Using dynamical systems methods to solve minimization problems (Q1902083): Difference between revisions

Many interesting problems in physics, chemistry, biology, economics and engineering are modelled by a parameter-dependent mathematical object, e.g. a parameter-dependent dynamical system or a parameter-dependent algebraic equation. Very often it is a natural task to determine parameter values such that a certain objective function is minimized. One possibility to compute a local minimum of a real-valued function \(f\) of \(N\) unknowns is to solve the gradient differential equation \(\dot x= - \nabla f(x)\). In the present paper a convergence result for minimization problems by discretizing this equation via fixed time-stepping one-step methods is derived. The asymptotic features of the numerical and the exact solutions are compared. Furthermore, it is shown that for a certain class of one-step methods the totality of the discrete and the continuous \(\omega\)-limit sets coincide if the stepsize is sufficiently small and if all equilibria of the gradient differential equation are regular. The parameters determination of the Michaelis-Menten kinetics model, describing how the degradation of a substrate is governed by an enzyme, is presented as numerical application.