On Converse and Saturation Results for Tikhonov Regularization of Linear Ill-Posed Problems (Q4337452)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1011187
Language Label Description Also known as
default for all languages
No label defined
    English
    On Converse and Saturation Results for Tikhonov Regularization of Linear Ill-Posed Problems
    scientific article; zbMATH DE number 1011187

      Statements

      On Converse and Saturation Results for Tikhonov Regularization of Linear Ill-Posed Problems (English)
      0 references
      0 references
      14 October 1997
      0 references
      linear ill-posed problems
      0 references
      Tikhonov regularization
      0 references
      Hilbert spaces
      0 references
      minimum norm solution
      0 references
      For a bounded linear operator \(T:X\to Y\) between Hilbert spaces \(X\) and \(Y\) with a non-closed range \(R(T)\) and for fixed \(y\in Y\), the equation under investigation is \(Tx=y\) which is supposed to have a minimum norm solution \(x^\dagger\in X\). It is well known that for \(0<\nu\leq 1\), the condition (1) \(x^\dagger\in R((T^*T)^\nu)\) is sufficient for \[ \sup\Big\{\inf_{\alpha>0}|x(\alpha,y_\delta)-x^\dagger|\mid y_\delta\in Y,\;|Q(y_\delta-y)|\leq\delta\Big\}=O(\delta^\frac{2\nu}{2\nu+1})\tag{2} \] to hold. Here \(Q\) denotes the orthogonal projection onto the closure of \(R(T)\), the operator \(T^*:Y\to X\) is the adjoint of \(T\), and \(x(\alpha,y_\delta)=(T^*T+\alpha I)^{-1}T^*y_\delta\) are the Tikhonov approximations. If \(\nu<1\) then condition (2) does not imply condition (1), however, and it is one subject of this paper to provide conditions that are equivalent to (2). In fact, it is shown for \(0<\nu<1\) that \[ (2)\text{ holds}\quad\Longleftrightarrow\quad|x(\alpha,y)-x^\dagger|=O(\alpha^\nu)\quad\Longleftrightarrow\quad\int_0^\mu d|E_\lambda x^\dagger|^2=O(\mu^{2\nu}),\tag{3} \] and then also \(x^\dagger\in R((T^*T)^\rho)\) holds for each \(0<\rho<\nu\). Here \(\{E_\lambda\}\) denotes the resolution of the identity corresponding to \(T^*T\), and the Landau symbol \(O\) is applied with respect to small positive parameters, respectively. The statement (3) remains valid if all arising \(O\)'s in (2)--(3) are replaced by \(o\)'s, respectively. The paper contains also a saturation result which yields a generalization for non-compact operators: if the condition \[ \sup\Big\{\inf_{\alpha>0}|x(\alpha,y_\delta)-x^\dagger|\mid y_\delta\in Y,\;|Q(y_\delta-y)|\leq\delta\Big\}=o(\delta^{2/3}) \] is fulfilled then necessarily \(x^\dagger=0\) holds.
      0 references

      Identifiers