Indexed text attachment
differential-topology-in-higher-categories.tex
Source for Differential Topology in Higher Categories
View paper · Download original file
Lines 881–980 of 6119. Source text is displayed literally.
881choose a bump function equal to one near the center and882supported compactly within the neighborhood. Add a883polynomial of degree at most $r$ times this bump to the884target coordinates of $f$. For small polynomial885coefficients all perturbed points still lie in the target886chart, since the change is made on a compact support.887Continue by $f$ outside the supports. At $x_i$ the888derivatives of the change are exactly the derivatives of889the chosen polynomial, and at all other chosen points the890jet is unchanged. Therefore the derivative of multijet891evaluation in the polynomial coefficients surjects onto892the product of the vertical tangent spaces of the jet893bundle at the configuration. This finite-dimensional894family is the input to multijet transversality arguments895\cite[Ch.~II]{golubitsky-guillemin}.896 897At order zero and $q=2$, the pair of values898$(f(x_1),f(x_2))$ can be tested against the diagonal of899$N\times N$. If this test is transverse, the double-point900locus away from the source diagonal is a submanifold of901dimension $2\dim M-\dim N$ when nonempty, by the transverse902pullback calculation of Chapter~\ref{ch:tangent}. The903rank condition is a one-point test. Accidental equality904of values is a two-point test. The distinction explains905why injectivity questions require multijets even when the906immersion condition is already settled.907\par908 909\bookpart{II}{Transversality and normal geometry}910\chapter{Critical values and smooth pullbacks}911\label{ch:sard}912 913A level equation \(f(x)=y\) asks for a pullback by the point914\(y:*\to N\). To identify this pullback with a smooth manifold915of the expected dimension, the crucial hypothesis is a rank916condition on the derivative along the fiber. Sard's theorem917makes that condition available for almost every choice of the918point. The point of view in this chapter is to keep the rank919test and the genericity argument distinct, so that each can920subsequently be applied to families.921 922\section{The fiber of a regular value}923 924Let \(M,N\) be finite-dimensional second-countable smooth925manifolds without boundary, with \(\dim M=m\) and \(\dim N=n\).926A point \(x\in M\) is \emph{critical} for a smooth map \(f\)927if \(df_x:T_xM\to T_{f(x)}N\) fails to be surjective.928The \emph{critical values} form the image of the critical929points. A regular value is a point outside this image, so a930point outside \(f(M)\) is a regular value with empty fiber.931 932\begin{theorem}[Regular preimages]933\label{thm:regular-preimage}934If \(y\) is a regular value, then \(f^{-1}(y)\) is an embedded935submanifold, of dimension \(m-n\) when nonempty. At a point936\(x\in f^{-1}(y)\) its tangent space is \(\ker df_x\).937More generally, for an embedded submanifold \(W\subseteq N\)938of codimension \(c\), suppose \(df_x(T_xM)+T_{f(x)}W=T_{f(x)}N\)939whenever \(f(x)\in W\). Then \(f^{-1}(W)\) is an embedded940submanifold of codimension \(c\) when nonempty, with tangent941space \(df_x^{-1}(T_{f(x)}W)\).942\end{theorem}943 944\begin{proof}945For the general statement choose a local submersion946\(q:U\to\mathbb R^c\) defining \(W\) near \(f(x)\), so that947\(W\cap U=q^{-1}(0)\). The displayed hypothesis says exactly948that \(d(qf)_x\) is surjective. In source coordinates whose949first \(c\) columns give an invertible minor, augment \(qf\)950by the remaining source coordinates. The derivative of the951augmented map is invertible, so by the inverse function theorem952its components are local coordinates. In these coordinates953the preimage is the zero set of the first \(c\) coordinates.954Differentiating the equations identifies its tangent space955with \(\ker d(qf)_x=df_x^{-1}(T_{f(x)}W)\). The case \(W=\{y\}\)956has \(c=n\). The local descriptions use the subspace topology957and thus assemble to the asserted embedded structures.958\end{proof}959 960The square in the theorem is a smooth pullback.961\[962\begin{tikzcd}[column sep=large]963 f^{-1}(W) \arrow[r] \arrow[d,hook] & W \arrow[d,hook]\\964 M \arrow[r,"f"'] & N .965\end{tikzcd}966\]967Indeed, a smooth test map into \(M\) whose image lands in968\(f^{-1}(W)\) factors smoothly through it. This follows from969the adapted coordinates in the proof. Testing against all970smooth manifolds gives the same pullback square of971representable sheaves. Such sheaves are \(0\)-truncated, so972their pullback in sheaves of spaces is also computed by the973ordinary fiber-product condition. Representability of that sheaf pullback requires a separate974argument. For975\(f(x)=x^2\) and \(y=0\), the fiber sheaf is represented by a976point, but \(df_0=0\) and its kernel is one-dimensional.977A derived enhancement of a smooth equation can record this978failure of the differential rank test.979 980\section{Sard's theorem in the smooth case}