UWSpace is currently experiencing technical difficulties resulting from its recent migration to a new version of its software. These technical issues are not affecting the submission and browse features of the site. UWaterloo community members may continue submitting items to UWSpace. We apologize for the inconvenience, and are actively working to resolve these technical issues.
 

The pie sharing problem: Unbiased sampling of N+1 summative weights

Thumbnail Image

Files

The pie sharing problem- Unbiased sampling of N+1 summative weights.pdf (2.64 MB)

Date

2022-02

Authors

Mai, Juliane
Craig, James R.
Tolson, Bryan A.

Journal Title

Journal ISSN

Volume Title

Publisher

Elsevier

Abstract

A simple algorithm is provided for randomly sampling a set of N+1 weights such that their sum is constrained to be equal to one, analogous to randomly subdividing a pie into N+1 slices where the probability distribution of slice volumes are identically distributed. The cumulative density and probability density functions of the random weights are provided. The algorithmic implementation for the random number sampling are made available. This algorithm has potential applications in calibration, uncertainty analysis, and sensitivity analysis of environmental models. Three example applications are provided to demonstrate the efficiency and superiority of the proposed method compared to alternative sampling methods.

Description

Keywords

weight sampling, identical distributions, random number generation, constrained random numbers

LC Keywords

Citation

URI

https://doi.org/10.1016/j.envsoft.2021.105282
 http://hdl.handle.net/10012/18028

Collections

Waterloo Research
Civil and Environmental Engineering