Independent scholarship · Open accessThe monthly digest

Indexed text attachment

differential-topology-in-higher-categories.tex

Source for Differential Topology in Higher Categories: Mapping Objects, Derived Intersections, Spaces of Choices, and Cobordism

View paper · Download original file

Lines 1091–1190 of 3851. Source text is displayed literally.

1091the form that generalizes. A $D$-open subobject of a mapping object has the same local structure as the1092mapping object, and so inherits whatever infinite-dimensional manifold structure the latter has1093\cite{KM97}.1094 1095\section{Genericity}1096 1097\begin{theorem}[Baire property]\label{thm:baire}\Scl[Hirsch 2.4.4]1098Every weakly closed subset of $C^r_S(M,N)$ is a Baire space in the strong topology: a countable1099intersection of open dense subsets is dense.1100\end{theorem}1101 1102A subset is \term{residual} if it contains a countable intersection of open dense sets. The theorems of the1103next chapter assert that certain sets are residual, and \cref{thm:baire} is what turns such an assertion1104into one about density. The proof realizes $C^r(M,N)$, by the prolongation1105$j^r\colon C^r(M,N)\to C^0(M,J^r(M,N))$, as a weakly closed set of sections of the jet bundle; the strong1106$C^r$ topology is by definition the one induced from the strong $C^0$ topology on sections, and the Baire1107property is proved there by a diagonal argument.1108 1109\begin{remark}[What the topos does not see]\label{rem:nobaire}1110The $D$-topology is defined from the plots, so openness is a notion internal to smooth sets.1111Residuality is not. A residual set need not be $D$-open, and a set meeting every plot in a residual set1112of parameters need not be residual. We know of no formulation of the Baire property of1113$C^\infty_S(M,N)$ in terms of the sheaf $[M,N]$ alone. What the sheaf does see is the parametric form of1114genericity (\cref{thm:parametric}): in a family rich enough to be a submersion onto its target, almost every1115member is generic. In applications that statement is usually the one wanted.1116\end{remark}1117 1118\section{Jets and partial differential relations}1119 1120The prolongation is a map of smooth sets $j^r\colon[M,N]\to\Gamma(J^r(M,N))$ into the smooth set of sections1121of the jet bundle over $M$. It is a monomorphism, and its image consists of the \term{holonomic} sections. A1122condition on maps that is defined by a subset $\mathcal R\subset J^r(M,N)$, such as being an immersion (the1123$1$-jet is injective) or a Morse function (the $1$-jet meets the zero locus only where the $2$-jet is1124non-degenerate), gives a pullback square1125\[\begin{tikzcd}1126{[M,N]_{\mathcal R}}\arrow[r]\arrow[d] & \Gamma(\mathcal R)\arrow[d]\\1127{[M,N]}\arrow[r,"j^r"] & \Gamma(J^r(M,N))1128\end{tikzcd}\]1129in which the right-hand side involves no differential equation: $\Gamma(\mathcal R)$ is the object of1130\term{formal solutions}. If $\mathcal R$ is open, the left-hand subobject is $D$-open when $M$ is compact, by1131the argument of \cref{prop:Dopen}. Whether the top map induces an equivalence of shapes is the question1132answered, for immersions, by the Smale--Hirsch theorem (\cref{thm:SH}).1133 1134\section*{Notes}1135 1136The weak and strong topologies and their properties are developed in Hirsch's Chapter~2 and, for1137$C^\infty$ and jets, in Golubitsky--Guillemin and Mather. The convenient setting of \cite{KM97} gives1138$C^\infty(M,N)$ the structure of an infinite-dimensional manifold whose smooth curves are the smooth families1139with locally compact support in the parameter; \cref{prop:strongpath} is the reason for that restriction.1140The comparison of the $D$-topology with the classical ones is \cite{CSW14}. \Cref{thm:oka} is called the1141smooth Oka principle by Sati and Schreiber; earlier special cases are due to Madsen and Weiss.1142 1143\section*{Exercises}1144 1145\begin{exercise}\label{exer:weakopen}1146Verify the claims of \cref{ex:weakfail}. Show also that $\Diff(\R)$ is not closed in $C^\infty_W(\R,\R)$.1147\end{exercise}1148 1149\begin{exercise}\label{exer:strongpath}1150Prove \cref{prop:strongpath} for $M=N=\R$. Deduce that the path component of $f$ in $C^0_S(\R,\R)$ is the set1151of maps that agree with $f$ outside a compact set.1152\end{exercise}1153 1154\begin{exercise}\label{exer:Dopen}1155Show that a plot of $[M,N]$ is continuous as a map into $C^1_W(M,N)$, and give an example of a smooth family1156$F\colon\R\times\R\to\R^2$ of maps for which $\{s:F_s\text{ is an embedding}\}$ is not open.1157\end{exercise}1158 1159\begin{exercise}\label{exer:sitting}1160Let $X$ be a smooth set. Show that concordance is an equivalence relation on $X(\pt)$.1161\end{exercise}1162 1163\begin{exercise}\label{exer:morseweak}1164A function $f\colon\R\to\R$ is Morse if $f''(x)\ne0$ wherever $f'(x)=0$. Show that the set of Morse functions1165is dense but not open in $C^\infty_W(\R,\R)$.1166\end{exercise}1167%% ---- end ch/03-function.tex ----1168%% ---- begin ch/04-transversality.tex ----1169\chapter{Sard's Theorem and Transversality}\label{ch:transversality}1170 1171This is the chapter on which the source turns. Its results say that the transverse cospans of1172\cref{thm:transpull} are dense among all cospans, and that the same holds in families. The proofs reduce, by1173one piece of linear algebra, to a single analytic theorem. The higher-categorical counterpart is a universal1174property: there is a smallest enlargement of $\Mfd$ in which all fibre products exist and the transverse ones1175are unchanged, and the genericity theorems say that every object of the enlargement is cobordant to an1176object of $\Mfd$.1177 1178\section{The Morse--Sard theorem}1179 1180\begin{theorem}[Morse--Sard]\label{thm:sard}\Scl[Morse, Sard, Brown; Hirsch 3.1.3]1181Let $f\colon M\to N$ be of class $C^r$ with $r>\max\{0,\dim M-\dim N\}$. Then the set of critical values of $f$1182has measure zero in $N$, and the set of regular values is residual.1183\end{theorem}1184 1185The proof is an induction on $\dim M$ by slicing with Fubini's theorem, with the differentiability hypothesis1186used to control the image of the set where many derivatives vanish. Nothing in this book adds to it. The1187theorem is the source of every statement of the form ``for almost all'' below.1188 1189\begin{corollary}\label{cor:noretract}\Scl[Hirsch 3.1.4]1190A compact manifold with non-empty boundary admits no smooth retraction onto its boundary.