Space-Filling Designs

Space-filling designs are useful in situations where run-to-run variability is of far less concern than the form of the model. Space filling designs include sphere packing Latin hypercube, uniform, minimum potential, maximum entropy, Gaussian process, and fast flexible designs.

Consider a sensitivity study of a computer simulation model. In this situation, and for any mechanistic or deterministic modeling problem, any variability is small enough to be ignored. For systems with no variability, replication, randomization, and blocking are irrelevant.

The Space Filling platform provides multiple design types for situations with only continuous factors. A special design type exists for designs that include categorical, discrete numeric or mixture factors. For continuous factors, space-filling designs have two objectives:

• maximize the distance between any two design points

• space the points uniformly

Figure 21.1 Space-Filling Design 

Contents

Overview of Space-Filling Designs

Example of a Space-Filling Design

Build a Space-Filling Design

Responses

Factors

Define Factor Constraints

Space-Filling Design Methods

Design

Design Diagnostics

Design Table

Space Filling Design Options

Additional Examples of Space Filling Designs

Example of Creating a Latin Hypercube Design

Example of Creating a Uniform Design

Example Comparison of Sphere-Packing, Latin Hypercube, and Uniform Methods

Example of a Minimum Potential Design

Example of a Constrained Fast Flexible Filling Design

Example of a Space-Filling Design for a Map Shape

Example of a Sphere-Packing Design

Statistical Details for Space Filling Designs

Minimum Potential

Maximum Entropy

Gaussian Process IMSE

Fast Flexible Filling Design Details

Want more information? Have questions? Get answers in the JMP User Community (community.jmp.com).