Numerical solution of an optimization problem with unknown boundary (Q760179)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Numerical solution of an optimization problem with unknown boundary
scientific article

    Statements

    Numerical solution of an optimization problem with unknown boundary (English)
    0 references
    0 references
    1984
    0 references
    Two numerical algorithms are proposed for the following optimization problem: Given \(S_ 0>0\) and a bounded simple closed curve \(\Gamma_ 0\subset R^ 2\), in the family \({\mathcal D}\) of all bounded double connected domains \(\Omega \subset R^ 2\) that have area \(S_ 0\) and have \(\Gamma_ 0\) as interior boundary, find \(\Omega^*\in {\mathcal D}\) that minimizes \(F(\Omega)=\int_{\Omega}| \nabla u_{\Omega}|^ 2dx_ 1dx_ 2\) where \(u_{\Omega}\) is harmonic on \(\Omega\) and satisfies the conditions: \(u_{\Omega}| \Gamma_ 0=1\), \(u_{\Omega}| \Gamma_ c=0\), \(\Gamma_ c\) being the outside boundary of \(\Omega\). The first algorithm, based on some necessary optimality conditions, starts from an admissible \(\Omega_ 0\in {\mathcal D}\) and finds a sequence of admissible domains, \(\Omega_ n\in {\mathcal D}\) for which \(F(\Omega_ n)\) is decreasing; the second algorithm is a variant of the first one in the case \(\Gamma_ 0\) is star-shaped with respect to the origin. Numerical experiments on four examples are presented.
    0 references
    0 references
    0 references
    0 references
    0 references
    free boundary optimization problems
    0 references
    necessary optimality conditions
    0 references
    numerical algorithms
    0 references
    Dirichlet problems
    0 references
    Numerical experiments
    0 references