Convex sweeping process in the framework of measure differential inclusions and evolution variational inequalities (Q484127): Difference between revisions

In the early 70's, J. J. Moreau studied sweeping processes and their applications. As an example, in such processes a point \(u(t)\) must remain in a closed convex set \(C(t)\) and its velocity must belong to the inward normal cone \(-N(C(t);u(t))\) at \(u(t)\) (of course this cone can be the zero set). The mathematical law of motion is then a differential inclusion with side conditions. That inclusion was generalized by adding a set-valued mapping \(F(t;u(t))\) on the set side of the inclusion and many results under different conditions exist, obtained in the last 40 years. A good readable overview can be found in this paper under review. Here the authors give existence und unity theorems for two new versions of such inclusions followed by motivations and applications. First, instead of \(F\) they add a Lipschitz perturbation function \(f\) and the variation of \(C(t)\) is given by a Radon measure. So they get a measure differential inclusion. Second, introducing positive semi-definite linear bounded operators on the left side of the inclusion -- \(f\) is continuous and the normal cone \(N\) now depends on the velocity of \(u\) -- they get (for almost all \(t\)) an evolution variational inclusion of the form \[ A_1 \dot{u}(t) + A_0 u(t) - f(t) \subset -N(C(t);\dot{u}(t)), \;u(0)=u_0 \in H, \] where \(H\) is a Hilbert space. For both the mentioned problems they give full proofs for the existence and uniqueness of (well-defined) solutions (under special conditions esp. for \(f\) and the operators \(A_0, A_1\)) using especially the methods of convex analysis, some results from measure theory, Sobolev's embedding theorem, Stampacchia's classical result, Gronwall's lemma and Moreau's catching up algorithm. Counter examples and remarks on numerical experiments supplement the paper.