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 1591–1690 of 3851. Source text is displayed literally.

1591 1592A bundle is \term{orientable} if its transition functions can be chosen in $\GL_k^+(\R)$; an1593\term{orientation} is such a choice up to equivalence. In terms of classifying maps, an orientation is a1594lift of $M\to B\Or(k)$ through the double cover $B\SO(k)\to B\Or(k)$, and the obstruction to its existence is1595the class $w_1(\xi)\in H^1(M;\Z/2)$ pulled back from the generator of $H^1(B\Or(k);\Z/2)$. A manifold is1596orientable if its tangent bundle is.1597 1598\begin{definition}\label{def:theta}1599A \term{tangential structure} is a map of spaces $\theta\colon B\to B\Or(d)$. A \term{$\theta$-structure} on a1600$d$-manifold $M$ is a lift of the classifying map $\tau_M\colon M\to B\Or(d)$ of $TM$ through $\theta$.1601\end{definition}1602 1603Orientations ($B=B\SO(d)$), framings ($B=\pt$), spin structures, almost complex structures, and maps to a1604fixed background space $X$ ($B=B\Or(d)\times X$) are all of this form. Because $\theta$-structures pull back1605along open embeddings and are defined by a lifting problem in spaces, the stack of $d$-manifolds with1606$\theta$-structure is again an object to which \cref{thm:concordance} applies; this is how the tangential1607structures of \cref{ch:cobordism} enter.1608 1609\begin{theorem}\label{thm:nonorient}\Scl[Hirsch 4.4.7]1610A closed non-orientable $n$-manifold does not embed in a simply connected $(n+1)$-manifold. In particular1611$\RP^{2n}$ does not embed in $\R^{2n+1}$.1612\end{theorem}1613 1614\begin{proof}1615Let $M\subset N$ be such an embedding with normal line bundle $\nu$. A loop in $N$ transverse to $M$ meets1616it in an even number of points, since the count modulo $2$ is a homotopy invariant1617(\cref{thm:intersection}) and the loop is null-homotopic. A loop in $M$ along which $\nu$ is non-trivial1618can be pushed off $M$ to a loop meeting $M$ once. So $\nu$ is trivial, and1619$TM\oplus\nu\cong TN|_M$ gives $w_1(TM)=w_1(TN|_M)=0$, as $N$ is simply connected.1620\end{proof}1621 1622\section{The tangent bundle as a map}1623 1624The tangent bundle of an $m$-manifold is a point $\tau_M$ of $\Map(M,B\Or(m))$. For a submanifold1625$M\subset\R^{m+k}$ the classifying map can be taken to be the Gauss map $x\mapsto T_xM\in\Gr_m(\R^{m+k})$, and1626the normal bundle is classified by $x\mapsto(T_xM)^\perp$. The normal bundle depends on the embedding, but1627its class in $[M,B\Or]$, where $B\Or=\colim_kB\Or(k)$, does not; this \term{stable normal bundle} $\nu_M$ is1628inverse to $\tau_M$ under direct sum, and \cref{ch:tubular} shows that it is well defined as a point of a1629contractible space of choices. Characteristic classes of $M$ are the pullbacks along $\tau_M$ or $\nu_M$1630of classes in $H^*(B\Or;\Z/2)=\Z/2[w_1,w_2,\dots]$. Every invariant of manifolds used in1631\cref{ch:degree,ch:cobordism} is obtained from $\tau_M$, as a consequence of the remark after1632\cref{prop:jets}.1633 1634\section*{Notes}1635 1636Steenrod's book is the classical source for classification by maps to Grassmannians; Hirsch's treatment by1637transversality is his own. The stack-theoretic formulation is standard; see \cite{Sch13}. The equivalence1638with projective modules is Swan's, after Serre. \Cref{thm:concordance} has a long prehistory: Brown1639representability, Gromov's sheaf-theoretic h-principle, and Madsen and Weiss's use of concordance in the1640proof of the Mumford conjecture.1641 1642\section*{Exercises}1643 1644\begin{exercise}\label{exer:mobius}1645Show that the tautological line bundle over $\RP^1$ is not trivial, and that its total space is the open1646Möbius band.1647\end{exercise}1648 1649\begin{exercise}\label{exer:swan}1650Let $e\colon M\to\R^{s\times s}$ be smooth with $e^2=e$ and constant rank $k$. Show that the image of $e$ is a1651vector bundle whose module of sections is $e\,C^\infty(M)^s$, a direct summand of a free module.1652\end{exercise}1653 1654\begin{exercise}\label{exer:TS}1655Show that $TS^n\oplus\R$ is trivial, and conclude that the composite $S^n\to B\Or(n)\to B\Or$ classifying the1656stable tangent bundle is null-homotopic.1657\end{exercise}1658 1659\begin{exercise}\label{exer:forms}1660Show that two closed $1$-forms on $M$ are concordant (are the restrictions to $M\times\{0\}$ and1661$M\times\{1\}$ of a closed $1$-form on $M\times\R$) if and only if they differ by an exact form.1662\end{exercise}1663 1664\begin{exercise}\label{exer:lines}1665Show that isomorphism classes of line bundles over $M$ correspond to $H^1(M;\Z/2)=\Hom(\pi_1M,\Z/2)$, and1666count them for the torus and for the Klein bottle. Which one is the determinant of the tangent bundle?1667\end{exercise}1668%% ---- end ch/05-bundles.tex ----1669%% ---- begin ch/06-tubular.tex ----1670\chapter{Tubular Neighbourhoods, Collars, and Spaces of Choices}\label{ch:tubular}1671 1672The second half of Hirsch's fourth chapter proves that a submanifold has a tubular neighbourhood, that the1673boundary of a manifold has a collar, and that each is unique up to isotopy. The uniqueness statements are1674what allow later constructions (connected sums, gluing of cobordisms, the Pontryagin--Thom collapse) to be1675called well defined. ``Unique up to isotopy'' says that a space of choices is connected. In every case in1676this chapter the space of choices is in fact contractible, and that stronger statement is what makes the1677constructions functorial.1678 1679\section{Tubular neighbourhoods}1680 1681\begin{definition}\label{def:tub}1682Let $M\subset N$ be a submanifold, both without boundary. A \term{tubular neighbourhood} of $M$ in $N$ is a1683vector bundle $\xi=(p,E,M)$ with an embedding $\varphi\colon E\to N$ onto an open set that restricts to the1684identity on the zero section. The \term{normal bundle} is $\nu=\nu(M,N)=TN|_M/TM$.1685\end{definition}1686 1687The differential of $\varphi$ along the zero section identifies $\xi$ with $\nu$, so one may take $\xi=\nu$.1688 1689\begin{theorem}[Tubular neighbourhoods]\label{thm:tubular}\Scl[Hirsch 4.5.2, 4.5.3]1690Every submanifold $M\subset N$ has a tubular neighbourhood, and any two are isotopic relative to $M$.