Thesis A MILP BASED SCHEDULING MODEL FOR ASTRONOMY
Loading...
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Tecnica Federico Santa Maria
Abstract
Los desafíos en torno a la astronomía han llevado a invertir en grandes proyectos astronómicos para observar con más detalle los rincones más remotos del universo. Algunos grandes proyectos como ALMA o E-ELT se construyen gracias a una enorme inversión y por lo tanto, es relevante considerar varias estrategias de optimización que permitan reducir al mínimo los costos y maximizar los resultados esperados. En este trabajo. se describe un algoritmo de planificación basado en programación lineal entera mixta con el fin de proporcionar una solución para el problema de calendarización astronómica basada en la situación de ALMA que permita optimizar el impacto científico de las observaciones programadas y el tiempo de utilización del telescopio. Teniendo en cuenta la complejidad del problema, se han realizado algunas simplificaciones del problema original. La validación de la solución propuesta se llevó a cabo utilizando un prototipo desarrollado en C++ y evaluado mediante el uso de instancias sintéticas basadas en datos reales y volúmenes de datos que ALMA espera tener en los próximos años. Por último, los resultados muestran que la solución propuesta ofrece una primera aproximación para resolver eficazmente el problema de calendarización de ALMA.
The challenges around astronomy have led to invest in large astronomical projects in order to observe with more detail the most remote corners of the universe. Some large projects like ALMA or E-ELT are built thanks to a huge investment and so, is relevant to consider several optimization strategies that allow to minimize the involved costs and maximize the expected results. In this work, a scheduling algorithm based on mixed integer linear programming is described seeking to provide a solution for the astronomical scheduling problem based on ALMA's situation that optimize the scientific impact of the scheduled observations and the utilization time of the telescope. Considering the problem's complexity. some simplifications of the original problem have been made. The validation of the proposed solution was carried out using a prototype implementation developed in C++ and tested by using synthetical instances based on realistic data characterizations and expected volumes of demand that ALMA will have in the next years. Finally, the results shows that the proposed solution provides an initial approach to solve efficiently the scheduling problem of ALMA.
The challenges around astronomy have led to invest in large astronomical projects in order to observe with more detail the most remote corners of the universe. Some large projects like ALMA or E-ELT are built thanks to a huge investment and so, is relevant to consider several optimization strategies that allow to minimize the involved costs and maximize the expected results. In this work, a scheduling algorithm based on mixed integer linear programming is described seeking to provide a solution for the astronomical scheduling problem based on ALMA's situation that optimize the scientific impact of the scheduled observations and the utilization time of the telescope. Considering the problem's complexity. some simplifications of the original problem have been made. The validation of the proposed solution was carried out using a prototype implementation developed in C++ and tested by using synthetical instances based on realistic data characterizations and expected volumes of demand that ALMA will have in the next years. Finally, the results shows that the proposed solution provides an initial approach to solve efficiently the scheduling problem of ALMA.
Description
Digitalizado de su versión en papel
Keywords
OBSERVATORIO ALMA, C++ (LENGUAJE DE PROGRAMACIÓN), PROGRAMACION LINEAL, ALGORITMOS MODELO MATEMATICO BIG DATA (VOLUMENES DE DATOS)
Citation
Campus
Universidad Técnica Federico Santa María UTFSM. Campus San Joaquín