We present the basic elements and first results of an end-to-end simulation package whose purpose is to test the validity of the Space Interferometer Mission design. The fundamental simulation time step is one millisecond, with substructure at 1/8 ms, and the total duration of the simulation is five years. The end product of a given 'wide-angle' astrometry run is an estimated grid star catalog over the entire sky with an accuracy of about 4 micro-arcseconds.
SIMsim is divided into five separate modules that communicate via data pipes. The first generates the 'truth' data on the spacecraft structure and laser metrology. The second module generates uncorrupted fringes for the three SIM interferometers, based on the current spacecraft orientation, target stars' positions, etc. The third module reads out the CCD white light fringe data at specified times, corrupting that and the metrology data with appropriate errors. The data stream out of this module represents the basic data stream on the simulated spacecraft. The fourth module performs fringe-fitting tasks on this data, recovering the total path delay, and the fifth and final module inverts the entire metrology/delay dataset to ultimately determine the instantaneous path delay on a fiducial baseline fixed in space. (Pathlength feed forward is used every few milliseconds to re-position the interferometer to keep the fringes in the delay window.) The average of all such delays over an integration time (typically 30 s) is reported as one of several hundred thousand measured stellar delays over the five-year period, which are then inverted to produce the simulated catalog. Future plans include taking into account more sources of error from the SIM error budget and including narrow angle observations in the observing plan.