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 1881–1980 of 10051. Source text is displayed literally.
1881\begin{scope}1882\fill[fwd!10] (0,0)--(2,0)--(2,1.6)--cycle; \fill[bwd!10] (0,0)--(0,1.6)--(2,1.6)--cycle;1883\draw[thick] (0,0)--(2,0)--(2,1.6)--(0,1.6)--cycle; \draw[thick] (0,0)--(2,1.6);1884\foreach \p/\l/\pos in {(0,0)/v_0/below,(2,0)/v_1/below,(0,1.6)/v_0'/above,(2,1.6)/v_1'/above} {\fill \p circle (1.7pt); \node[\pos] at \p {\small$\l$};}1885\node at (1,-0.95) {\small two triangles};1886\end{scope}1887\begin{scope}[xshift=5.2cm]1888\coordinate (a) at (0,0); \coordinate (b) at (2.3,0); \coordinate (c) at (1.3,0.75);1889\coordinate (A) at (0,1.9); \coordinate (B) at (2.3,1.9); \coordinate (C) at (1.3,2.65);1890\fill[fwd!10] (a)--(b)--(B)--(A)--cycle;1891\draw[thick] (a)--(b)--(B)--(A)--cycle; \draw[thick] (A)--(C)--(B);1892\draw[thick,densely dashed] (a)--(c)--(b) (c)--(C);1893\draw[thick,fwd] (a)--(B); \draw[thick,fwd,densely dashed] (a)--(C) (b)--(C);1894\foreach \p/\l/\pos in {a/v_0/below,b/v_1/below,c/v_2/right,A/v_0'/left,B/v_1'/right,C/v_2'/above} {\fill (\p) circle (1.7pt); \node[\pos] at (\p) {\small$\l$};}1895\node at (1.15,-0.95) {\small three tetrahedra};1896\end{scope}1897\end{tikzpicture}1898\figcap{The prisms over an edge and over a triangle. The three tetrahedra are $[v_0,v_0',v_1',v_2']$,1899$[v_0,v_1,v_1',v_2']$ and $[v_0,v_1,v_2,v_2']$; the added diagonals are in colour.}\label{fig:prism}1900\end{center}1901 1902\begin{theorem}[Prism operator]\label{thm:prism}1903\Scl[\src{I,\,\S14}] Define $P\colon C_r(K)\to C_{r+1}(K\times[0,1])$ by1904$P[v_0,\dots,v_r]=\sum_{i=0}^r(-1)^i[v_0,\dots,v_i,v_i',\dots,v_r']$. Then1905\[ \pd P+P\pd=i_1-i_0, \]1906where $i_0,i_1$ are the inclusions of the bottom and the top.1907\end{theorem}1908 1909\noindent \chk{P1} The cylinder on the projective plane has homology $\Z,\Z/2,0,0$, that of the projective plane,1910and the cylinder on the M\"obius band has Betti numbers $(1,1,0,0)$. \chk{P2}1911 1912\begin{theorem}[Homotopy invariance]\label{thm:hinv}1913\Scl[\src{I,\,\S13--15, Thms.~20--24}] Homotopic maps of polyhedra induce the same homomorphisms of homology1914groups. Homotopy equivalent polyhedra have isomorphic homology groups.1915\end{theorem}1916\begin{proof}1917A homotopy is a map $F\colon|K|\times[0,1]\to|L|$ with $F i_0=f$ and $F i_1=g$. Approximate it simplicially; then1918$F_\#P$ is a chain homotopy from $f_\#$ to $g_\#$, and on a cycle $z$ the difference $g_\#z-f_\#z=\pd F_\#Pz$ is a1919boundary.1920\end{proof}1921 1922\noindent The M\"obius band and its core circle both have Betti numbers $(1,1)$; a disc has those of a point.1923\chk{P4} A combinatorial form: two simplicial maps are \term{contiguous} if the images of each simplex under the1924two together span a simplex. Any two maps $\sdv K\to K$ that send each vertex $\sigma$ to a vertex of $\sigma$ are1925contiguous, and they induce the same map on homology; this was checked for twelve random pairs on four1926complexes. \chk{P3}1927 1928\section{What a chain homotopy is}1929 1930\[\begin{tikzcd}[column sep=5em]1931{|K|}\arrow[r,bend left=35,"f",""{name=U,below}]\arrow[r,bend right=35,"g"',""{name=D,above}] & {|L|}1932\arrow[Rightarrow,from=U,to=D,"F"]1933&[1em]1934C_*(K)\arrow[r,bend