scientific article; zbMATH DE number 19642

From MaRDI portal
Publication:3979135





[For the entire collection see Zbl 0732.00028.] Let \(w(x,t)\) be the solution to the following parabolic boundary value problem: \[ \begin{aligned} w_ t(x,t)=w_{xx}(x,t)-q(x)\cdot w(x,t) &\text{ for }(x,t)\in (0,1)\times (0,T),\\ h_ 1\cdot w(0,t)-k_ 1\cdot w_ x(0,t)=u(t) &\text{ for }t\in (0,T),\\ h_ 2\cdot w(1,t)+k_ 2\cdot w_ x(1,t)=0 &\text{ for } t\in (0,T),\\ w(x,0)= w_ 0(x)\in L^ 2(0,1) &\text{ for } x\in (0,1),\end{aligned} \] where \(q(x)\in C^ \infty[0,1]\) and \(h_ 1\), \(k_ 1\), \(h_ 2\), \(k_ 2\) are constants. Let \(y(t)=h\cdot w(x_ 0,t)+k\cdot w_ k(x_ 0,t)\), where \(x_ 0\in[0,1]\) and \(h\), \(k\) are constants. The problem of reconstructing \(u(t)\) on the basis of \(N\) measurements \(y_ 1,y_ 2,\dots,y_ N\) is considered, where \(| y_ i-y(t_ i)|<\varepsilon\) for \(i=1,2,\dots,N\) and \(\varepsilon>0\) is given. The problem is reduced to a Volterra integral equation of convolution type: \(y(t)=K* u(t)\). Tikhonov regularization is used to obtain a theoretical solution which is stable under small perturbation in \(y\). The regularization is based on a quadratic penalty function determined by a constant coefficient linear differential operator. A numerical procedure using cubic spline interpolation of \(\{y_ j\}^ N_{j=1}\) is proposed. Numerical examples are given.




Cited in
(29)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3979135)