No label defined (Q4940741)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1406840
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1406840

      Statements

      0 references
      27 February 2000
      0 references
      Hausdorff dimension
      0 references
      homoclinic \(\Omega\)-explosion
      0 references
      Lagrange spectrum
      0 references
      regular Cantor set
      0 references
      Markov spectrum
      0 references
      The booklet under review surveys and outlines the main ideas behind the proofs of some recent results by the author (partly co-authored by J.-C. Yoccoz), some of them still unpublished, some others published in [Ann. Inst. Henri Poincaré Anal. Non Linéaire 13, 747--781 (1996; Zbl 0865.58035)]. While these results are concerned with two completely different mathematical areas (homoclinic bifurcations for diffeomorphisms on surfaces, on the one hand, and continuous fractions, on the other hand), they are linked by the notion of regular Cantor set. Let \(I_1,\ldots, I_k\) be pairwise disjoint compact subintervals of \(\mathbb{R}\) and let \(I\) be their convex hull. Assume that \(\psi:I_1\cup\cdots \cup I_k\rightarrow I\) is an expanding function of the class \(C^q\) for some real number \(q>1\), that is, a \(C^1\) map satisfying \(|\psi'(x)|>1\) for any \(x\) and for which there is a constant \(C>0\) with \(|\psi'(x)-\psi'(y)|\leq C|x-y|^{q-1}\) for any \(x,y\). Further, assume that, for any \(j\), \(\psi(I_j)\) is the convex hull of the union of some of the intervals \(I_i\), and that \(\psi^n(I_{j,n})= I_1\cup\cdots \cup I_k\) for any number \(n\) large enough, where \(I_{j,n}\subset I_j\) is the set on which \(\psi^n\) is well defined. Then we call the Cantor set \(K\) of all points from \(I\) on which \(\psi^n\) is well defined for any \(n\) a regular Cantor set of the class \(C^q\). Regarding regular Cantor sets, the paper includes, among others, the following results: Theorem 1. (jointly with J.-C. Yoccoz) Fix \(1<q\leq \infty\) and let \(\Omega\) be the set of pairs \((K_1,K_2)\) of regular Cantor sets of the class \(C^q\) such that the sum of their Hausdorff dimensions \(HD(K_1)+HD(K_2)\) is greater than \(1\). Then there is an open dense set \(\mathcal U\subset \Omega\) in the \(C^q\)-topology such that, if \((K_1,K_2)\in \mathcal U\) and \(S(K_1,K_2)\subset \mathbb{R}\) is the set of numbers \(t\) with the property that any pair \((\widetilde{K}_1,\widetilde{K}_2)\) close enough to \((K_1,K_2+t)\) satisfies \(\widetilde{K}_1\cap\widetilde{K}_2\neq \emptyset\), then \(S(K_1,K_2)\) is open and dense in \(K_1-K_2\). In the above theorem we mean \(K_2+t=\{y+t: y\in K_2\}\) and \(K_1-K_2=\{x-y:x\in K_1,y\in K_2\}\). In the \(C^q\)-topology, two Cantor sets \(K\) and \(\widetilde{K}\) are close when the corresponding families of intervals \(I_1,\ldots,I_k\) and \(\widetilde{I}_1,\ldots,\widetilde{I}_k\) in their definitions have the same cardinality, the endpoints of the intervals \(I_i\) are close to those of the intervals \(\widetilde{I}_i\), the maps \(\psi\) and \(\widetilde{\psi}\) are close in the \(C^1\)-topology, and the constants \(C\) and \(\widetilde{C}\) are close. Theorem 2. If \(K_1\) and \(K_2\) are Gauss regular Cantor sets then \(HD(K_1+K_2)=HD(K_1)+HD(K_2)\). Here, so-called Gauss regular Cantor sets are those obtained by restricting the Gauss map \(g:(0,1]\rightarrow [0,1)\) defined by \(g(x)=1/x-[1/x]\) to a finite numbers of intervals having as their endpoints irrationals of the type \(r+\sqrt{s}\), with \(r,s\in \mathbb{Q}\). The above theorems have a number of consequences. Theorem~1, which proves a conjecture by J. Palis from the eighties, implies the following remarkable result: Theorem 3. Let \(\{\varphi_\mu\}_{\mu\in \mathbb{R}}\) a family of \(C^2\)-diffeomorphisms on a closed surface \(M^2\) having a homoclinic \(\Omega\)-explosion in \(\mu=0\). Then, generically (in the \(C^2\)-topology for diffeomorphisms on \(M^2\)), the set of parameters \(\mu\) such that \(\varphi_\mu\) is either persistently hyperbolic or has stable homoclinic tangencies, has Lebesgue full density in \(\mu=0\). Here a diffeomorphism \(\varphi\) is hyperbolic if its nonwandering set \(\Omega(\varphi)\) is hyperbolic, and has a homoclinic tangency if \(\varphi\) has a hyperbolic saddle fixed point \(p\) whose stable and unstable manifolds \(W^s(p)\) and \(W^u(p)\) intersect tangentially at some point \(x\); ``persistence'' and ``stability'' means that the properties above persist under small perturbations of \(\varphi\). Finally, \(\{\varphi_\mu\}_{\mu\in \mathbb{R}}\) has a homoclinic \(\Omega\)-explosion in \(\mu=0\) if, roughly speaking, \(\varphi_u\) is persistently hyperbolic for any \(\mu<0\) and \(\varphi_0\) has a homoclinic tangency as before; moreover \(\Omega(\varphi_0)=\Lambda\cup \mathcal O\), where \(\Lambda\) is hyperbolic and \(\mathcal O=\{\varphi^n(x)\}_{n=-\infty}^\infty\) is the set of (quadratic) homoclinic tangencies between \(W^s(p)\) and \(W^u(p)\). Theorem~2 can be applied to number theory as follows. For any \(\alpha\in \mathbb{R}\), let \( k(\alpha)=\sup \{k>0: \text{}|\alpha-p/q|<1/(kq^2)\) has infinitely many rational solutions \(p/q\)
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references