Continuous solutions to two iterative functional equations (Q2055259): Difference between revisions

Let \((\Omega,\mathcal{A},P)\) be a probability space and \((X,\rho)\) a separable metric space with the \(\sigma\)-algebra \(\mathcal{B}\) of all its Borel subsets. Let \(f:X\times \Omega \to X\) be a \(\mathcal{B}\otimes \mathcal{A}\) measurable function. The author's aim is to look for continuous solutions \(\varphi: X \to \mathbb{R}\) of the equations \begin{align*} \varphi(x)&=F(x)-\int_{\Omega} \varphi(f(x,\omega))P(d\omega), \tag{1} \\ \varphi(x)&=F(x)+\int_{\Omega} \varphi(f(x,\omega))P(d\omega). \tag{2} \end{align*} Define \[ f^0(x,\omega_1,\omega_2,\dots)=x, \qquad f^n(x,\omega_1,\omega_2,\dots)=f(f^{n-1}(x,\omega_1,\omega_2,\dots),\omega_n), \] and \[ \pi_n^f(x,B)=P^{\infty}(f^n(x,\cdot)\in B), \quad n\in \mathbb{N}\cup \{0\}, \ B\in \mathcal{B}. \] Under the following conditions \[ \int_{\Omega} \rho(f(x,\omega),f(z,\omega))P(d\omega)\le \lambda \rho(x,z), \quad x,z \in X, \ \lambda \in (0,1), \tag{3} \] and \[ \int_{\Omega} \rho(f(x,\omega),x)P(d\omega)< \infty, \] there exists a probability Borel measure \(\pi^f\) on \(X\) such that for every \(x\in X\) the sequence \((\pi_n^f(x,\cdot))\) converges weakly to \(\pi^f\). Assuming these conditions with a fixed \(\lambda \in (0,1)\), let \(\mathcal{F}(X)\) be defined as the set of all continuous functions \(F:X \to \mathbb{R}\) such that there are a sequence \((F_n)\) of real functions on \(X\) and constants \(\theta \in (0,1)\), \(L\in (0,1/\lambda)\) and \(\alpha, \beta \in (0, \infty)\) such that \[ |F(x)-F_n(x)|\le \alpha \theta^n, \quad x\in X,\ n\in \mathbb{N}, \] and \[ |F_n(x)-F_n(z)|\le \beta L^n\rho(x,z), \quad x,z \in X,\ n\in \mathbb{N}. \] The first result can now be stated. Theorem. Assume the previous conditions. If \(F \in \mathcal{F}(X)\) then \[ \varphi(x)=F(x)-\frac{1}{2}\int_X F(z)\pi^f(dz)+\sum_{n=1}^{\infty} (-1)^n\Big(\int_X F(z)\pi_n^f(x,dz)-\int_X F(z)\pi^f(dz)\Big), \quad x\in X, \] defines a continuous solution of (1). If additionally the condition \(\int_X F(x)\pi^f(dx)=0\) holds true, then the formula \[ \varphi_0(x)=F(x)+\sum_{n=1}^{\infty} \int_X F(z)\pi_n^f(x,dz), \quad x\in X, \] defines a continuous solution \(\varphi_0:X \to \mathbb{R}\) of (2). Concerning the problem of uniqueness of solution the following is proved. Theorem. Assume the previous conditions. Let \(F \in \mathcal{F}(X)\). \begin{itemize} \item[(i)] If \(\varphi_1, \varphi_2\in \mathcal{F}(X)\) are solutions of (1), then \(\varphi_1=\varphi_2\). \item[(ii)] If \(\varphi_1, \varphi_2\in \mathcal{F}(X)\) are solutions of (2), then \(\varphi_1-\varphi_2\) is a constant function. \end{itemize} The last problem investigated in the paper is about the number of functions \(F\) for which (1) and (2) have continuous solutions. Assume that \((X,\rho)\) is a compact metric space and that condition (3) holds true. Define \begin{align*} \mathcal{F}_1&=\{F\in C(X): \text{Eq. (1) has a continuous solution} \}, \\ \mathcal{F}_2&=\{F\in C_f: \text{Eq. (2) has a continuous solution} \}, \end{align*} where \[ C_f=\{F\in C(X): \int_X F(x)\pi^f(dx)=0 \}. \] Theorem. Under the previous assumptions, the following holds: \begin{itemize} \item[(i)] \(\mathcal{F}_1\) is a Borel and dense subset of \(C(X)\), and if \(\mathcal{F}_1\neq C(X)\), then \(\mathcal{F}_1\) is of first category in \(C(X)\) and a Haar zero subset of \(C(X)\). \item[(ii)] \(\mathcal{F}_2\) is a Borel and dense subset of \(C_f\), and if \(\mathcal{F}_2\neq C_f\), then \(\mathcal{F}_2\) is of first category in \(C_f\) and a Haar zero subset of \(C_f\). \end{itemize}