Indexed text attachment
connections-curvature-higher-geometry.tex
Source for Connections, Curvature, and Higher Geometry: From covariant differentiation to differential cohomology
View paper · Download original file
Lines 2753–2852 of 4453. Source text is displayed literally.
2753\(V\) is abelian. There are no arrows between distinct \(g\)'s, but2754tensoring with \(g\) acts on the automorphism arrows by the given2755representation. This provides a model in which the two-form2756connection takes values in an abelian group and still depends on2757transport by a nonabelian one-form connection.2758\end{example}2759\chapter{Two-connections and their curvatures}2760\label{ch:2connections}2761 2762A strict two-group assigns a boundary in \(G\) to each arrow2763label in \(H\). On a small surface, the path transport around the2764boundary is controlled by the curvature of an ordinary2765connection. A two-form suitable for surface transport must have2766that curvature as its infinitesimal boundary. This yields a2767compatibility condition between the one- and two-form data.2768 2769\section{Forms for a differential crossed module}2770 2771Differentiate the crossed module of Chapter~\ref{ch:crossed-modules}.2772Write \(\mathfrak h,\mathfrak g\) for the Lie algebras,2773\(\partial_*\) for the differential of the boundary, and2774\(X\triangleright b\) for the induced action of \(\mathfrak g\) on2775\(\mathfrak h\) by derivations. The differentiated identities are2776\[2777 \partial_*(X\triangleright b)=[X,\partial_*b],2778 \qquad2779 (\partial_*a)\triangleright b=[a,b].2780\]2781They extend to forms by combining wedge of scalar forms with2782the indicated bracket or action on coefficients. In particular,2783brackets of forms are graded antisymmetric. When both inputs have2784degree one, the factor \(\tfrac12\) in \(\tfrac12[A,A]\)2785accounts for the two cross terms.2786 2787On a trivialization over \(U\), a pair of potentials consists of2788\(A\in\Omega^1(U;\mathfrak g)\) and2789\(B\in\Omega^2(U;\mathfrak h)\). Write \(D_A=d+A\triangleright(-)\)2790on \(\mathfrak h\)-valued forms and \(D_A=d+[A,-]\) on2791\(\mathfrak g\)-valued forms. Our conventions are2792\begin{align}2793 F_A&=dA+\tfrac12[A,A],\label{eq:two-f}\\2794 K&=F_A-\partial_* B,\label{eq:two-fake}\\2795 \mathcal H&=D_AB.\label{eq:two-h}2796\end{align}2797The form \(K\) is the fake curvature and \(\mathcal H\) is the2798three-form curvature. We reserve the term \emph{fake-flat} for2799\(K=0\). In the path two-functor model this condition is part of2800the local correspondence; a pair of forms can also be considered2801without imposing it~\cite{baez-schreiber,schreiber-waldorf-surface}.2802 2803\begin{proposition}[Higher Bianchi identities]2804For a pair of potentials with conventions \eqref{eq:two-f}--\eqref{eq:two-h},2805\[2806 D_A K=-\partial_*\mathcal H,\qquad2807 D_A\mathcal H=K\triangleright B.2808\]2809If \(K=0\), then \(\mathcal H\) takes values in \(\ker\partial_*\)2810and is \(D_A\)-closed there. The operator \(D_A\) is flat on2811\(\ker\partial_*\).2812\end{proposition}2813 2814\begin{proof}2815Equivariance of \(\partial_*\) gives2816\(D_A\partial_*B=\partial_*D_AB\). The ordinary Bianchi identity2817\(D_AF_A=0\), applied to \eqref{eq:two-fake}, gives the first formula.2818On an \(\mathfrak h\)-valued form \(u\), expansion gives2819\[2820 D_A^2u2821 =dA\triangleright u+A\triangleright(A\triangleright u)2822 =F_A\triangleright u.2823\]2824The two terms involving \(du\) cancel, and the representation2825identity makes \(A\triangleright(A\triangleright u)\) equal to2826\(\tfrac12[A,A]\triangleright u\). In particular,2827\[2828 D_A\mathcal H2829 =K\triangleright B+(\partial_*B)\triangleright B2830 =K\triangleright B+[B,B].2831\]2832Here \([B,B]=0\) because \(B\) has even degree and the coefficients2833are over \(\mathbb R\). When \(K=0\), the first identity gives2834\(\partial_*\mathcal H=0\). The differentiated Peiffer identity2835makes \(\ker\partial_*\) central in \(\mathfrak h\), and2836equivariance makes it stable under the \(G\)-action. On that2837subspace \(D_A^2u=(\partial_*B)\triangleright u=[B,u]=0\).2838\end{proof}2839 2840Even under fake flatness \(D_A\)-closedness is a statement about2841the flat local system on the kernel, rather than ordinary2842closedness of each coefficient in an arbitrary frame. In a2843parallel local frame for that system it becomes ordinary2844closedness. The distinction is meaningful when the2845\(G/\partial(H)\)-action on \(\ker\partial_*\) is nontrivial.2846 2847\section{Two elementary changes of potentials}2848 2849The ordinary frame change by \(g:U\to G\) acts by2850\[2851 A^g=\operatorname{Ad}(g^{-1})A+g^{-1}dg,2852 \qquad B^g=g^{-1}\triangleright B.