Optimal control of an elliptic system with a nonsymmetric state operator (Q852247)

The authors consider a second-order equation: \[ \sum^n_{i,j=1} \partial/\partial x_i\bigl(k_{ij}(\partial y/\partial x_j)\bigr)+ \sum^n_{j=1}\partial/\partial x_i \bigl(b_i(x)y\bigr)+q(x)y=f(x),\tag{1} \] with \(k_{ij}=k_{ji}\), and \(q(x)\geq 0\), in bounded connected Lipschitz domains, respectively, \(\Omega_1\), \(\Omega_2\) in \(\mathbb{R}^n\), with Dirichlet boundary condition \(y\equiv 0\) on the boundary. Variable \(y\) is assumed to be element of \(H_0\). The authors introduce an obvious inner product \(a(y,v)\) in the space \(H_0\), containing also a boundary term caused by a cut in the domain containing imperfect contact conditions, thus causing the loss of symmetry. Authors refer to Mikhlin's 1970 classic monograph, to justify use of Friedrich's inequality and also justify validity of the Lax-Milgram lemma. They introduce control \(u\), which is an element of a Hilbert space \(H\). In fact all operations are assumed to be mappings of Hilbert spaces into Hilbert spaces, all of them being some versions of the \(L^2\) space. The lack of symmetry is obviously an inconvenient occurrence. The problem they tackle is really analysis of the static response of an \(n\)-dimensional ``membrane''. Deflection of this strange ``membrane'' is redefined in higher dimensions, subject to a control attempting to minimize a rather complex cost functional. Another Hilbert space of operators containing observation operator \(C\) is introduced, with \(C(y(u))=y(x,u)\). Authors select some specific element of \(C(y)=z_g\) and define the cost functional in terms of a sum of inner products containing the term \(\|(z-y(0))\|^2\) and other terms that represent distances from \(y(0)\). The paper is lengthy and rather difficult to read, unless one spends considerable amount of time reading some earlier papers of these authors. Yet it is well written, and treats in detail control of higher-dimensional second-order elliptic PDE. One is tempted to find some common trends with the sophisticated papers of Lasiecka, Triggiani, Lagnese, Avalos, Tataru, Shubova, and their students, pursuing control of membranes, beams, and shallow shells, whereby the Lasiecka and/or Triggiani and other cooperators, were forced to introduce some rather exotic topological spaces, to fit their problems. However, this is not a one-sided lack of exchanging information. The classic monographs of J. L. Lions and P. G. Ciarlet are the only references to work originating outside the former Soviet Union, in this and other recent publications of the authors. This is reciprocated by many western authors, who largely ignore publications of P.D.E. control theorists still residing in Russia or Ukraine.