The COMPASS platform is the assembling of an AO end-to-end simulation code, a real-time core, both optimized to run on GPUs, and an E-ELT observations simulator targeting imaging and spectroscopy on various astronomical object. In this section, we give an overview of the various components of the COMPASS platform.
COMPASS AO simulation platform is based on a novel original approach in which the simulation models and the real-time core are developed in a unified framework so as to provide optimal simulation performance and full flexibility for the integration of complex control schemes in real systems. It provides a unified environment to simulate efficiently various flavors and dimensionings of AO systems on GPU accelerated computers. Moreover, the development platform includes a prototype for a low latency acquisition system based on standard image acquisition protocols, and designed to optimize the data transfer throughput between the camera and the real-time core making COMPASS a real pathfinder for the use of accelerator-based architectures for real-time control in AO.
One of the goal of the COMPASS project is to produce end-to-end simulations to assist scientists and engineers in technical trade-off during instrument design studies. This strategy is depicted in the figure below : given a set of scientific inputs (from observations and/or numerical models), an end-to-end simulator produces fake data in FITS format, mimicking the product of real (reduced) observations. During this process, the Adaptive Optics system is taken into account through the delivered PSF which is generated using a dedicated pipeline. It is then possible to systematically explore the observational parameter space, which can be split into
parameters related to the telescope (e.g., diameter, throughput, thermal background), instrument (e.g., resolution, multiplex), and adaptive optics system (e.g., ensquared energy, number of actuators, number of GS). This parameter space can be constrained relatively to a given scientific goal (e.g., distinguishing a rotating disk from a galaxy merger). By successive iterations, one can find the optimal set of parameters to reach this science goal, which will in turn help deriving optimal trades-off between science and technical constrains.
