Independent scholarship · Open accessThe monthly digest

Indexed text attachment

galois-theory-in-higher-categories.tex

Source for Galois Theory in Higher Categories

View paper · Download original file

Lines 951–1050 of 4071. Source text is displayed literally.

951The homotopy exact sequence shows that the total space is also952\(\pi\)-finite. The space of morphisms for stage representations953with given endpoints is the fiber of954\[955 \operatorname{Map}(P,E)\longrightarrow956     \operatorname{Map}(P,C_{D_0}\times C_{D_1}).957\]958This fiber description follows at each stage and passes through the959filtered colimit. Both its source and target are preserved under960\eqref{eq:pro-completion}, which proves full faithfulness.961A connected stage maps to one component of the core, proving the final962assertion.963\end{proof}964 965For example, take the constant pro-space \(\mathbb N\), with966\(\mathbb N\) discrete. A family of finite discrete spaces with967fiber cardinality \(m+1\) over \(m\in\mathbb N\) belongs to968\(\operatorname{Rep}_{\mathcal S_\pi}(\mathbb N)\).969It cannot descend from its profinite completion by970Proposition \ref{prop:completion-comparison}. Each individual fiber971satisfies the requisite finiteness condition. The failure comes from the972infinitely many equivalence classes occurring in a single family.973 974\section{From one-stage actions to continuous discrete actions}975\index{profinite group!continuous action}\index{ind-completion}976 977Let \(H=\lim_i H/U_i\) be a profinite group, where the \(U_i\) form978the directed system of open normal subgroups, and set979\[980   B_{\mathrm{pro}}H=\{B(H/U_i)\}_{i}.981\]982A left action on a discrete set \(M\) is continuous exactly when983the stabilizer of each element is open. One direction follows by984taking the preimage of the singleton under each orbit map. For the985converse, the action map restricted to \(H\times\{m\}\) is locally986constant for each \(m\), and these pieces are open in987\(H\times M\).988 989The stage category \(\operatorname{Rep}_{\mathrm{Set}}(B_{\mathrm{pro}}H)\)990realizes the full subcategory of discrete actions with a common991open normal subgroup acting trivially. An action at one stage has992that property. Conversely the common subgroup allows the action993to be written over a quotient. If two actions come from994\(H/U_i\) and \(H/U_j\), then a function between them is995\(H\)-equivariant exactly when it is equivariant for996\(H/(U_i\cap U_j)\), using surjectivity of the quotient map.997Therefore passing to stages identifies their mapping sets. Every998finite continuous \(H\)-set belongs to this subcategory, as a999finite intersection of open stabilizers is open.1000 1001Every continuous discrete \(H\)-set \(M\) is a filtered union of1002finite continuous \(H\)-sets. Every orbit is finite by the compactness1003of \(H\) and the openness of its stabilizer, and one can take the1004finite unions of orbits. A finite \(H\)-set is compact among1005continuous discrete \(H\)-sets. To see this, transport a map1006into a filtered colimit to some stage as a function on its finite1007underlying set. Choose an open normal subgroup fixing all elements1008of the source and their chosen representatives at that stage.1009Equivariance can now be checked against the finitely many elements1010of the quotient by that subgroup. The required equalities hold after1011some further refinement, by the description of filtered colimits1012of sets. The same argument with finitely many pointwise equalities1013proves uniqueness in the colimit of mapping sets. Thus the category1014of all continuous discrete actions is the ind-completion of its1015finite subcategory.1016 1017The difference between stage representations and this ind-completion1018already occurs for \(H=\mathbb Z_p\). On the disjoint union1019\(\coprod_{m\geq 1}\mathbb Z/p^m\mathbb Z\), the translation1020action is continuous and requires arbitrarily deep quotients.1021Its kernel is trivial, which is a closed subgroup that is1022not open. This object is built from stage actions by taking1023a filtered union, but no single stage represents it.1024 1025A different construction first discards the topology and takes1026\(B(H^{\delta})\), the delooping of the underlying discrete group.1027For \(H=\prod_{m\geq 1}\mathbb F_2\), there is an abstract1028homomorphism \(H\to\mathbb F_2\) that annihilates1029\(\bigoplus_{m\geq 1}\mathbb F_2\) and takes the all-ones sequence1030to \(1\). Indeed, that sequence defines a nonzero vector in the1031quotient by the direct sum. Extend this vector to an algebraic basis1032and project to its coordinate. The resulting homomorphism is1033discontinuous because its kernel is dense and proper. Its action1034by translations on \(\mathbb F_2\) is an action of1035\(H^{\delta}\) with no continuous counterpart. Thus using the1036abstract classifying space instead of the pro-classifying space can1037change even the finite-coefficient theory.1038 1039\section[Closed stabilizers]{Closed stabilizers and the infinite correspondence}1040\index{profinite group!closed subgroup}\index{Galois correspondence!infinite}1041 1042The compact Hausdorff replacement of a finite discrete transitive1043action is a transitive action on a profinite space. For every1044closed subgroup \(L\leq H\), there is a homeomorphism of1045\(H\)-spaces1046\begin{equation}1047\label{eq:closed-stabilizer}1048 H/L\longrightarrow\lim_i (H/U_i)/(LU_i/U_i).1049\end{equation}1050Here \(H/L\) denotes the left \(H\)-space of cosets \(hL\), with