1: \begin{abstract}

2: Many naturally-occuring models in the sciences are well-approximated

3: by simplified models, using multiscale techniques. In such settings

4: it is natural to ask about the relationship between inverse problems

5: defined by the original problem and by the multiscale approximation.

6: We develop an approach to this problem and exemplify it in the context

7: of optical tomographic imaging.

8: 

9: Optical tomographic imaging is a technique for infering the properties of 

10: biological tissue via measurements of the incoming and outgoing light intensity; it may be used as a

11: medical imaging methodology. Mathematically, light propagation is modeled 

12: by the radiative transfer equation (RTE), and optical tomography amounts to 

13: reconstructing the scattering and the absorption coefficients in the RTE 

14: from boundary measurements. We study this problem in the Bayesian framework, 

15: focussing on the strong scattering regime. In this regime the forward 

16: RTE is close to the diffusion equation (DE). 

17: We study the RTE in the asymptotic regime where the forward problem

18: approaches the DE, and prove convergence of the inverse RTE to the inverse DE in both nonlinear 

19: and linear settings. Convergence is proved by studying the distance between 

20: the two posterior distributions using the Hellinger metric, and using

21: Kullback-Leibler divergence.

22: 

23: 

24: \end{abstract}

25: