Liapunov and Lagrange stability: inverse theorem for discontinuous systems (Q1265425): Difference between revisions

The authors consider general systems of the form \[ \dot x=F(t,x) \tag{1} \] where \(F\) is not continuous with respect to \(x\). The solution in this case may be defined in several senses among which the Filippov sense showed to be the most adequate. This allows to consider the system in the framework of differential inclusions of the form \[ \dot x\in F(t,x).\tag{2} \] The main purpose of the paper is to consider Lyapunov-type theorems (both direct and converse) to establish Lagrange and Lyapunov stability for differential inclusions. Especially, the converse theorems require the smoothness of equations' right-hand side and this is the reason why these topics are not tackled for differential inclusions. The main stability tool in this paper is given by the weak Lyapunov function \(V:\mathbb{R}_+\times Q\to\mathbb{R}\), where \(Q \subseteq \mathbb{R}^n\) is an open subset with the property that if \(t_1\leq t_2\) then \(V(t_1,x(t_1))\geq V(t_2,x(t_2))\) for each solution \(x:I\to \mathbb{R}^n\) to (2) and each pair \(t_1,t_2\in I\) such that \(x(t)\in Q\), \(t\in[t_1,t_2]\). The differential inclusion (2) is said to be robustly Lagrange stable if there exists a sequence \(\{G_i\}_{i\in \mathbb{N}}\) of open sets in \(\mathbb{R}_+\times \mathbb{R}^n\) and two sequences of real numbers \(\{a_i\}_{i\in\mathbb{N}}\), \(\{b_i\}_{i\in\mathbb{N}}\) such that \[ 0<a_i\leq b_i<a_{i+1},\;i=0,1,2, \dots;\;\lim_{i@>\text{II}>> \infty} a_i=+\infty, \tag{i} \] \[ \mathbb{R}_+\times \bigl\{x: | x|< a_i\bigr\} \subseteq G_i \subset\mathbb{R}_+ \times\bigl\{x: | x| <b_i\bigr\},\;i=0,1,2, \dots, \tag{ii} \] \[ \text{for each }i=0,1,2, \dots \text{ each initial pair } (t_0,x_0) \in G_i,\tag{iii} \] and for each solution \(x(\cdot; t_0,x_0)\) one has \((t,x(t; t_0,x_0))\in G_i\) for each \(t\geq t_0\). The paper contains for direct and converse theorems on Lagrange and Lyapunov stability. They may be summarized as follows: Assume there exists a weak Lyapunov function for (2) and two continuous maps \(a\): \(\mathbb{R}_+ \to\mathbb{R}_+\), \(b:\mathbb{R}_+ \to\mathbb{R}_+\) such that \[ V \text{ is defined and continuous on }\mathbb{R}_+\times \mathbb{R}^n\setminus \{0\},\tag{I} \] \[ a(0)=0,\;a(r)>0 \text{ for }r>0 \text{ and }\lim_{r\to\infty} a(r)=\infty, \tag{II} \] \[ a\bigl(| x| \bigr)\leq V(t,x)\leq b\bigl(| x|\bigr),\;(t,x)\in \mathbb{R}_+\times \mathbb{R}^n \setminus\{0\}.\tag{III} \] Then (2) is robustly Lagrange stable. Conversely, if (2) is robustly Lagrange stable, than there exist a weak Lyapunov function and two continuous maps such that (I)--(III) hold. Moreover \(V\) is locally Lipschitzian with respect to both arguments. It is obvious that Lagrange stability is a global result. Two other theorems contain local versions of two parts of the above theorem. The fifth theorem is unifying all these results. The proofs are inspired by ideas of \textit{J. Kurzweil} [Czechosl. Math. J. 5(80), 382-396 (1955; Zbl 0068.29901); Am. Math. Soc. Transl. II. Ser. 24, 19-77 (1963), translation from Czechosl. Math. J. 6(81), 217-259, 455-473 (1956; Zbl 0127.30703)] and of \textit{J. Kurzweil} and \textit{I. Vrkoć} [Transl., II. Ser., Am. Math. Soc. 29, 271-288; translation from Czechosl. Math. J. 7(82), 254-272 (1957; Zbl 0127.04302)] with additional technicalities required by the multivalued character of the RHS of (2).