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 2531–2630 of 3851. Source text is displayed literally.

2531that the Euler characteristic modulo $2$ is a ring homomorphism $\mathfrak N_*\to\Z/2$, and that $\RP^{2m}$ is2532not a boundary.2533\end{exercise}2534 2535\begin{exercise}\label{exer:ranks}2536Compute $\dim\mathfrak N_n$ for $n\le10$ from \cref{thm:thom}.2537\end{exercise}2538 2539\begin{exercise}\label{exer:sw}2540Verify the Stiefel--Whitney numbers in \cref{ex:sw}.2541\end{exercise}2542 2543\begin{exercise}\label{exer:lowdim}2544Show that $\mathfrak N_1=0$ and that the Klein bottle is a boundary. Using the classification of surfaces,2545conclude that $\mathfrak N_2=\Z/2$, generated by $\RP^2$.2546\end{exercise}2547 2548\begin{exercise}\label{exer:framed0}2549Show directly that the Pontryagin--Thom collapse of a single point of $\R^k$ with the standard framing is a2550map $S^k\to S^k$ of degree $1$, and that reversing one framing vector gives degree $-1$. Conclude2551$\Omega^{\mathrm{fr}}_0\cong\Z$.2552\end{exercise}2553%% ---- end ch/09-cobordism.tex ----2554%% ---- begin ch/10-isotopy.tex ----2555\chapter{Isotopy}\label{ch:isotopy}2556 2557Hirsch's eighth chapter is about paths in spaces of embeddings: an isotopy of a compact submanifold extends2558to an isotopy of the ambient manifold; embeddings of discs are unique up to isotopy; manifolds glued by2559isotopic maps are diffeomorphic. Each statement is about $\pi_0$ or about lifting paths. Each has a form2560that concerns the whole homotopy type of a space of embeddings, and those forms are where the subject went2561after 1976. This chapter states them and ends with the functor calculus that organizes spaces of2562embeddings in high codimension. The companion volume uses this chapter more than any other.2563 2564\section{Isotopies and vector fields}2565 2566\begin{definition}\label{def:isotopy}2567An \term{isotopy} of embeddings of $V$ in $M$ is a plot $I\to\Emb(V,M)$: a smooth map2568$F\colon V\times I\to M$ with every $F_t$ an embedding. A \term{diffeotopy} of $M$ is an isotopy2569$F\colon M\times I\to M$ of diffeomorphisms with $F_0=\id$. Embeddings are \term{isotopic} if joined by an2570isotopy, and \term{ambient isotopic} if one is carried to the other by the end of a diffeotopy.2571\end{definition}2572 2573So isotopy is concordance of points of the smooth set $\Emb(V,M)$ in the sense of \cref{def:shape}. (The2574word concordance is used differently in the classical literature on embeddings, for embeddings of2575$V\times I$ in $M\times I$ that need not preserve levels; we avoid that usage.)2576 2577\begin{theorem}\label{thm:flow}\Scl[Hirsch 8.1.1]2578A time-dependent vector field $G$ on $M$ of bounded velocity, for instance one with compact support,2579generates a unique diffeotopy $F$ with $\partial_tF_t(x)=G(F_t(x),t)$. Every diffeotopy with compact support2580arises in this way from its velocity field.2581\end{theorem}2582 2583So paths in $\Diff_c(M)$ starting at the identity are the same thing as paths in the vector space2584$\mathfrak X_c(M)$ of compactly supported vector fields, which is the statement that $\Diff_c(M)$ is a Lie2585group with that Lie algebra \cite{KM97}. The hypothesis on the velocity cannot be dropped: the field2586$x^2\partial_x$ on $\R$ has solutions $x_0/(1-x_0t)$, which leave every compact set in finite time.2587 2588\section{Isotopy extension}2589 2590\begin{theorem}[Isotopy extension]\label{thm:isoext}\Scl[Thom, Palais, Cerf; Hirsch 8.1.3]2591Let $V\subset M$ be a compact submanifold and $F\colon V\times I\to M$ an isotopy with image in2592$M\setminus\partial M$. Then $F$ extends to a diffeotopy of $M$ with compact support.2593\end{theorem}2594 2595The proof extends the velocity field of $F$, which is defined along the track of the isotopy in2596$M\times I$, to a compactly supported field on $M\times I$ using a tubular neighbourhood of the track, and2597integrates. The extension is a contractible choice. Carrying parameters through the argument gives the2598stronger statement.2599 2600\begin{theorem}[Restriction is a fibration]\label{thm:palais}\Scl[Palais \cite{Pal60}; Cerf; Lima]2601Let $V\subset M$ be a compact submanifold. The restriction map2602$\Diff_c(M)\to\Emb(V,M)$ is a locally trivial fibre bundle with fibre the group $\Diff_c(M\text{ rel }V)$ of2603diffeomorphisms fixing $V$. More generally, for compact $V\subset V'$ the restriction2604$\Emb(V',M)\to\Emb(V,M)$ is a fibration.2605\end{theorem}2606 2607\Cref{thm:isoext} is the path-lifting property of this map; \cref{thm:palais} says that all homotopies2608lift. The consequence is a long exact sequence of homotopy groups. For $V$ a point it reads2609\[\cdots\to\pi_1\Diff(M)\to\pi_1(M)\to\pi_0\Diff(M\text{ rel }x)\to\pi_0\Diff(M)\to\pi_0(M),\]2610and the middle map sends a loop to the diffeomorphism that pushes the base point around it. For compact2611$V$ the quotient of $\Emb(V,M)$ by $\Diff(V)$ is the space of submanifolds of $M$ diffeomorphic to $V$, and the2612theorem says that the group $\Diff_c(M)$ acts on it with orbits that are unions of components.2613 2614Compactness is essential. Let $F_t\colon\R\to\R^3$ be a long knot whose knotted part sits at distance2615$1/(1-t)$ along the line, for $t<1$, and let $F_1$ be the straight line. This is an isotopy in the sense of2616\cref{def:isotopy}, smooth because on every compact set it is eventually constant, from a knotted line to an2617unknotted one; it cannot be ambient (\cref{exer:noncompact}). In the language of \cref{ch:function}, it is a2618plot of the mapping object that is not a path in the strong topology (\cref{prop:strongpath}).2619 2620\section{Embeddings of discs}2621 2622\begin{theorem}[Discs]\label{thm:discs}\Scl[Cerf, Palais; Hirsch 8.3.1 on $\pi_0$]2623For an $n$-manifold $M$, the map $\Emb(D^n,M)\to\Fr(TM)$, $f\mapsto(f(0),df_0)$, to the frame bundle is a2624homotopy equivalence. More generally $\Emb(D^k,M)$ is equivalent to the bundle of $k$-frames in $TM$, and2625$\Emb(\coprod_jD^n,M)$ to the pullback of $\Fr(TM)^j$ to the configuration space $\Conf_j(M)$.2626\end{theorem}2627 2628The proof is the Alexander-type shrinking of \cref{thm:tubcontract}: $f_t(x)=f(tx)$ rescaled deforms an2629embedding to its linear part (\cref{exer:discRn}). On components this is Hirsch's theorem that two2630embeddings of $D^k$ in a connected $M$ are isotopic, provided, when $k=n$ and $M$ is orientable, they induce