Optimal design under the one-dimensional wave equation (Q933107)

Given the one-dimensional domain \(\Omega=(0,1)\), two materials \(m_\alpha\) and \(m_\beta\), a finite horizon \(T>0\), the authors are interested in the following optimal design problem: to find the best distribution of the material \(m_\alpha\) inside \(\Omega\) (represented by the characteristic function \(\chi(t,\cdot)\) on \(\Omega\)) such that the functional \[ I(\chi)=\int_0^T\int_\Omega\left(u_t(t,x)^2+a\left(t,x,\chi(t,x)\right)u_x(t,x)^2\right)dxdt, \] is minimized subject to a constraint given by a suitable initial/boundary value problem for the one-dimensional wave equation \[ u_{tt}-\text{div}\left(\left(\alpha\chi+(1-\beta)\chi\right)u_x\right)=0\;\text{in } (0,T)\times\Omega. \] Here \(a(\cdot,\cdot,\cdot)\), \(\alpha\) and \(\beta\) are given coefficients. A constraint on the total amount of material \(m_\alpha\) at disposal is also given. It is well known that such a problem does not admit a solution (i.e., a minimizing distribution). Hence the authors propose a suitable relaxation of the problem in the sense of a quasi-convexification by using also Young measures. The authors investigate the nature of such a relaxation in connection with the original problem, give some results about the microstructures of the original optimal design problem and formulate a conjecture concerning further results. Numerical experiments are also presented.