The Riemannian quantitative isoperimetric inequality (Q6172666)

In this interesting paper, the authors prove that a direct analogue of the Euclidean quantitative isoperimetric inequality is, in general, false on a closed Riemannian manifold. More precisely, the first main result of the paper proves that there exists a closed manifold \(M^n\) with a real analytic Riemannian metric \(g\), a uniquely isoperimetric region \(\Omega\subset M\) and sets \(E_k\) with smooth boundary, such that \(|\Omega\Delta E_k|_g\rightarrow 0\), but \[ \frac{(|\Omega\Delta E_k|_g)^2}{\mathcal{P}^g(E_k)-\mathcal{P}^g(\Omega)}\rightarrow\infty. \] Here, \(\mathcal{P}^g\) denotes the distributional perimeter and \(\Omega\Delta E_k\) denotes the symmetric difference. Moreover, the authors show that there is a smooth, but not real analytic Riemannian metric \(g\) on \(M^n\), such that one cannot bound \(\mathcal{P}^g(E_k)-\mathcal{P}^g(E)\) by any power of \(|\Omega\Delta E_k|_g\). The second main result of the paper shows that the Riemannian quantitative isoperimetric inequality is true generically. A key ingredient of this proof is a bumpiness result with a volume constraint inspired by \textit{B. White} [Indiana Univ. Math. J. 40, No. 1, 161--200 (1991; Zbl 0742.58009); Indiana Univ. Math. J. 36, No. 3, 567--602 (1987; Zbl 0770.58005)]. The authors are also inspired by \textit{M. Cicalese} and \textit{G. P. Leonardi} [Arch. Ration. Mech. Anal. 206, No. 2, 617--643 (2012; Zbl 1257.49045)], using the selection principle approach. This allows one to reduce to the case where \(\partial \Omega\) is a small \(C^{1,\alpha}\) graph over \(\partial B\), and thus use [\textit{M. Colombo} et al., J. Reine Angew. Math. 768, 149--182 (2020; Zbl 1452.35033)]. Finally, the authors prove that a sharp, modified version of the quantitative isoperimetric inequality holds for a real analytic metric. That is done through the Lojasiewicz-Simon inequality.