On the structure of the solution set for first order differential equations (Q1078389)

Consider the periodic boundary value problem: (*) \(u'+g(u)=e(t)\), \(u(0)=u(2\pi),\) where \(g\in C({\mathbb{R}},{\mathbb{R}})\) is a monotone function and \(e\in L^ 2(0,2\pi)\). The problem (*) has a solution if and only if \((1/2\pi)\int^{2\pi}_{0}e(t)dt=\omega \in Range g.\) Moreover if \(\omega \in Int(Range g)\) then the solution set is nonempty, compact, connected and acyclic, on the other hand if \(\omega\in \partial (Range g)\), then the solution set is either empty or connected and unbounded. The author presents a lot of similar results and shows that the solution set may be ''chaotic'' if any of the assumptions is slightly modified.