Hubbard HDR Library
Type: Library, Extra
The HDR is a simple, open-source pseudo-random number generation algorithm. It has been embraced by organizations such as ProbabilityManagement.org as a standard for efficiently exchanging exact Monte Carlo samples. Given 4 integer seed values, the sequence is deterministic, so two parties with these same 4 values can reproduce precisely the same sequence up to a length of 100M samples, even from Excel. Select a different combination of seeds to prevent any two independent simulations from inadvertently using the same random number stream. This ensures that independent simulation results can be combined without unintended correlations. Or use the same combination of seeds to intentionally produce identical simulations of the same stream. This allows entire Monte Carlo simulation samples to be communicated in just a few numbers.
This library provides the functions for using HDR in an Analytica model. The Hubbard_HDR(...) function enables you to obtain the nth random number in a sequence, while the HDR_U(...) function acts as a Uniform(0,1) distribution as is convenient to use in any Inverse Cumulative Probability Function (e.g., CumKeelinInv, CumNormalInv, CumBetaInv, etc) to encode a specific distribution inside a model.
Reference
- Douglas Hubbard (Aug 2019), "A Multi-Dimensional, Counter-Based Pseudo Random Number Generator as a Standard for Monte Carlo Simulations", Winter Simulation Conference, Washington DC, DOI: 10.1109/WSC40007.2019.9004773.
Download: Hubbard HDR Library
Authors: Doug Hubbard, Hubbard Decision Research, and Lonnie Chrisman, Lumina Decision Systems
Keywords: Uncertainty, Statistics