1: \begin{abstract}

2:   In this paper, we propose an algorithm for the construction of

3:   low-rank approximations of the inverse of an operator given in

4:   low-rank tensor format. The construction relies on an updated greedy

5:   algorithm for the minimization of a suitable distance to the inverse

6:   operator. It provides a sequence of approximations that are defined

7:   as the projections of the inverse operator in an increasing sequence

8:   of linear subspaces of operators. These subspaces are obtained by

9:   the tensorization of bases of operators that are constructed from

10:   successive rank-one corrections. In order to handle high-order

11:   tensors, approximate projections are computed in low-rank

12:   Hierarchical Tucker subsets of the successive subspaces of

13:   operators. Some desired properties such as symmetry or sparsity can

14:   be imposed on the approximate inverse operator during the correction

15:   step, where an optimal rank-one correction is searched as the tensor

16:   product of operators with the desired properties. Numerical examples

17:   illustrate the ability of this algorithm to provide efficient

18:   preconditioners for linear systems in tensor format that improve the

19:   convergence of iterative solvers and also the quality of the

20:   resulting low-rank approximations of the solution.

21:  

22: \end{abstract}

23: