Skip to content

Commit 6cede9f

Browse files
author
Marc Bezem
committed
Done naturality of r_G
1 parent 1fec3b5 commit 6cede9f

1 file changed

Lines changed: 65 additions & 47 deletions

File tree

absgroup.tex

Lines changed: 65 additions & 47 deletions
Original file line numberDiff line numberDiff line change
@@ -653,7 +653,7 @@ \section{Groups: from abstract to concrete and back}
653653
is an equivalence of sets.\footnote{Indeed, conversely, $\mu(\blank,\inv u)$
654654
satisfies the condition for $\pi$. Prove this! The reason for
655655
using $\inv u$ here, and not $u$, becomes clear in the next paragraph.
656-
\wip{We may have to reconsider this}.}
656+
}
657657

658658
We have to promote $r_{\agp G}$ from an equivalence of sets to
659659
an isomorphism of abstract groups, with $\agp G$ as domain.
@@ -992,7 +992,7 @@ \section{Homomorphisms, from abstract to concrete and back}
992992
X \mapsto \absprtor[\agp H] \times_{\agp G} X,
993993
\]
994994
where $\absprtor[\agp H] \times_{\agp G} X$\footnote{%
995-
\wip{OK with $\times_{\varphi}$??}}
995+
\wip{Better with $\times_{\varphi}$??}}
996996
is the set quotient $T \times X/\sim$
997997
for the equivalence relation on $T\times X$ defined by
998998
\[
@@ -1008,7 +1008,7 @@ \section{Homomorphisms, from abstract to concrete and back}
10081008
inverse of the pointing path that we choose.
10091009
Note that the $g$ is uniquely determined, since $X$ is a $\agp G$-torsor;
10101010
this makes it particularly easy to check that $\sim$ is an equivalence
1011-
relation.\footnote{\wip{Yes, but why?} The notation $gy$ in the definition stands for
1011+
relation.\wip{Hardly, works for $\agp G$-sets too?} \footnote{The notation $gy$ in the definition stands for
10121012
the action of $g$ on $y$ as given by the abstract homomorphism that
10131013
comes with $X$; if $X$ is the principal $\agp G$-torsor this is indeed
10141014
left multiplication.}
@@ -1032,16 +1032,31 @@ \section{Homomorphisms, from abstract to concrete and back}
10321032
q : \id_{\Group} \isoto \concr\circ\abstr, \quad
10331033
r : \id_{\absGroup} \isoto \abstr\circ\concr.
10341034
\]
1035+
1036+
\begin{figure}[h]%\small
1037+
\[
1038+
\begin{tikzcd}
1039+
\sh_G \ar[dddd,mapsto,bend right=30] \ar[rrr,mapsto] &
1040+
&& (\USymG,g\mapsto(g\blank)) \ar[dddr,mapsto]\\
1041+
&\BG \ar[r,equivr,"\Bq_G"]\ar[d,"{\Bf}"'] &
