HDR Random Number Generator
The HDR Random Number Generator removes the need to distribute actual trials. Designed by Doug Hubbard, the HDR pseudo random number generator allows simulations running on diverse computer platforms to generate either identical or independent streams of random numbers as required though a multi-dimensional seed. This allows simulation results to be networked together while maintaining statistical coherence. The generator performs well under the respected Dieharder random number tests and has formulas that fit into a single cell of Excel to perform interactive, as opposed to recursive, simulations. This last property is perhaps the most important, as it maintains access to over a billion users, as opposed to more sophisticated generators that require more complex computations.
View slide presentation: A Simple Pseudo Random Number Generator
HDR 2.0 Standard Seed Assignment
To maintain coherence among virtual SIPs, a standard seed assignment is assumed for the HDR 2.0 generator. There are five inputs to the HDR formula regardless of environment.
| counter |
The Monte Carlo Iteration Counter, PM_Index when using the SIPmath tools or ChanceCalc. |
|---|---|
| entity |
A seed reserved for an entity such as a department or a company or government organization to ensure that they will not inadvertently create statistical dependence when not needed, but may enforce dependence when needed, such as with global commodity prices. |
| varId |
A seed reserved for each variable in a model. |
| seed3 |
An optional seed that may be assigned to unique identifiers such part or project numbers. |
| seed4 |
A second optional seed. |
This figure shows the positions in the formula of the Monte Carlo iteration counter
Download HDR_2_0 Seed Standard.xlsx to double-check results for your implementation in other environments.