Absence of robust rigidity for circle maps with breaks (Q2450580)

A classical theorem of \textit{A. Denjoy} [J. Math. Pures Appl. (9) 11, 333--375 (1932; JFM 58.1124.04)] says that a \(C^{1+\alpha}\), \(\alpha>0\), diffeomorphism of the circle with irrational rotation number \(\rho\) is conjugate to a rotation by \(\rho\). A natural question to ask is under what conditions the conjugacy is smooth. \textit{V. I. Arnol'd} proved in 1961 [Transl., Ser. 2, Am. Math. Soc. 46, 213--284 (1965; Zbl 0152.41905); translation from Izv. Akad. Nauk SSSR, Ser. Mat. 25, 21--86 (1961)] that an analytic orientation preserving diffeomorphism of the circle whose rotation number satisfies a certain Diophantine condition is analytically conjugated to a rotation. This was later extended by \textit{M. R. Herman} [Publ. Math., Inst. Hautes Étud. Sci. 49, 5--233 (1979; Zbl 0448.58019)] and blossomed into \textit{rigidity theory}. Typically, rigidity results depend strongly on the Diophantine properties of the rotation numbers. The authors study the smoothness of the conjugacy of orientation-preserving diffeomorphisms of the circle with the same rotation number \(\rho\). A \textit{circle map with a break singularity} (with \textit{break} for short) is an orientation-preserving homeomorphism \(T\) of the circle \(\mathbb{T}^1=\mathbb{R}/\mathbb{Z}\) which is \(C^r\)-smooth for some \(r\in [1,\infty]\cup\{\omega\}\) everywhere except a single point where the derivative has a jump discontinuity. Broadly speaking maps with breaks are situated in between circle diffeomorphisms and critical circle maps. More precisely, let \(\tau:\mathbb{R}\to\mathbb{R}\) be a lift of \(T\). We say that \(T\) has a \textit{break singularity} if there exists a point \(x_{\text{br}}\in[0,1)\) such that: {\parindent=0.7cm\begin{itemize}\item[--] \(\tau \in C^r\) on \([x_{\text{br}},x_{\text{br}}+1]\) for some \(r\in [1,\infty]\cup\{\omega\}\), \item there exists \(C >0\) such that \(\tau'(x)>C>0\) for all \(x\in[x_{\text{br}},x_{\text{br}}+1]\), \item[--] the one-sided derivatives at \(x_{\text{br}}\), denoted \(\tau'_-(x_{\text{br}})\) and \(\tau'_+(x_{\text{br}})\) satisfy \[ \sqrt{\frac{\tau'_-(x_{\text{br} })}{\tau'_+(x_{\text{br}})}}=c \] for some \(c\in\mathbb{R}^+\setminus\{1\}\) (\(c\) is called the \textit{size} of the break). \end{itemize}} Here \(C^\omega\) denotes the family of functions \(\tau\) such that the restriction of \(\tau\) to \([x_{\text{br}},x_{\text{br}}+1]\) has analytic extension to some disk in the complex plane containing \([x_{\text{br}},x_{\text{br}}+1]\). In a previous paper, the first author and \textit{A. Teplinsky} [Commun. Math. Phys. 320, No. 2, 347--377 (2013; Zbl 1280.37037)] (announced by the authors in [Invent. Math. 169, No. 1, 193--218 (2007; Zbl 1151.37035)]) proved the rigidity theorem stating that for diffeomorphisms with breaks and a certain class of strongly non-Diophantine (Liouville) rotation numbers, the conjugacy is still at least \(C^1\)-smooth, provided that the sizes of the breaks are the same. In the paper under review, the authors construct examples showing that maps with breaks are ``as bad as they can be'' in light of the positive results mentioned above: {\parindent=0.7cm\begin{itemize}\item[--] There are analytic circle maps with singularities of break type with the same rotation number and the same size of the break for which no conjugacy is Lipschitz continuous. \item[--] There are maps with a rotation numbers in the class considered in the rigidity theorem of Khanin and Teplinsky for which a \(C^1\)-smooth conjugacy is not \(C^{1+\alpha}\) for any \(\alpha>0\). \end{itemize}}