1 Introduction

Over the last decade optical interferometry has grown out of its pure experimental state. Optical interferometers are running routinely now, producing excellent science (Baldwin et al. 1998; Mourard et al. 1998; Traub 1998; Townes et al. 1998; Johnston 1998; Wallace et al. 1998; Davis et al. 1998). Soon new object classes will become observable with interferometers which provide large apertures as VLTI and Keck (Mariotti et al. 1998; Colavita et al. 1998). Within the next ten years space borne interferometry will offer many advantages including the absence of atmospheric absorption and turbulence, and the possibility to achieve very long baselines (Shao 1998; Penny et al. 1998).

Modern interferometers are highly complex systems combining subsystems of different engineering disciplines, mainly active/adaptive optics, control engineering, electronics and structural mechanics. They typically include a set of nested control loops for various tasks, e.g. image tracking, tip-tilt control, higher-order adaptive optics, fringe tracking, laser metrology or attitude and orbit control system (AOCS). Most of these loops depend intimately on the object observed, and it is very difficult, if not impossible, to calibrate all their effects by observing unresolved stars (Colavita 1999). In combination with experimental work numerical modeling is a powerful technique to study the interactions between the subsystems mentioned above. Furthermore it can be used for preparation of astronomical observations.

The work presented here is a versatile approach to model the imaging process of an interferometer. A computer program package for simulating the dynamic behavior of the VLTI has been developed at ESO. Though targeting mainly at the VLTI the open and flexible structure of the program modules allows their application to a wide range of astronomical interferometers. The total process of imaging an extended astronomical object in the aperture synthesis observational mode under the influence of various perturbations can be simulated.

In Sect.2 we describe the philosophy behind our modeling approach. The description of an interferometer by the incoherent space invariant imaging equation is analyzed, taking pupil mapping under special consideration. Following in Sect.3 is the description of how the dynamic response of an interferometer is calculated. Building on this result Sect.4 investigates the computation of the interferometer's point spread function, taking into account multiaxial and coaxial beam combination, and looking into temporal coherence and observation of polarized objects. Imaging of extended objects is explained in Sect.5, followed by some illustrative examples from the model (Sect.6) and the summary (Sect.7).