Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command

Large-scale simulation-based studies rely on at least three properties of pseudorandom number sequences. Since they behave like random numbers, they can be replicated and generated in parallel. However, there has been some divergence between empirical techniques employing random numbers and the stan...

Full description

Bibliographic Details
Main Author:	Owen Ozier
Format:	Policy Research Working Paper
Language:	en_US
Published:	World Bank, Washington, D.C. 2013
Subjects:	Applied Mathematics bootstrap bug Command confidence intervals cores Data Analysis Development Research Econometrics empirical methods Estimators integer Kurtosis Monte Carlo methods probability produce Public Services random numbers reliability Research Working Papers researchers sciences seed Seeds Simulation Simulations Skewness Stata statistical tests techniques yields
Online Access:	http://hdl.handle.net/10986/16374

id	okr-10986-16374
recordtype	oai_dc
spelling	okr-10986-163742021-04-23T14:03:28Z Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command Owen Ozier Applied Mathematics bootstrap bug Command confidence intervals cores Data Analysis Development Research Econometrics empirical methods Estimators integer Kurtosis Monte Carlo methods probability produce Public Services random numbers reliability Research Working Papers researchers sciences seed Seeds Simulation Simulations Skewness Stata statistical tests techniques yields Large-scale simulation-based studies rely on at least three properties of pseudorandom number sequences. Since they behave like random numbers, they can be replicated and generated in parallel. However, there has been some divergence between empirical techniques employing random numbers and the standard battery of tests used to validate them. A random number generator that passes tests for any single stream of random numbers may fail the same tests when it is used to generate multiple streams in parallel. The lack of systematic testing of parallel streams leaves statistical software with important potential vulnerabilities. This paper reveals one such vulnerability in Stata's rnormal function which went unnoticed for almost four years and how this error was detected. Furthermore, the paper discusses practical implications for the use of parallel streams in existing software. 2013-12-09T20:01:31Z 2013-12-09T20:01:31Z 2012-11 http://hdl.handle.net/10986/16374 en_US Policy Research Working Paper;No.6278 CC BY 3.0 IGO http://creativecommons.org/licenses/by/3.0/igo World Bank World Bank, Washington, D.C. Publications & Research :: Policy Research Working Paper Publications & Research
repository_type	Digital Repository
institution_category	Foreign Institution
institution	Digital Repositories
building	World Bank Open Knowledge Repository
collection	World Bank
language	en_US
topic	Applied Mathematics bootstrap bug Command confidence intervals cores Data Analysis Development Research Econometrics empirical methods Estimators integer Kurtosis Monte Carlo methods probability produce Public Services random numbers reliability Research Working Papers researchers sciences seed Seeds Simulation Simulations Skewness Stata statistical tests techniques yields
spellingShingle	Applied Mathematics bootstrap bug Command confidence intervals cores Data Analysis Development Research Econometrics empirical methods Estimators integer Kurtosis Monte Carlo methods probability produce Public Services random numbers reliability Research Working Papers researchers sciences seed Seeds Simulation Simulations Skewness Stata statistical tests techniques yields Owen Ozier Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command
relation	Policy Research Working Paper;No.6278
description	Large-scale simulation-based studies rely on at least three properties of pseudorandom number sequences. Since they behave like random numbers, they can be replicated and generated in parallel. However, there has been some divergence between empirical techniques employing random numbers and the standard battery of tests used to validate them. A random number generator that passes tests for any single stream of random numbers may fail the same tests when it is used to generate multiple streams in parallel. The lack of systematic testing of parallel streams leaves statistical software with important potential vulnerabilities. This paper reveals one such vulnerability in Stata's rnormal function which went unnoticed for almost four years and how this error was detected. Furthermore, the paper discusses practical implications for the use of parallel streams in existing software.
format	Publications & Research :: Policy Research Working Paper
author	Owen Ozier
author_facet	Owen Ozier
author_sort	Owen Ozier
title	Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command
title_short	Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command
title_full	Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command
title_fullStr	Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command
title_full_unstemmed	Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command
title_sort	perils of simulation : parallel streams and the case of stata’s rnormal command
publisher	World Bank, Washington, D.C.
publishDate	2013
url	http://hdl.handle.net/10986/16374
_version_	1764433008127901696

Perils of Simulation : Parallel Streams and the Case of Stata’s Rnormal Command

Similar Items