Equivalence of geometric and combinatorial Dehn functions. (Q1866411)

Let \(X\) be a simply connected 2-complex, and let \(w\) be an edge circuit in \(X^{(1)}\). If \(D\) is a van Kampen diagram for \(w\), then the area of \(D\) is defined as the number of 2-cells on \(D\), and the area of \(w\), denoted \(a(w)\), is defined as the minimum of the areas of all van Kampan diagrams for \(w\). The Dehn function of \(X\) is then defined to be \(\delta_X(n)=\max a(w)\), where the maximum is taken over all loops \(w\) of length \(l(w)\leq n\). Given two functions \(f\) and \(g\) from \(\mathbb{R}^+\) to \(\mathbb{R}^+\), we say that \(f\prec g\) if there exist positive constants \(A,B,C,D,E\) so that \(f(n)\leq Ag(Bn+C)+Dn+E\). Two such functions are called equivalent if \(f\prec g\) and \(g\prec f\). Let \(G\) be a finitely presented group, and \(\mathcal P\) be a finite presentation for \(G\). Let \(K=K(\mathcal P)\) be the 2-complex associated to \(\mathcal P\), i.e., the 2-complex with a single vertex, an oriented edge for every generator of \(\mathcal P\), and a 2-cell for every relator, attached to the edges according to the spelling of the relator. Then the Dehn function of \(\mathcal P\) is, by definition, the Dehn function \(\delta_{\widetilde K}\) of the universal covering of \(K\). Two finite presentations \(\mathcal P\) and \(\mathcal Q\) for the same group \(G\) yield 2-complexes \(\widetilde K(\mathcal P)\) and \(\widetilde K(\mathcal Q)\) with quasi-isometric 1-skeleta, and hence equivalent Dehn functions. Thus the Dehn function of the group \(G\) is defined to be the equivalence class of the Dehn functions of any of its presentations. Given a Lipschitz loop \(\gamma\) in a simply connected Riemannian manifold \(M\), we define the area of \(\gamma\) to be the infimum of the areas of all Lipschitz discs bounded by \(\gamma\). We then define the geometric Dehn function of \(M\) by \[ \delta_M(x)=\max_{l(\gamma)\leq x}\text{area}(\gamma) \] where \(l(\gamma)\) represents the length of \(\gamma\). The main purpose of the paper under review is to consider the natural question of whether the Dehn function of a simply connected Riemannian manifold \(M\) and of a finitely presented group \(G\) acting properly discontinuously and cocompactly by isometries on \(M\) agree. The fact that they effectively agree has been implicitly assumed in the literature, though no proof has been given. A closely related statement is given in Theorem 10.3.3 of [\textit{D. B. A. Epstein} et. al., Word processing in groups. Boston, MA etc.: Jones and Bartlett Publishers (1992; Zbl 0764.20017)], applying the Deformation Theorem of Geometric Measure Theory (see 4.2.9 of [\textit{H. Federer}, Geometric measure theory. Classics in Mathematics. Berlin: Springer-Verlag (1996; Zbl 0874.49001) and \textit{L. Simon}, Lectures on geometric measure theory. Proceedings of the Centre for Mathematical Analysis, Australian National University, Vol. 3. Canberra: Centre for Mathematical Analysis, Australian National University (1983; Zbl 0546.49019)]) to this setting, and which provides the basis of the Pushing Lemma (Lemma 2.1 of the paper under review). This paper is devoted to providing a complete and detailed proof that the combinatorial and geometric Dehn functions are equivalent. The authors state that it is known to them that M. Bridson has lectured on an alternate, unpublished proof of the same result. The authors, to obtain the greatest generality, state their result not only for Riemannian manifolds, but for simplicial complexes with certain metric properties.