Indexed text attachment
algebraic-and-differential-topology-in-higher-categories.tex
Source for Algebraic and Differential Topology in Higher Categories
View paper · Download original file
Lines 661–760 of 10051. Source text is displayed literally.
661$1\to2\to3\to1$ is a cycle $z$. The boundary of the sum of the ten triangles, all oriented alike, runs round the662hexagon and is $2z$. So $z$ is not a boundary but $2z$ is: a torsion class, visible over $\Fp_2$ and invisible over663$\Fp_3$ and over the rationals.}\label{fig:rp2}664\end{center}665 666\section{Why chains: the free abelian group on a space}667 668The zero-dimensional case is elementary and shows the pattern.669 670\begin{proposition}\label{prop:h0}671\Scl[\src{I,\,\S5, Thms.~6--8}] $H_0(K)$ is the free abelian group on the set $\pi_0(K)$ of components.672\end{proposition}673 674\noindent \chk{H5} The functor $\pi_0$ is left adjoint to the inclusion of sets into complexes, and the free675abelian group is left adjoint to the forgetful functor, so $H_0$ is a composite of left adjoints: it is the676universal way of turning a complex into an abelian group. The higher groups are the same statement with sets677replaced by homotopy types.678 679\begin{theorem}[Dold--Kan and Dold--Thom]\label{thm:doldkan}680\Scl[\cite{Dol58,GoJa99,DoTh58}] \emph{(a)} Normalized chains give an equivalence between simplicial abelian groups681and non-negatively graded chain complexes. \emph{(b)} For a simplicial set $X$ let $\Z[X]$ be the free simplicial682abelian group on it. Then $H_n(X)=\pi_n\,\Z[X]$.683\end{theorem}684 685\begin{theorem}[Chains as a left adjoint]\label{thm:leftadjoint}686\Srf[\cite{Lur17}] Let $\Spc$ be the $\infty$-category of homotopy types and $\DD(\Z)$ the derived687$\infty$-category of abelian groups. The functor $C_*\colon\Spc\to\DD(\Z)$ is the unique colimit-preserving functor688sending a point to $\Z$. It is left adjoint to the functor that sends a chain complex to its underlying space.689\end{theorem}690 691\noindent So the chain complex is not a device that happens to produce invariants. It is the free module over692$\Z$ on a homotopy type, as $\Z[S]$ is the free module on a set $S$, and the boundary operator is what freeness693looks like after \Cref{thm:doldkan}. Part Two will make the same construction with $\Z$ replaced by another ring,694and \Cref{ch:stems} compares the two.695 696\section{Euler and Poincar\'e}697 698\begin{theorem}[Euler--Poincar\'e]\label{thm:ep}699\Scl[\src{I,\,\S6, Thm.~9}] If $\alpha_r$ is the number of $r$-simplices of $K$, then700$\sum_r(-1)^r\alpha_r=\sum_r(-1)^r\beta_r$, and the same holds with Betti numbers over any field.701\end{theorem}702\begin{proof}703Rank is additive on the short exact sequences $0\to Z_r\to C_r\to B_{r-1}\to0$ and $0\to B_r\to Z_r\to H_r\to0$,704and the ranks of the $B_r$ cancel in the alternating sum.705\end{proof}706 707\noindent \chk{H3} The proof uses one property of rank, additivity, and the Euler characteristic is the708universal invariant with that property: it is the class of $C_*(K)$ in the Grothendieck group of perfect709complexes, which is $\Z$. Additivity for unions is the same fact one level down:710$\chi(A\cup B)=\chi(A)+\chi(B)-\chi(A\cap B)$. \chk{H7} In \Cref{ch:traces} the Euler characteristic becomes the711trace of the identity, and \Cref{thm:ep} a special case of a formula for traces.712 713\section{Cones, suspensions, products}714 715\begin{proposition}\label{prop:cone}716\Scl[\src{I,\,\S8}] \emph{(a)} A cone has the homology of a point. \emph{(b)} The suspension $\Sigma K$, two cones717glued along $K$, has $\tilde H_{r+1}(\Sigma K)\cong\tilde H_r(K)$.718\end{proposition}719 720\begin{center}721\begin{tikzpicture}[scale=0.95]722\begin{scope}723\draw[thick] (0,0) ellipse (1.1 and 0.32);724\node at (0,-0.75) {\small$K$};725\end{scope}726\begin{scope}[xshift=4.2cm]727\fill[fwd!10] (-1.1,0)--(0,1.7)--(1.1,0) arc (0:-180:1.1 and 0.32);728\draw[thick] (-1.1,0)--(0,1.7)--(1.1,0);729\draw[thick] (-1.1,0) arc (180:360:1.1 and 0.32);730\draw[thick,densely dashed] (1.1,0) arc (0:180:1.1 and 0.32);731\fill (0,1.7) circle (1.7pt);732\node at (0,-0.75) {\small cone $CK$};733\end{scope}734\begin{scope}[xshift=8.4cm]735\fill[fwd!10] (-1.1,0)--(0,1.7)--(1.1,0)--(0,-1.7)--cycle;736\draw[thick] (-1.1,0)--(0,1.7)--(1.1,0)--(0,-1.7)--cycle;737\draw[thick] (-1.1,0) arc (180:360:1.1 and 0.32);738\draw[thick,densely dashed] (1.1,0) arc (0:180:1.1 and 0.32);739\fill (0,1.7) circle (1.7pt); \fill (0,-1.7) circle (1.7pt);740\node at (2.1,0) {\small$\Sigma K$};741\end{scope}742\end{tikzpicture}743\figcap{A complex, its cone and its suspension. The cone kills all homology; the suspension moves it up by one744dimension.}745\end{center}746 747\noindent The cones on the six complexes are acyclic; the suspension of the torus has Betti numbers $(1,0,2,1)$,748and the suspension of the projective plane has $H_2=\Z/2$ and nothing else in positive degrees. \chk{H6} Statement749(b) says that homology cannot tell a complex from its suspension except by a shift. Invariants with this property750are called \emph{stable}; homotopy groups do not have it, and the second monograph of the source is about the751range of dimensions in which they acquire it (\Cref{ch:framed}).752 753\begin{theorem}[K\"unneth]\label{thm:kunneth}754\Scl[\cite{Hat02}] Over a field, $H_*(K\times L)\cong H_*(K)\otimes H_*(L)$.755\end{theorem}756 757\noindent With the product triangulation of \Cref{thm:sset}, the torus $(\Z/3)^2$ has Betti numbers $(1,2,1)$, and758the $3$-torus $(\Z/4)^3$, with $64$ vertices and $384$ tetrahedra, has $(1,3,3,1)$ over $\Fp_2$, $\Fp_3$ and a759large prime field. \chk{H8} In the language of \Cref{thm:leftadjoint}, $C_*$ is symmetric monoidal: the free760module on a product is the tensor product of the free modules.