Unwinding spirals I (Q2010769)

The authors study the possibility to ``unwind'' a spiral by a bi-Lipschitz homeomorphism \(h\) of \(\mathbb{R}^2\). Here, a spiral is defined as \(C_\phi(t) = \phi(t)\mathrm{e}^{it}\), \(t \in [0,\infty)\), where \(\phi\colon [0,\infty) \to (0,1]\) is a continuous, monotonically decreasing function with \(\lim_{t\to\infty} = 0\). A point \(u \in S^1\) is called \textit{asymptotic direction\/} of a curve at \(0\), if there exists a sequence \(a_n\) of curve points with \(a_n \to 0\) and \(a_n/\Vert a_n \Vert \to u\). The \(h\)-image of \(C_\phi\) is called \textit{unwound} if the set of its asymptotic directions at \(0\) is a proper subset of~\(S^1\). For a logarithmic spiral, given by \(\phi(t) = \mathrm{e}^{-at}\), unwinding turns out to be possible. In fact, it is even possible to map a logarithmic spiral to a line segment by a bi-Lipschitz homeomorphism \(h\). For spirals defined by a function \(\phi\) with sub-exponential decay (meaning \(\log(\phi(n))/n \to 0\) as \(n \to \infty\)), this is easily seen to be impossible. Substantially harder to prove is the paper's main result: Spirals with sub-exponentially decaying function \(\phi\) cannot be unwound. For a consecutive paper the authors announce further investigations on unwinding spirals and in particular an example of a spiral that can be unwound but not to a line segment.