The LSST is a the nation's top priority next-gen ground-based astronomy project, with the objective ofconducting observations of the entire accessible sky, with about 800 visits per field. The scheduler for the projectwill determine the order in which these fields are observed, with the goal of maximizing some scientific merit function.A portion of that merit function has to do with Fourier coverage in the time domain.The variability in sky conditions (cloud cover, sky brightness and atmospheric seeing) makes the schedulingproblem non-trivial. It's a traveling salesman problem with a stochastic component, subject to certain constraints.Our goals are to1. devise a sensible quantitative framework that would accommodate various merit functions.2. assess whether an instantaneous (nightly) sequence of observations optimization will achieve a global optimum,3. build some numerical tools to make a toy model and try out some implementation schemes.
Some references
Hubble Space Telescope scheduler
genetic_edge_recombination_operator
genetic_alg_scheduler_SPIE2012