Density of hyperbolicity for classes of real transcendental entire functions and circle maps (Q2347583)

In one-dimensional dynamics, the density of hyperbolicity -- the question of where hyperbolic maps are dense in a given parameter space -- is one of the most central questions in the field. The conjecture is still open in any complex parameter space, however, a lot of progresses was made in term of real parameter space. In particular, \textit{M. Lyubich} [Acta Math. 178, No. 2, 185--297 (1997; Zbl 0908.58053)], and, independently, \textit{J. Graczyk} and \textit{G. Światek} [Ann. Math. (2) 146, No. 1, 1--52 (1997; Zbl 0936.37015)] confirmed the conjecture for the real quadratic family \(x\to x^2+c\). Moreover, \textit{O. Kozlovski} et al. [Ann. Math. (2) 166, No. 1, 145--182 (2007; Zbl 1138.37013)] solved the question for real polynomials with real critical points and for general interval and circle maps. In this paper, the authors first successfully prove density of hyperbolicity for the following families of transcendental functions. The first result is about the density of hyperbolicity in families of bounded functions. Theorem 1. Let \(n\geq 1\), and let \(N\) be an \(n\)-dimensional manifold. Suppose that \((f_\lambda)_{\lambda\in N}\) is a continuous family of functions \[ f_\lambda\in S_{\mathbb{R}}=\{f:\mathbb{C}\to \mathbb{C}, f(\mathbb{R})\subset \mathbb{R} \text{ and the set of singular values is finite and contained in } \mathbb{R}\} \] such that 1. \(f_\lambda|_\mathbb{R}\) is bounded for all \(\lambda\in N\), 2. the number of singular values is less than \(n\) for all \(\lambda \in N\), 3. no two maps \(f_{\lambda_1}\) and \(f_{\lambda_2}\) are conjugate by a map \(M(z)=az+b\). Then the set \(\{\lambda\in N: f_\lambda \text{ is hyperbolic}\}\) is dense and open in \(N\). The result implies the density of hyperbolicity for trigonometric polynomials. The set of parameters \((a, b)\) for which the map \(C(a,b)(x)=a\sin x+b\cos x\) is hyperbolic forms an open and dense subset in \(\mathbb{R}^2\). More generally, let \(n\geq 1\), then hyperbolicity is dense in the space of the real trigonometric polynomials \[ f(x)=a_0+\sum_{j=1}^n (a_j\cos(jx)+b_j\sin(jx)) \] for which all critical values are real. The second result is about functions with unbounded singular values. Theorem 2. Let \(n\geq 1\), and let \(N\) be a \(n\)-dimensional manifold. Suppose that \((f_\lambda)_{\lambda\in N}\) is a continuous family of functions \[ f_\lambda\in S_{\mathbb{R}}=\{f:\mathbb{C}\to \mathbb{C}, f(\mathbb{R})\subset \mathbb{R} \text{ and the set of singular values is finite and contained in } \mathbb{R}\} \] such that 1. the number of singular values is less than \(n\) for all \(\lambda \in N\), 2. no two maps \(f_{\lambda_1}\) and \(f_{\lambda_2}\) are conjugate by a map \(M(z)=az+b\). 3. \(f_\lambda\) has no asymptotic values and the degree of critical points of \(f\) is uniformly bounded for every \(\lambda \in N\) or satisfies a certain geometric condition Then the set of \(\lambda\in N\) such that every singular value of \(f_\lambda\) either belongs to a basin of attraction or tends to infinity under iteration, is dense and open in \(N\). In particular, real-hyperbolicity is dense in the space of functions \(f\in S_\mathbb{R}\) that have \(k\) singularities, and real-hyperbolicity is dense in the family \[ S_{a,b}: x\to axe^x+b, a\in \mathbb{R}\setminus \{0\}, b\in \mathbb{R}. \] Another result is about the density of hyperbolicity for circle maps. Theorem 3. Let \(n\geq 1\), and let \(N\) be a \(n\)-dimensional manifold. Suppose that \((f_\lambda)_{\lambda\in N}\) is a continuous family of functions \[ f_\lambda\in S_{S^1}=\{f:\mathbb{C^*}\to \mathbb{C*}, f(S^1)\subset S^1 \text{ and the set of singular values is finite and contained in } S^1\} \] such that 1. the number of singular values is less than \(n\) for all \(\lambda \in N\), 2. no two maps \(f_{\lambda_1}\) and \(f_{\lambda_2}\) are conjugate by a rotation. Then the set \(\{\lambda\in N: f_\lambda \text{ is hyperbolic}\}\) is dense and open in \(N\). A natural family of circle maps of degree \(D\) can be described as follows. For \(\mu\in \mathbb{R}^{2m}\) and \[ F_{\mu}(t)=Dt+\mu_1+\mu_{2m}\sin (2\pi mt)+\sum_{j=1}^{m-1}(\mu_{2j}\sin(2\pi jt) +\mu_{2j+1} \cos (2\pi jt)). \] Hence \(F_{\mu}(t)\) induces a circle map \(f_\mu:S^1\to S^1\) and the map is naturally parameterized by \[ \mu=\{\mu_1, \dots, \mu_{2m}\}\in \Delta, \] where \[ \Delta=\{\mu\in \mathbb{R}/\mathbb{Z}\times \mathbb{R}^{2m-1}, \mu_{2m}>0\}. \] The famous Arnold family \[ f_{\mu_1, \mu_2}(t)=t+\mu_1+\mu_2\sin (2\pi t), \mu_1\in \mathbb{R}, \mu_2>0 \] is a special case when \(m=1\). Following the theorem above, the authors prove the following theorem: Theorem 4. The set of parameters in \(\Delta\) for which \(f_\mu\) is hyperbolic is dense. Furthermore, let \(\mu_0\in \Delta\). The set of \(\mu\) for which \(f_\mu\) is topologically conjugate to \(f_{\mu_0}\) has at most \(m\) components. If \(f_{\mu_0}\) has no periodic attractors on the circle, then each component of topologically conjugacy of \(\mu_0\) is equal to a point.