Estimates for the Bergman kernel and metric of convex domains in Cⁿ (Q4484549)

The geometry of a bounded convex domain in \(\mathbb{C}^n\) is well understood from the point of view of complex analysis: at each point \(z\) of the domain there is a special polydisc (not uniquely determined) whose shape controls the function theory. Namely, take an affine complex disc centered at \(z\) that extends out to the closest boundary point of the domain; slice the domain by the \((n-1)\)-dimensional affine complex subspace through \(z\) orthogonal to the disc; repeat the construction in the slice; iterate; and obtain the canonical polydisc as the Cartesian product of the resulting \(n\) mutually orthogonal affine complex discs centered at \(z\). This construction apparently was introduced in the unpublished dissertation of \textit{Jui-Hsin Chen} directed by David Catlin [Estimates of the invariant metrics on convex domains in \(\mathbb{C}^n\), Purdue University (1989)], where it was used to estimate the Bergman kernel function on the diagonal and the Bergman metric near convex, finite-type boundary points of pseudoconvex domains. Subsequently \textit{Jeffery D. McNeal} went much further to obtain precise estimates for the Bergman kernel function and its derivatives both on and off the diagonal near such boundary points [Adv. Math. 109, 108-139 (1994; Zbl 0816.32018)]. In this article, the authors give short geometrical proofs of estimates for the Bergman kernel function on the diagonal and the Bergman metric for any convex domain in \(\mathbb{C}^n\), whether finite type or not. The only hypothesis on the convex domain \(D\) is that \(D\) can be mapped biholomorphically to a bounded domain (equivalently, \(D\) contains no complex line). First the authors give a precise estimate for how the Bergman kernel function \(K(z)\) on the diagonal compares to the reciprocal of the volume of the canonical polydisc centered at \(z\). If \(V(z)\) denotes that volume, then \(1/4^n \leq K(z) V(z) \leq (2n)!/2^n\) for all points \(z\) in \(D\), and the constants are the best possible. Then the authors estimate the Bergman metric \(B(z;X)\) in terms of the ``largest disc'' contained in \(D\) in the direction \(X\). Namely, let \(d(z,X)\) denote the supremum of the set of positive numbers \(\lambda\) for which \(z+\alpha X\) lies in \(D\) whenever \(|\alpha|<\lambda\). Then \[ 1/2 \leq B(z;X) d(z,X) \leq n \sqrt{2^{n-1}(2(n+1))!/3}. \] It follows that if \(F(z;X)\) denotes either the Carathéodory metric or the Kobayashi metric (by Lempert's theorem these two metrics coincide on convex domains), then \[ 1 \leq B(z;X)/F(z;X) \leq 2n \sqrt{2^{n-1}(2(n+1))!/3}. \]