1042+
\absGTor[\abstr(G)]\ar[d,"{\Bconcr(\abstr(f))}"] \\
1043+
&\BH \ar[r,equivl,"\Bq_H"'] & \absGTor[\abstr(H)] \\
1044+
\sh_H \ar[d,"{\Bfpt}"]\ar[rrr,mapsto] &&&
1045+
(\USymH,...))
1046+
\ar[d,eqr,"\Bq_H(\Bfpt)"]
1047+
\ar[r,eqr,"{\Bconcr(\abstr(f))_\pt}"]&
1048+
\absprtor[\abstr(H)] \times_{\abstr(G)} (\USymG,...)
1049+
\ar[ld,eqr,"c_{\sh_G}"] \\
1050+
\Bf(\sh_G)\ar[rrr,mapsto] &&& ((\Bf(\sh_G)\eqto\sh_H),...)
1051+
\end{tikzcd}
1052+
\]
1053+
\caption{\label{fig:q-natural}Commuting inner square of maps and
1054+
lower right triangle of pointing paths.}
1055+
\end{figure}
10351056
For naturality of $q$, consider a homomorphism $f : \Hom(G,H)$
10361057
classified by a pointed map $\Bf : \BG \ptdto \BH$.
1037-
We need to show that the square of pointed maps
1038-
\[
1039-
\begin{tikzcd}
1040-
\BG \ar[d,"\Bf"']\ar[r,"\Bq_G"] & \absGTor[\abstr(G)]\ar[d,"\Bconcr(\abstr(f))"]\\
1041-
\BH \ar[r,"\Bq_H"'] & \absGTor[\abstr(H)]
1042-
\end{tikzcd}
1043-
\]
1044-
commutes. For $z:\BG$, we have
1058+
We need to show that the inner square of pointed maps
1059+
in \cref{fig:q-natural} commutes. For $z:\BG$, we have
10451060
\begin{align*}
10461061
\Bconcr(\abstr(f))(\Bq_G(z))
10471062
&\jdeq \absprtor[\abstr(H)] \times_{\abstr(G)} \pathsp{z}(\sh_G) \\
@@ -1059,15 +1074,17 @@ \section{Homomorphisms, from abstract to concrete and back}
10591074
&h'\inv\Bfpt \Bf(p')=
10601075
h\USymf(g) \inv\Bfpt \Bf(p')=\\
10611076
&h\inv\Bfpt\Bf(g)\Bf(p')=
1062-
h\inv\Bfpt \Bf(p)
1077+
h\inv\Bfpt \Bf(p).
10631078
\end{align*}}
10641079
\[
10651080
c_z : \bigl((\sh_H \eqto \sh_H) \times_{\abstr(G)} (z \eqto\sh_G)\bigr)
1066-
\isoto (\Bf(z)\eqto \sh_H )
1081+
\isoto (\Bf(z)\eqto \sh_H ),
10671082
\]
1083+
commuting with the homomorphisms which are both based on left multiplication.
10681084
And for $z\jdeq\sh_G$, this agrees with our chosen pointing paths,
10691085
see \cref{fig:q-natural}.\footnote{%
1070-
Indeed, $\Bq(\Bfpt)$ is right multiplication by $\inv\Bfpt$,
1086+
Indeed, $\Bq(\Bfpt)$ in \cref{fig:q-natural} is
1087+
right multiplication by $\inv\Bfpt$,
10711088
which makes up for the difference between
10721089
\cref{xca:Bconcr-OK}\ref{it:Bconcr_pt} and the
10731090
function defining $c_{\sh_G}$. The dotted second components, the
@@ -1076,50 +1093,51 @@ \section{Homomorphisms, from abstract to concrete and back}
10761093

10771094
For naturality of $r$, consider an abstract group homomorphism
10781095
$\varphi : \agp G \to \agp H$, where again $S$ and $T$ are the underlying
1079-
sets of $\agp G$ and $\agp H$, respectively. It suffices to check that
1080-
square of underlying sets commutes:
1096+
sets of $\agp G$ and $\agp H$, respectively.
1097+
Recall from \cref{thm:Groupsareidentitytypes}
1098+
that $r(s)$ is right multiplication by $\inv s$.
1099+
It suffices to check that
1100+
the square of underlying sets commutes:
10811101
\[
10821102
\begin{tikzcd}
1083-
S \ar[r,"r_{\agp G}"]\ar[d,"\varphi"'] &
1103+
S \ar[rr,equivr,"r_{\agp G}"]\ar[d,"\varphi"'] &&
10841104
(\absprtor[\agp G] \eqto \absprtor[\agp G]) \ar[d,"\abstr(\concr(\varphi))"] \\
1085-
T \ar[r,"r_{\agp H}"] &
1105+
T \ar[rr,equivl,"r_{\agp H}"'] &&
10861106
(\absprtor[\agp H] \eqto \absprtor[\agp H])
10871107
\end{tikzcd}
10881108
\]
1089-
For $g:S$, this amounts to showing that the following square commutes,
1090-
where the left isomorphism is $r_{\agp H}(\varphi(g))$, and the right--down--left
1091-
detour is $\abstr(\concr(\varphi))(r_{\agp G}(g))$:
1109+
For $s:S$, this amounts to showing that $\abstr(\concr(\varphi))$
1110+
maps $\mu_{\agp G}(\blank,\inv s)$
1111+
to $\mu_{\agp H}(\blank,\inv{\varphi(s)})$.%
1112+
\footnote{The application of $\B\concr(\varphi)$ on an identification
1113+
$e:X\eqto X'$ can be analyzed step for step.
1114+
First, between the products $\absprtor[\agp H] \times X$
1115+
and $\absprtor[\agp H] \times X$ we get the equivalence
1116+
$\id_{\absprtor[\agp H]}\times e$.
1117+
1118+
Second, concerning the quotients modulo $\sim$ and $\sim'$, respectively,
1119+
we observe that $x=gy$ in $X$ and $e(x)=ge(y)$ in $X'$ are equivalent,
1120+
as the respective homomorphisms of $X$ and $X'$ correspond via $e$.
1121+
All this means that we can completely focus on the sets.
1122+
}
1123+
For this we have to show that the following square commutes,
1124+
where the left isomorphism is $r_{\agp H}(\varphi(s))$, and the right--down--left
1125+
detour is $\abstr(\concr(\varphi))(r_{\agp G}(s))$:
10921126
\[
10931127
\begin{tikzcd}
1094-
\absprtor[\agp H] \ar[r,equivl]\ar[d,equivl,"\preinv(\varphi(g))"']
1095-
& \absprtor[\agp H] \times_{\agp G} \absprtor[\agp G]
1096-
\ar[d,equivr,"\id \times_{\agp G} \preinv(g)"] \\
1097-
\absprtor[\agp H] \ar[r,equivr]
1098-
& \absprtor[\agp H] \times_{\agp G} \absprtor[\agp G]
1128+
\absprtor[\agp H] \ar[rr,eqr,"{\B\concr(\varphi)_{\pt}}"]\ar[d,equivl,"{\mu_{\agp H}(\blank,\inv{\varphi(s)})}"']
1129+
&& \absprtor[\agp H] \times_{\agp G} \absprtor[\agp G]
1130+
\ar[d,equivr,"{\id \times_{\agp G} \mu_{\agp G}(\blank,\inv s)}"] \\
1131+
\absprtor[\agp H] \ar[rr,eql,"{\B\concr(\varphi)_{\pt}}"']
1132+
&& \absprtor[\agp H] \times_{\agp G} \absprtor[\agp G].
10991133
\end{tikzcd}
11001134
\]
1101-
The horizontal isomorphisms map $h$ to $[(h,e)]$, so the square commutes
1102-
since $(h,\inv g) \sim (h\inv{\varphi(g)},e)$.
1135+
The pointing path of $\B\concr(\varphi)$ corresponds with
1136+
the inverse of the map in \cref{xca:Bconcr-OK}\ref{it:Bconcr_pt},
1137+
and maps $t$ to $[(t,e_{\agp G})]$. Hence the square commutes
1138+
since $(t,\inv s) \sim (t\inv{\varphi(s)},e_{\agp G})$.
11031139
\end{proof}
1104-
\begin{figure}[h]\small
1105-
\[
1106-
\begin{tikzcd}
1107-
\sh_G \ar[dddd,mapsto,bend right=30] \ar[rrr,mapsto] &
1108-
&& (\USymG,g\mapsto(g\blank)) \ar[dddr,mapsto]\\
1109-
&\BG \ar[r,"\Bq_G"]\ar[d,"{\Bf}"'] &
1110-
\absGTor[\abstr(G)]\ar[d,"{\Bconcr(\abstr(f))}"] \\
1111-
&\BH \ar[r,"\Bq_H"'] & \absGTor[\abstr(H)] \\
1112-
\sh_H \ar[d,"{\Bfpt}"]\ar[rrr,mapsto] &&&
1113-
(\USymH,...))
1114-
\ar[d,"\Bq_H(\Bfpt)"]
1115-
\ar[r,eqr,"{\Bconcr(\abstr(f))_\pt}"]&
1116-
\absprtor[\abstr(H)] \times_{\abstr(G)} (\USymG,...)
1117-
\ar[ld,eqr,"c_{\sh_G}"] \\
1118-
\Bf(\sh_G)\ar[rrr,mapsto] &&& ((\Bf(\sh_G)\eqto\sh_H),...)
1119-
\end{tikzcd}
1120-
\]
1121-
\caption{\label{fig:q-natural}The lower right triangle commutes.}
1122-
\end{figure}
1140+
11231141
\wip{
11241142
\section{Benefits from having the categorical equivalence}
11251143

0 commit comments

Comments
 (0)