Sparse initial data identification for parabolic PDE and its finite element approximations (Q2356553)

In this paper, optimal control and approximate controllability arguments are used to solve the problem of inverse source identification for parabolic equations. The authors prove that approximate controllability can be achieved by sparse controls (measures) that can be activated in a finite number of pointwise locations. Optimal initial measures are characterized with the help of minimization of some quadratic functional on the solutions of the adjoint equation. Also, a numerical approach to the considered inverse control problem is presented. The authors use the finite element approximation and discuss some convergence results. The ample bibliography contains 32 items.