A comparison of the Carathéodory and Filippov solution sets (Q1916800)

A function \(\varphi: [0, T]\to \mathbb{R}^n\) is called a Carathéodory solution of \((1)\) \(x'(t)= f(x(t))\), \(t\in [0, T]\) and \(x(0)= x_0\), if \(\varphi\) is absolutely continuous, \(\varphi'(t)= f(\varphi(t))\) a.e. on \([0, T]\) and \(\varphi(0)= x_0\), where \(T> 0\), \(f: \mathbb{R}^n\to \mathbb{R}^n\) and \(x_0\in \mathbb{R}^n\). A function \(\psi: [0, T]\mapsto \mathbb{R}^n\) is called a Filippov solution of \((1)\) if \(\psi\) is absolutely continuous, \(\psi'(t)\in F(\psi(t))\) a.e. on \([0, T]\) and \(\psi(0)= x_0\), with \(F: \mathbb{R}^n\mapsto {\mathcal P}(\mathbb{R}^n)\), \(F(y):= \bigcap_{\varepsilon> 0} \bigcap_{Z: m(Z)= 0} \overline{\text{conv}} f(B_\varepsilon(y)\backslash Z)\), where \(m\) represents the Lebesgue measure, \(\overline{\text{conv}}\) denotes closed convex hull and \(B_\varepsilon(y)\) is the open ball of radius \(\varepsilon> 0\) centered at \(y\). Two fundamental questions are addressed: I) under what conditions are the Carathéodory and Filippov solution sets of \(x'= f(x)\) the same? II) in the case in which the solution sets are not the same, what is the nature of a solution that is of one type but not the other ?