left=35,"f_\#",""{name=V,below}]\arrow[r,bend right=35,"g_\#"',""{name=E,above}] & C_*(L)1935\arrow[Rightarrow,from=V,to=E,"F_\#P"]1936\end{tikzcd}\]1937 1938\noindent A homotopy is a $2$-cell between maps, and the theorem sends it to a $2$-cell between chain maps. The1939construction continues: a homotopy between homotopies gives a map $C_r(K)\to C_{r+2}(L)$ with the analogous1940identity, and so on. Spaces with maps, homotopies and higher homotopies form an $\infty$-category, as do chain1941complexes with chain maps, chain homotopies and higher ones, and the assertion that all of these constructions1942are compatible is:1943 1944\begin{theorem}\label{thm:infty}1945\Srf[\cite{Lur17}; \Cref{thm:leftadjoint}] Chains extend to a functor of $\infty$-categories1946$C_*\colon\Spc\to\DD(\Z)$, and homology is its composite with $H_*\colon\DD(\Z)\to\cat{grAb}$.1947\end{theorem}1948 1949\noindent \Cref{thm:hinv} is the shadow of this on homotopy categories. The prism operator itself is a product:1950$P\sigma$ is the product of $\sigma$ with the fundamental chain of the interval, under the shuffle map1951$C_*(K)\otimes C_*(L)\to C_*(K\times L)$ of Eilenberg and Zilber \cite{EiZi53}. The identity of \Cref{thm:prism}1952is the Leibniz rule for that product, since the boundary of the interval is top minus bottom.1953 1954\section{Degree}1955 1956For a map $f\colon S^n\to S^n$ the homomorphism $f_*$ on $H_n(S^n)=\Z$ is multiplication by an integer, the1957\term{degree} \src{I,\,\S15}. By \Cref{thm:hinv} it is a homotopy invariant. Simplicially it is the signed number1958of cells mapped onto a given cell, which is how the degree of the labelling was counted in \Cref{thm:sperner}.1959 1960\begin{proposition}\label{prop:degree}1961\Scl \emph{(a)} $\deg(g\circ f)=\deg g\cdot\deg f$. \emph{(b)} The antipodal map of $S^n$ has degree1962$(-1)^{n+1}$. \emph{(c)} The suspension of a map has the same degree.1963\end{proposition}1964 1965\noindent For (b), on the boundaries of the cross-polytopes of dimensions $2,3,4$ the antipodal map acts on top1966homology by $1,-1,1$. \chk{L6} Statement (c) is \Cref{prop:cone}\emph{(b)} applied to maps. The converse of1967homotopy invariance for spheres, that maps of the same degree are homotopic, is Hopf's theorem; it is not1968accessible to the methods of Part One and is proved in \Cref{ch:hopf} with framed points.1969 1970\section{What homology cannot see}1971 1972Homology is a functor on the homotopy category, but it is far from faithful. The simplest witness is the1973\term{Hopf map} $h\colon S^3\to S^2$, $(z_1,z_2)\mapsto z_1/z_2$. Since $S^3$ and $S^2$ have no homology in a common1974positive degree, $h_*$ is the same as the map induced by a constant, in every degree and with every coefficient1975group. Yet $h$ is not homotopic to a constant (\Cref{thm:hopf}). The invariant that detects it is not a property1976of the homomorphisms $h_*$ but of how the chains fit together: the preimages of two points are two circles in1977$S^3$, and they are linked.1978 1979In the language of \Cref{thm:infty}: the functor $C_*$ is faithful on neither objects nor morphisms, and what it1980forgets is measured by replacing the coefficient ring $\Z$ with a larger one through which every such functor