=H(\epsilon)$ and $\epsilon^{-1}=F(\eta)$. Thus an adjunction between two groups consists of a pair of group homomorphisms $F\colon\mathcal G_1\rightleftarrows\mathcal G_2\colon H$ whose composites are inner automorphisms so that $F$ sends the inner automorphism of $\mathcal G_1$ to the inverse inner automorphism of $\mathcal G_2$ and reversly for $H$.

$\endgroup$
2
$\begingroup$

Every natural transformation between functors between groups is an equivalence, so every adjunction is an equivalence.

$\endgroup$
  • $\begingroup$ And equivalences between groups viewed as categories are isomorphisms of the groups, for they induce isomorphisms between the endomorphism monoid of the objects. $\endgroup$ – Mariano Suárez-Álvarez Jan 13 '17 at 19:58

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.