Exponential and Gaussian concentration of 1-Lipschitz maps (Q1034703): Difference between revisions

A metric measure space \(X= (X,dx,\mu x)\) is a complete separable metric space \((X,dx)\) with a Borel probability measure \(\mu x\). Let \(N\) be an \(m\)-dimensional, complete and simply connected Riemannian manifold with non-positive sectional curvature, namely, a Hadamard manifold. Given a Borel measurable map \(f: X\to N\) such that the induced measure \(f_*(\mu x)\) of has finite moment of order 2, one defines its expectation value \(E(f)\) by the central mass of the measure \(f_*(\mu x)\). Considering a closed Riemannian manifold \(M\) in place of \(X\) with volume measure \(\mu_M\) normalized as \(\mu_M(M)= 1\), and denoting by \(\lambda_1(M)\) the first nonzero eigenvalue of the Laplacian on \(M\), \textit{M. Gromov} [Metric structures for Riemannian and non-Riemannian spaces. Progress in Mathematics (Boston, Mass.). 152. Boston, MA: Birkhäuser (1999; Zbl 0953.53002)] proved that \[ \mu_M(\{x\in M\mid d_N(f(x), E(f))\geq r\})\leq m/(\lambda_1(M) r^2) \] for any \(I\)-Lipschitz map \(f: M\to N\), and asked that if the right-hand side of the above inequality may be improved by the form \(C, e^{-C_2\sqrt{m/\lambda_1(M)r}}\) or the form \(C,e^{-C_2(m/\lambda_1(M))r^2}\). In this paper the answer to this problem is given affirmatively.