A decomposition approach for correlated random vector generation

"When doing simulation studies, it is necessary to consider correlation in input variables in order to obtain correct results. Generating correlated random nurnbers is not always an easy task, as the accuracy and computational stability of some methods depend of the probabilit, y distribution o...

Full description

Autores:
Guaje Acosta, Oscar Orlando
Tipo de recurso:
Fecha de publicación:
2016
Institución:
Universidad de los Andes
Repositorio:
Séneca: repositorio Uniandes
Idioma:
spa
OAI Identifier:
oai:repositorio.uniandes.edu.co:1992/61040
Acceso en línea:
http://hdl.handle.net/1992/61040
Palabra clave:
Correlación (Estadística)
Métodos de simulación
Optimización matemática
Rights
openAccess
License
https://repositorio.uniandes.edu.co/static/pdf/aceptacion_uso_es.pdf
Description
Summary:"When doing simulation studies, it is necessary to consider correlation in input variables in order to obtain correct results. Generating correlated random nurnbers is not always an easy task, as the accuracy and computational stability of some methods depend of the probabilit, y distribution of input variables. We present a method based on mixed-integer programrning to generate correlated random numbers. Since this method does not perform acceptably, we show a column generation procedure used to accelerate the MIP. We implemented our method and found significant computational improvements over the base MIP while improving the accuracy of the solution over known methods.".-- Tomado del resumen.