The chain control set of discrete-time linear systems on the affine two-dimensional Lie group (Q6580106)

Let \(G=\mathrm{Aff}\left( E,\mathbb{R}\right) \) be the two-dimensional Lie group, i.e.\N\[\N\mathrm{Aff}\left( E,\mathbb{R}\right) =\left\{ \left[ \begin{array}{cc} a & b \\\N0 & A \end{array} \right] :a>0,b\in \mathbb{R}\right\}\N\]\Nso that \(\mathrm{Aff}\left( E,\mathbb{R}\right) \cong \mathbb{R}^{+\ast }\times \mathbb{R}\) endowed with the semidirect product \(\left( a,b\right) .\left( c,d\right) =\left( ac,d+bc\right) .\) The distance of two elements \( g,h\in G\) is denoted as \(\delta \left( g,h\right) .\) (The notation of the paper is modified in this review to avoid some confusions.) The discrete-time system \(\Sigma \) is given by\N\[\Nx_{k+1}=f_{u_{k}}\left( x_{k}\right) ,\quad u_{k}\in U,\quad x_{k}\in G\text{ for all integers }k\N\]\Nwhere \(U\) is a compact neighborhood of \(0\) and \(f_{\upsilon }\) is a diffeomorphism for all \(\upsilon \in U.\) By induction, the solution at time \( k\) can be written\N\[\Nx_{k}=\varphi \left( k,g,u\right) ,\quad x_{0}=g\N\]\Nwhere \(u=\left( u_{i}\right) _{i\in \mathbb{Z}}\in \mathcal{U}:=\prod_{i\in \mathbb{Z}}U.\) The system \(\Sigma \) is called linear if \(f_{\upsilon }\left( g\right) =f_{\upsilon }\left( 0\right) f_{0}\left( g\right) ,\) which implies \(\varphi \left( k,g,u\right) =\varphi \left( k,e,u\right) f_{0}^{k}\left( u\right) .\)\N\NReachability and controllability are defined as usual and the set of all points \(h\) reachable from \(g\) is denoted as \(\mathcal{R}\left( g\right) .\) Furthermore, a subset \(D\neq \varnothing \) of \(G\) is called a \textit{control set} if (i) for all \(g\in D,\) \(\exists u\in \mathcal{U}:\varphi \left( k,g,u\right) \in D\) for all \(k\geq 0;\) (ii) for all \(g,\) \(D\) is included in the closure of \(\mathcal{R}\left( g\right) ;\) \(D\) is maximal with these properties.\N\NWeakening this notion, given \(g,h\in G\) and a real number \(\varepsilon >0,\) an \(\left( \varepsilon ,k\right) \)-controllable chain between \(g\) and \(h\) is the set\N\[\N\mu _{\left( \varepsilon ,k\right) }=\left\{ \begin{array}{c} n\in \mathbb{N} \\\Nx_{0},x_{1},\dots,x_{n}\in G \\\Nu_{0},\dots,u_{n-1}\in \mathcal{U} \\\Nk_{0},\dots,k_{n-1}\geq k \end{array} \right\}\N\]\Nwhere \(x_{0}=g,\) \(x_{n}=h\) and \(\delta \left( \varphi \left( k_{i},x_{i},u_{i}\right) ,x_{i+1}\right) <\varepsilon \) \(\left( i=0,\dots,n-1\right) .\) The pair \(\left( g,h\right) \) is called \textit{chain-controllable} if for all \(\left( \varepsilon ,k\right) \in \mathbb{R} ^{+\ast }\times \mathbb{N}\), there is an \(\left( \varepsilon ,k\right) \)-controllable chain between \(g\) and \(h.\) A subset \(E\neq \varnothing \) of \(G\) is called a \textit{chain-control set} if (i) for all \(g\in E,\) \(\exists u\in \mathcal{U}:\varphi \left( k,g,u\right) \in E\) for all \(k\in \mathbb{Z};\) (ii) every pair \(\left( g,h\right) \in E\times E\) is chain-controllable\(;\) (iii) \(E\) is maximal with these properties.\N\NFrom a result obtained in [\textit{V. Ayala} and \textit{A. Da Silva}, J. Differ. Equations 268, No. 11, 6683--6701 (2020; Zbl 1440.93025)], it is shown that \(f_{u}\) is of the form\N\[\Nf_{u}\left( a,b\right) =\left( \mathfrak{h}\left( u\right) a,\alpha \left( a-1\right) +\gamma b+\mathfrak{g}\left( u\right) a\right) ,\quad \gamma \neq 0\N\]\Nwhere \(\mathfrak{h},\mathfrak{g}\) are smooth and \(\mathfrak{h}\left( 0\right) =1,\mathfrak{g}\left( 0\right) =0.\)\N\NAssuming that \(\mathfrak{h}\equiv 1,\) control sets \(D_{a}\) \(\left( a\in \mathbb{R}^{+\ast }\right) \)\ with empty interior are determined when \( \left\vert \gamma \right\vert <1.\) In that case, \(E\) is the union of the \( D_{a}\) \(\left( a>0\right) ,\) hence unique. Furthermore, if \(\gamma =\pm 1,\) then the sets \(D_{a}=\left\{ a\right\} \times \mathbb{R}\) \(\left( a>0\right) \) are control sets and, if \(\gamma =-1,\) \(G\) is chain-controllable.