1: \begin{abstract}

2: Block coordinate  descent (BCD) methods   approach  optimization problems by  performing gradient steps along alternating  subgroups of coordinates.

3: This is in contrast to full gradient descent,  where  a gradient step

4: updates all coordinates simultaneously.  BCD has  been demonstrated to accelerate the  gradient

5: method in many practical large-scale applications. Despite its success

6: no convergence analysis  for inverse problems is known so far.

7: In  this paper, we investigate the BCD method for solving linear  inverse problems. As main theoretical result, we show that for operators having a particular tensor product form,

8: the BCD method combined with an appropriate stopping criterion yields a convergent regularization method.

9: To illustrate the theory, we  perform numerical experiments comparing the BCD and the full gradient descent method   for a system  of integral equations.

10: We also present numerical tests for a non-linear inverse problem not covered by our theory, namely one-step inversion in multi-spectral X-ray tomography.

11: 

12: \medskip \noindent \textbf{Keywords:} ill-posed problems, convergence analysis, regularization theory, coordinate descent, multi-spectral CT

13: 

14: \medskip \noindent \textbf{MSC2010:} 65J20, 44A12, 47J06.

15: 

16: 

17: \end{abstract}

18: