Optimization of telescope scheduling: algorithmic research and scientific policy

Abstract

The use of very expensive facilities in Modern Astronomy has demonstrated the importance of automatic modes in the operation of large telescopes. As a consequence, several mathematical tools have been applied and developed to solve the (NP − hard) scheduling optimization problem: from simple heuristics to the more complex genetic algorithms or neural networks. In this work, the basic scheduling problem is translated into mathematical language and two main methods are used to solve it: neighborhood search methods and genetic algorithms; both of them are analysed. It is shown that the algorithms are sensitive to the scientific policy by means of the definition of the objective function (F) and also by the assignment of scientific priorities to the projects. The definition of F is not trivial and requires a detailed discussion among the Astronomical Community.