MultiDOE: sampling technics on MATLAB/OCTAVE - v3.2

MultiDOE

This MATLAB/OCTAVE toolbox regroups many existing tools for generating sample points using many specific DOE techniques:

• Full factorial sampling
• Latin Hypercube Sampling
• Optimal Latin Hypercube sampling
• Improved Hypercube Sampling
• Random Sampling

Many techniques proposed in this toolbox are based on

• the R software. More precisely, the library lhs is used.
• the IHS matlab's softwares proposed by John Burkardt.
• the LHS proposed by Budiman Minasny.

NOTICE: sampling techniques based on the R software are only usable on a Unix OS.

Features

The MultiDOE toolbox are able to

• generate sampling
• sort the sample points using many techniques
• add points to an existing sampling (based on the LHS and IHS procedures of the R software)
• normalize/renormalize the sample points to be well defined in the specified design space
• call automatically the R software (on Unix/Linux/OSX systems only)
• compute scores (minimal distances, discrepancy...), see for instance the Franco's PhD thesis (2008)

Updates for v3.2:

• Fix issues in examples
• Fix issues in R DOE techniques
• Remove test function loading (due to new external class optiGTest)