The asymptotic behavior of functions extremal for Baernstein's cos \(\beta\lambda\) theorem (Q791710)

Suppose that f is an entire function of order \(\lambda\). The cos \(\pi\lambda\) theorem of Wiman and Valiron (1914) asserts that if \(\lambda<1\), then \(\overline{\lim}_{r\to \infty}\frac{\log m^*(r)}{\log M(r)}\geq \cos \pi \lambda\) where M and \(m^*\) denotes respectively the maximum and minimum moduli of f. This theorem was generalized by the reviewer (1974) as follows: If \(0\leq \lambda<\infty\) and the parameter \(\beta\) satisfies \(\beta \lambda<\pi,\quad 0<\beta \leq \pi,\) then \(\overline{\lim}_{r\to \infty}\frac{\log m_{\beta}(r)}{\log M(r)}\geq \cos \beta \lambda,\) where \[ m_{\beta}(r)=\sup_{\phi}\inf_{| \theta | \leq \beta}\log | f(re^{i(\theta +\phi)})|. \] Drasin and Shea (1976) studied functions extremal for the cos \(\pi\lambda\) theorem in the sense that \(\log m^*(r)\leq((\cos \pi \lambda)+o(1))\log M(r).\) They proved that outside a small exceptional r-set log M(r) must grow regularly, in a certain sense, and f(z) behaves locally like a Lindelöf function. In the paper under review the authors study functions which are similarly extremal for the cos \(\beta\lambda\) theorem. They prove that again, outside an exceptional r-set, log M(r) grows regularly and f(z) is locally Lindelöfian, but only in sectors of width 2\(\beta\). Outside these sectors many different behaviours are possible. The log M(r) result follows easily by combining the Drasin-Shea techniques with the reviewer's ''*-function'' techniques, but the f(z) results require an intricate and ingenious analysis. The cos \(\beta\lambda\) theorem is sharp if \(\beta \lambda \leq frac{1}{2}\pi\) or \(\lambda \leq 1.\) The authors point out that if \(\lambda>1\) and \(frac{1}{2}\pi<\beta \lambda<\pi\) extremal function seem to be known only in certain isolated cases. This leads to a probably difficult open question: Is the cos \(\beta\lambda\) theorem really sharp for every \(\beta\) and \(\lambda\) with \(\beta \lambda<\pi,\quad \lambda>1?\)