Scheduling Exoplanet Observations Across A Network Of Telescopes

Abstract

Given the vast amount of observations that astronomers require for their work and the extremely limited resource that is observing time on a capable telescope, the problem of telescope scheduling is a vital one for observational astronomy. There is considerable work on developing and optimizing a schedule for observations on a single telescope. However, the literature largely lacks models for designing an optimal allocation of observations across a network of several telescopes. This paper addresses that need by creating a scheduler for exoplanet observations that is not restricted to a single telescope. We use multiprocessor scheduling as a model to access this problem and develop a heuristic-based approach. The basic problem is solved on a deadline-free set of observations using processing time heuristics. Over the course of the paper, strategies that account for the time remaining until an observation's deadline are incorporated. Weighting schemes are introduced to prioritize observations deemed more scientifically valuable. Furthermore, we explore the possibility that observations may conflict and be mutually exclusive for a single telescope. Finally, our algorithm includes a model for telescope unavailability, both due to planned and spontaneous reasons. This topic has obvious applications in today's world, where networks of observatories can be pooled together to further optimize the collective gathering of scientific knowledge. This paper also explores the applications of this scheduler to problems in finance, where large construction contractors must account for physical and financial constraints when planning projects.