Inverse problems for variational inequalities

From MaRDI portal





[For the entire collection see Zbl 0752.00054.] This paper considers the problem of determining the motion \(x_ r(\cdot)\) of a system which takes the form of a parabolic variational inequality, posed on a Hilbert space, and depending on an unknown input variable \(u_ r(t)\). The time interval is partitioned, and at discrete times \(\tau_ i\) approximations \(\psi_{\varepsilon,i}\) to the \(x_ r(\tau_ i)\) are measured; the parameter \(\varepsilon\) measures the closeness of the approximations, that is, \(\varepsilon\) is the level of informational noise. The problem, then, is to calculate an approximate input \(u^*\); this is done by using an algorithm which is a dynamical modification of the discrepancy method. The major part of the paper is taken up with analysing the algorithm, and with showing that the family of approximations to \(u^*\), denoted by \(v^ \varepsilon\), is a good approximation in the sense that \(v^ \varepsilon\to u^*\) as \(\varepsilon\to 0\), in an appropriate norm. The paper concludes with an example of a parabolic variational inequality arising from the obstacle problem. To calculate \(u^*\) the method of closed-loop control with a model is used. In this procedure an auxiliary system is introduced, with control \(v^ \varepsilon\). Then an algorithm for the formation of \(v^ \varepsilon\) by feedback is designed. The connection to the original problem is that \(v^ \varepsilon\) approximates the unknown control \(u^*\). It is shown in fact that \(v^ \varepsilon\) converges weakly to the unknown control.











This page was built for publication: Inverse problems for variational inequalities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4028862)