Abstract
This paper is devoted to the development of an astrometric solver aiming to reveal precise stellar parameters by mining the data provided by a satellite telescope. First, we focus on the self-calibration model of the satellite telescope intended to absorb the telescope’s optical and mechanical distortions and the satellite’s imprecise attitude. Our experiments reveal that the calibration model based on the Legendre polynomials can cope with different types of input data distortions. Second, we demonstrate the performance of the solver for different test cases on the CPU cluster. So far we showed that 275 millions of stellar observations can be processed on 64 cluster nodes within 2.5 hours. The viability of this approach is proven along with the indication of bottlenecks and challenges for future work.