Processes with long memory: Regenerative construction and perfect simulation
Comets, Francis ; Fernández, Roberto ; Ferrari, Pablo A.
Ann. Appl. Probab., Tome 12 (2002) no. 1, p. 921-943 / Harvested from Project Euclid
We present a perfect simulation algorithm for stationary processes indexed by $\mathbb{Z}$, with summable memory decay. Depending on the decay, we construct the process on finite or semi-infinite intervals, explicitly from an i.i.d. uniform sequence. Even though the process has infinite memory, its value at time 0 depends only on a finite, but random, number of these uniform variables. The algorithm is based on a recent regenerative construction of these measures by Ferrari, Maass, Martínez and Ney. As applications, we discuss the perfect simulation of binary autoregressions and Markov chains on the unit interval.
Publié le : 2002-08-14
Classification:  Perfect simulation,  regeneration,  chains with complete connections,  binary autoregressions,  60G99,  68U20,  60K10,  62J02
@article{1031863175,
     author = {Comets, Francis and Fern\'andez, Roberto and Ferrari, Pablo A.},
     title = {Processes with long memory: Regenerative construction and perfect simulation},
     journal = {Ann. Appl. Probab.},
     volume = {12},
     number = {1},
     year = {2002},
     pages = { 921-943},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1031863175}
}
Comets, Francis; Fernández, Roberto; Ferrari, Pablo A. Processes with long memory: Regenerative construction and perfect simulation. Ann. Appl. Probab., Tome 12 (2002) no. 1, pp.  921-943. http://gdmltest.u-ga.fr/item/1031863175/