GTD, together with Institute IEEC, presented the Research study on schedulers for astronomical observatories within the theme "Observatory Operations: Strategies, Processes, and Systems IV ". This study was an outcome of the SCTA (Scheduler for the Cherenkov Telescope Array) that is being developed in collaboration with the Institute of Space Studies de Catalunya (IEEC) and funded by the CDTI. SCTA aims to study a scheduling algorithm that allows to provide a scheduler innovative and competitive network of Cherenkov telescopes (CTA), a international project which includes hundreds of scientists and engineers from more than 20 countries. CTA has already been completed design study phase (2006-2010) and the project now in the phase school (2010-2013) that precedes the construction phase (2013-2018). Phase one will decide the design CTA and exact start prototyping.. CTA is a ground-based observatory to study the universe in gamma rays of very high energy that will feature dozens of telescopes in two or three different sizes covering a surface around a square kilometer in the northern hemisphere and around ten square kilometers in the southern hemisphere. The planner required by the CTA will have to optimize different objectives as example maximize the scientific return and minimize the time between observations. Also have to deal with a more or lesser restrictions and priorities as may be environmental conditions, time constraints defined by the very scientific or zenith angle at which a target is observed. To achieve the objectives are being studied in conjunction with different algorithms multiobjective techniques for the treatment of constraints (CSPs). In the framework of the SCTA, the system being developed by Pau Colomer (GTD) is designed to generate schedules for the short and long term. Planning for the long term seeks to create a set of schedules that distribute the different blocks within an observation period longer or shorter time (1 to 6 months) focusing, above all, to optimize all the objectives to optimize rather than to perform all restrictions. For example, a time period may contain more than one block observation. In contrast, the short term is focused on planning a new estimate night of observation in near real time to adapt to changing different restrictions and / or priorities. For example, a change of weather may involve one or more observations can not be executed. For this reason, planning is necessary to recalculate the night to take into account new conditions. In order to compute optimized schedules that take into account all the restrictions and priorities is working on a stochastic algorithm based on neural networks called GDSN (Guarded Discrete Stochastic Network) created initially to plan Hubble Space Telescope. This algorithm is used in both planning cycles (short and long term). To optimize all objectives simultaneously to avoid conflicts between them, work is in the combination of GDSN algorithm, described previously, with algorithms multiobjective that allow us to generate optimized joint planning within the cycle of long-term planning. Finally, to take into account the long-term planning within the planning add a short term priority in the short term, containing the information calculated in the long term. For more information, please see the following: • The official website of CTA: http://www.cta-observatory.org/ • The Spanish website of CTA: http://www.observatorio-cta.es/ • The official website of the SPIE conferences: http://spie.org/x13662.xml
