Indexed text attachment
differential-topology-in-higher-categories.tex
Source for Differential Topology in Higher Categories
View paper · Download original file
Lines 2079–2178 of 6119. Source text is displayed literally.
2079leaves its Thom class and pulls back the second class by its2080zero section. The latter pullback is $e(E_{\mathrm{exc}})$.2081Extending the resulting relative class to $B$ and forgetting2082the complement gives precisely $j_!e(E_{\mathrm{exc}})$.2083\end{proof}2084 2085An ordinary clean pullback displays the manifold $Z$, but2086its tangent sequence also contains the cokernel bundle.2087Transversality makes that cokernel vanish and recovers the2088rank-zero Thom formula. A tangent-complex description keeps2089it in a second degree, as developed later for derived2090intersections. The formula above identifies a classical2091contribution of that bundle without replacing the clean2092intersection by a finite set.2093 2094\section{Self-intersection and tangent zeros}2095 2096Take $i$ to be the inclusion of $A$ itself. The intersection2097inside $B=A$ is all of $A$, and the excess bundle is2098$\nu_{A/W}$. Thus restricting $\operatorname{PD}_W[A]$2099to $A$ gives the Euler class of the normal bundle. In2100$\mathbb{CP}^2$, let $A$ be the projective line $z_0=0$.2101The line $z_1=0$ is isotopic to $A$, through the rotations of2102the first two homogeneous coordinates. The two lines meet2103in one point $[0:0:1]$. In the affine chart at that point2104they are complementary complex coordinate lines. Complex2105linear identifications preserve the underlying real2106orientation, so the intersection sign is positive. The2107normal Euler number of $A$ is therefore $1$. The2108self-intersection can be computed by a transverse motion2109even though the original pullback is a two-dimensional2110manifold.2111 2112When $E=TB$ for a closed oriented manifold, a vector field2113with nondegenerate zeros computes the tangent Euler number2114by the same formula. Take a Morse function and a metric,2115using its gradient as the section. At a critical point2116of index $\lambda$, the derivative of the gradient is the2117inverse metric composed with the Hessian, so its determinant2118sign is $(-1)^\lambda$. Each corresponding Morse handle2119contributes that same sign to the Euler characteristic2120by Chapter~\ref{ch:morse}. Thus2121$\langle e(TB),[B]\rangle=\chi(B)$, the2122Poincar\'e--Hopf equality in the transverse setting2123\cite[Ch.~5]{hirsch}. For example, a height function of2124$S^2$ has one minimum and one maximum. Both indices are2125even, and the two contributions give $2$.2126\par2127\chapter{Morse functions and handle filtrations}2128\label{ch:morse}2129 2130The sublevel sets of a smooth real-valued function form an2131increasing family of spaces. A regular band contributes a2132product, while an isolated nondegenerate critical point2133adds a handle. This turns the real-valued function into a2134description of the topology of its source. We use the regular2135and critical level arguments of differential topology2136\cite[Ch.~6]{hirsch} to describe the filtration and its2137homotopy pushouts explicitly.2138 2139\section{The Hessian and the local form}2140 2141Let $f:M\to\mathbb R$ be smooth and let $p$ satisfy $df_p=0$.2142Second partial derivatives in a chart define a symmetric2143bilinear form on $T_pM$. Under a change of coordinates2144the Hessian is conjugated by the Jacobian, since the term2145involving the second derivatives of the coordinate change2146is multiplied by $df_p=0$. A critical point is2147\emph{nondegenerate} if this form is nonsingular, and its2148\emph{index} is the dimension of a maximal negative definite2149subspace. A Morse function has only nondegenerate critical2150points. They are isolated by the inverse function theorem2151applied to the coordinate components of $df$. For a compact2152source the critical set is closed and therefore finite.2153 2154\begin{theorem}[Morse local form]2155\label{thm:morse-lemma}2156If $p$ has index $\lambda$ in an $n$-manifold, there are2157local coordinates $(u,v)$ centered at $p$, of dimensions2158$\lambda$ and $n-\lambda$, such that2159\[2160 f(u,v)=f(p)-\|u\|^2+\|v\|^2 .2161\]2162\end{theorem}2163 2164\begin{proof}2165In coordinates $x$ near $p=0$, Taylor's integral formula2166writes the difference as $x^{\mathsf T}A(x)x$ with2167\[2168 A(x)=\int_0^1(1-t)\,D^2 f(tx)\,dt .2169\]2170This is a smooth symmetric matrix and $A(0)$ is nonsingular.2171A fixed linear change of basis makes $A(0)$ diagonal with2172entries $-1$ and $1$. In a neighborhood of $0$ all successive2173pivots in the symmetric elimination of $A(x)$ are nonzero.2174Eliminating one row and column at a time is the completion2175of a square, using only division by the current nonzero2176pivot. The signs of the pivots are locally constant. Taking2177square roots of their absolute values gives a smooth2178factorization $A(x)=B(x)^{\mathsf T} D B(x)$, where $D$ is the