A Tikhonov-based projection iteration for nonlinear ill-posed problems with sparsity con\-straints (Q2503143)

Let \(T:X\to Y\) be an ill-posed nonlinear operator. The computation of the solution to \(Tx=y\), where a sparse expansion is assumed, may be expressed as a minimization (over \(\ell_2\)) problem for a (nonlinear) functional like \(g\vdash J_\alpha (g):=| | y^\delta-T(F^*g)| | _Y^2 +2\alpha\Psi(Lg)\). To solve it, the authors develop an iterative scheme involving weighted \(\ell_p\) penalties on the coefficients. The usefulness of the obtained results is illustrated by solving the nonlinear SPECT problem with respect to quadratic and sparsity constraints.