Extremal problems for Fredholm eigenvalues (Q731369)

Denote by \(\Sigma\) the class of univalent holomorphic functions \(f(z)=z+b_0+b_{-1}z^{-1}+\dots\), \(z\in\Delta^*=\{z:|z|>1\}\), \(f(z)\neq0\). The Grunsky coefficients \(\alpha_{mn}\) of \(f\in\Sigma\) are defined by \[ \log\frac{f(z)-f(\zeta)}{z-\zeta}=-\sum_{m,n=1}^{\infty}\alpha_{mn}z^{-m}\zeta^{-n}, \] and the quantity \[ \varkappa(f):=\sup\left\{\left|\sum_{m,n=1}^{\infty}\sqrt{mn}\alpha_{mn}x_mx_n \right|:x=(x_n)\in S(l^2)\right\}, \] where \(S(l^2)\) is the unit sphere of the Hilbert space \(l^2\) with \(\|x\|^2=\sum_1^{\infty}|x_n|^2\), is called the Grunsky norm of \(f\). Let \(\Sigma_k\) be the subclass of all \(f\in\Sigma\) having \(k\)-quasiconformal extensions to \(\mathbb C\cup\{\infty\}\). Consider the subclass \(\Sigma\langle k\rangle\) which consists of all \(f\in\Sigma_k\) with \(\varkappa(f)\leq k\). \(\Sigma^0:=\bigcup_k\Sigma_k=\bigcup_k\Sigma\langle k\rangle\). Denote by \(f^{\mu}(z)=z+b_0+b_{-1}z^{-1}+\dots\) any solution of the Beltrami equation \(\partial_{\overline z}f=\mu\partial_zf\), \(\mu\in L_{\infty}(\mathbb C)\), \(\mu|\Delta^*=0\), \(\|\mu\|<1\). The minimal dilatation is \(k(f):=\inf\{k(w^{\mu})=\|\mu\|_{\infty}: w^{\mu}|\partial\Delta^*=f\}\). Let \(F(f)\) be a holomorphic functional on \(\Sigma\). Assume that the derivative \(\varphi_0(z)=F'_I(g(I,z))\) at the identity map \(I\) is a meromorphic function on \(\mathbb C\), which is holomorphic and integrable on the unit disk \(\Delta\). Here \(g(w,\zeta)\) is the kernel of variation for \(f\in\Sigma^0\) with the normalization conditions which distinguish a unique quasiconformal extension of \(f\in\Sigma^0\) with a fixed \(\mu\). The main result is given in Theorem 3.1: If the range domain of the functional \(F\) on \(\Sigma^0\) has more than two boundary points and its derivative \(F'_I(g(I,z))=\varphi_0(z)\) has only zeros of even order in \(\Delta\), then there exists a number \(k_0(F)>0\) such that, for all \(k\leq k_0( F)\), the extremal functions for \(F\) on the class \(\Sigma\langle k\rangle\) are the same as in the smaller class \(\Sigma_k\); in other words, \[ \max_{\varkappa(f)\leq k}|F(f)-F(I)|=\max_{k(f)\leq k}|F(f)-F(I)|=\max_{|t|=k}|F(f^{t|\varphi_0|/\varphi_0}-F(I)|. \] Theorem 3.1 is derived from Theorem 3.2: If a holomorphic functional \(F\) on \(\Sigma^0\) with \(F(I)=0\) satisfies \(\sup_{\|\mu\|_{\infty}<1}|F(f^{\mu})|=1\), and its derivative \(F'_I(g(I,z))=\varphi_0(z)\) has only zeros of even order in \(\Delta\), then, for any \(k\in[0,1]\) and any \(t\) with \(|t|=k\), \(\max_{\varkappa(f)\leq k}|F(f)|=|F(f^{t|\varphi_0|/\varphi_0})|=k\). Theorems 3.1 and 3.2 are illustrated by several examples.