DocumentCode :
3746879
Title :
Efficient simulation for branching linear recursions
Author :
Ningyuan Chen;Mariana Olvera-Cravioto
Author_Institution :
Industrial Engineering and Operations Research, Columbia University, New York, 10027, USA
fYear :
2015
Firstpage :
2716
Lastpage :
2727
Abstract :
We provide an algorithm for simulating the unique attracting fixed-point of linear branching distributional equations. Such equations appear in the analysis of information ranking algorithms, e.g., PageRank, and in the complexity analysis of divide and conquer algorithms, e.g., Quicksort. The naive simulation approach would be to simulate exactly a suitable number of generations of a weighted branching process, which has exponential complexity in the number of generations being sampled. Instead, we propose an iterative bootstrap algorithm that has linear complexity; we prove its convergence and the consistency of a family of estimators based on our approach.
Publisher :
ieee
Conference_Titel :
Winter Simulation Conference (WSC), 2015
Electronic_ISBN :
1558-4305
Type :
conf
DOI :
10.1109/WSC.2015.7408378
Filename :
7408378
Link To Document :
بازگشت