Approximate solution to a time optimal boundary control problem for the wave equation (Q267915)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Approximate solution to a time optimal boundary control problem for the wave equation
scientific article

    Statements

    Approximate solution to a time optimal boundary control problem for the wave equation (English)
    0 references
    0 references
    0 references
    12 April 2016
    0 references
    The authors present a numerical scheme for a time optimal boundary control problem involving the wave equation \(y_{tt}=y_{xx}-\theta (x)y\) for \(t>0\) and \(0<x<l\), with the boundary conditions \(-\beta _{0}y_{x}+\sigma _{0}y\mid _{x=0}=u_{0}(t)\), \(\beta _{1}y_{x}+\sigma _{1}y\mid _{x=l}=u_{1}(t)\) and the initial conditions \(y\mid _{t=0}=0=y_{t}\mid _{t=0}\). The control problem consists to find a pair \((u_{0\ast },u_{1\ast })\) such that the target \( f=(f^{0},f^{1})\in H^{1}(0,l)\times L^{2}(0,l)\) is attainable by \((y,y_{t})\) in the minimal time \(T_{\ast }\). The authors first prove that \( (f^{0},f^{1})\in H^{1}(0,l)\times L^{2}(0,l)\) is attainable in finite time \(T_{\ast }(f)\) if and only if the solution \(y\) to the problem \(y_{tt}=y_{xx}-\theta (x)y\), \(y\mid _{t=T}=f^{0}\), \(y_{t}\mid _{t=T}=f^{1}\) satisfies \(y(t,x)=0\) on \(\Omega _{T}\cap \Omega _{0}\) where \(\Omega _{s}=\{0\leq x\leq l\), \(s\leq t\leq s+\frac{l}{2}-\left| x-\frac{l}{2} \right| \}\times \{0\leq x\leq l\), \(s-\frac{l}{2}+\left| x-\frac{l}{2 }\right| \leq t\leq s\}\). Then they represent the optimal time \(T_{\ast }(f)\) as \(\inf_{f\in N(\mathfrak{B}_{T})}T\) where \(\mathfrak{B}_{T}\) is the linear operator defined through \(\mathfrak{B}_{T}f=y\) where \(y\) is the solution to the Cauchy problem \(y_{tt}=y_{xx}-\theta (x)y\), \(y\mid _{t=0}=f^{0}\), \(y_{t}\mid _{t=0}=-f^{1}\). The main idea of the numerical scheme consists to introduce modified controls using cuts and shifts. The authors prove a convergence result associated to these modified controls. They then introduce a discretization procedure considering square cells \( (t^{j}-\frac{\tau }{2},t^{j}+\frac{\tau }{2})\times (x_{i}-\frac{\tau }{2} ,x_{i}+\frac{\tau }{2})\) and representing the solution of the last Cauchy problem using d'Alembert's formula. In the last part of their paper, the authors describe the results they obtained using this numerical scheme in an attainable case and in a non attainable case.
    0 references
    time optimal boundary control problem
    0 references
    attainability
    0 references
    wave equation
    0 references
    Cauchy problem
    0 references
    numerical scheme
    0 references
    modified control
    0 references

    Identifiers