Multidimensional radiative transfer modeling: Indispensable tool for interpretation of interferometry observations

In order to better understand and correctly interpret complex astrophysical reality, it is necessary to invest time and money not only into the extremely complex and costly Darwin interferometer itself, but also into the development of multidimensional radiative transfer modeling. New observational technologies require much more sophisticated numerical tools for routine interpretation of the rapidly increasing amounts of high-resolution images of complex circumstellar structures. We need very efficient two- and three-dimensional radiative transfer methods coupled with an automated search through a very large parameter space and with realistic estimates of model error bars. We have to learn how to construct accurate, self-consistent models which would take into account and quantitatively reproduce the entire set of spatially-resolved measurements accumulated to date for any individual object. We illustrate the ideas by presenting recent detailed models of two young stellar objects (L1551 IRS 5, HL Tau) and two evolved post-AGB stars (the Red Rectangle, IRC + 10 216) as examples which demonstrate both advantages and difficulties of self-consistent radiative transfer modeling. Being at present a complex approach, requiring a lot of computational resources, it is the only way to reduce the otherwise unacceptably large uncertainties of the physical interpretations and parameters derived on the basis of much simpler considerations. One can expect suitable computational power relatively soon, at least by the time Darwin starts its observations.