Sensitivity of dynamical systems to parameters in a convex subset of a topological vector space (Q2470935)

The authors study the Cauchy problem \[ \dot x(t)=f(t,x(t),u)\;(t\geq t_0),\quad x(t_0) =x_0 \] in a complex Banach space. By several interesting examples they illustrate that a parameter \(u\) may belong to a convex subset of a topological vector space. For the solution \(x\) to the above problem (uniquely existing under usual assumptions on \(f)\) they define the sensitivity of the solution \(x\) with respect to \(u\) as the directional derivative \[ y(t)=\delta x(t,u;\nu-u)=\lim_{\varepsilon \to 0^+}\frac {x(t,u+\varepsilon(\nu-u))-x(t,u)}{\varepsilon} \] and obtain the sensitivity equation \[ \dot y (t)=f_x(t,x(t,u),u)y(t)+\delta f(t,x(t,u),u;\nu-u)\quad\text{ for } t\geq t_0,\;y(t_0)=0. \] (Here \(f_x\) denotes the Fréchet derivative and \(\delta f(t,x(t,u),u;\nu-u)\) is defined as above). This result is applied to equations of the form \(\dot x(t)=F(t,x(t),P)\), where \(P\) is a probability distribution (or measures). Interpreting \(P\) as a parameter in an appropriately chosen space the authors investigate the scalar linear delay problem \[ \dot x(t)=f(t)+\int^0_{-\tau}x(t+\tau)dP(\tau)\;(t\geq 0),\quad x(\theta)=\Phi(\theta)\;(\theta\in[-\tau,0)),x(0)=x_0. \] Numerical simulations are given. This is an interesting and deep paper, indeed.