How geodesics approach the boundary in a simply connected domain (Q1343316)

Let \(\Phi\) map the unit disk conformally onto the Jordan domain \(\Omega\). The author studies the problem whether the geodesic \(\Phi(te^{i\vartheta})\), \(0\leq t\leq 1\), approaches its endpoint more or less monotonically. Let \[ b(t, \vartheta)= \text{dist}\bigl[\Phi(te^{i\vartheta}),\partial\Omega\bigr],\quad e(t, \vartheta)= \bigl|\Phi (te^{i\vartheta})- \Phi(e^{i\vartheta})\bigr|, \] furthermore \(B(t,\vartheta)= \sup\{b(s, \vartheta): t\leq s\leq 1\}\) and similarly \(E(t, \vartheta)\). The author gives marvellously precise answers to this question. It is proved that, for almost all \(\vartheta\), \[ \lim_{r\to 1} {B(t, \vartheta)\over e(t, \vartheta) \varphi(e(t, \vartheta))}= \lim_{r\to 1} {E(t, \vartheta)\over e(t, \vartheta) \varphi(e(t, \vartheta))}= 0\tag{*} \] for any mildly regular decreasing function \(\varphi\) such that \[ \int^ 1_ 0 \varphi(t)^{- 9/2} t^{- 1} dt\tag{**} \] is finite. If this integral is however infinite then there exists \(\Phi\) such that (*) with \(\limsup\) is infinite for almost all \(\vartheta\) so that the approach is very far from being monotone. A similar behaviour holds when \(e(t, \vartheta)\) in (*) is replaced by \(b(t, \vartheta)\); then the exponent \(-9/2\) in (**) is replaced by \(-2\). Finally, \[ \limsup_{r\to 1} {E(t, \vartheta)\over B(t, \vartheta)\log\log [1/B(t, \vartheta)]} \] is bounded for almost all \(\vartheta\) but there exists \(\Phi\) such that this \(\limsup\) is positive for almost all \(\vartheta\). The author raises the interesting question whether \[ \liminf_{r\to 1} e(t, \vartheta)/b(t, \vartheta)\leq 1\quad\text{for almost all }\vartheta; \] it is easy to see that this holds with 1 replaced by 2.