1: \begin{abstract}

2: We propose a novel numerical inversion algorithm for the coefficients 

3: of parabolic partial differential equations, based on model reduction. 

4: The study is

5: motivated by the application of controlled source electromagnetic

6: exploration, where the unknown is the subsurface electrical

7: resistivity and the data are time resolved surface measurements of the

8: magnetic field. The algorithm presented in this paper considers

9: inversion in one and two dimensions. The reduced model is obtained

10: with rational interpolation in the frequency (Laplace) domain and a

11: rational Krylov subspace projection method. It amounts to a non-linear

12: mapping from the function space of the unknown resistivity to the

13: small dimensional space of the parameters of the reduced model. We use

14: this mapping as a non-linear preconditioner for the Gauss-Newton

15: iterative solution of the inverse problem. The advantage of the

16: inversion algorithm is twofold.  First, the non-linear preconditioner

17: resolves most of the nonlinearity of the problem.  Thus the iterations

18: are less likely to get stuck in local minima and the convergence is

19: fast. Second, the inversion is computationally efficient because it

20: avoids repeated accurate simulations of the time-domain response.  

21: We study the stability of the inversion algorithm

22: for various rational Krylov subspaces, and assess its performance with

23: numerical experiments.

24: \end{abstract}

25: