Dakota Reference Manual  Version 6.4
Large-Scale Engineering Optimization and Uncertainty Analysis
 All Pages

Uses Latin Hypercube Sampling (LHS) to sample variables


Alias: none

Argument(s): none


The lhs keyword invokes Latin Hypercube Sampling as the means of drawing samples of uncertain variables according to their probability distributions. This is a stratified, space-filling approach that selects variable values from a set of equi-probable bins.

Default Behavior

Latin Hypercube Sampling is the default sampling mode in most contexts (exception: multilevel_sampling). To explicitly specify LHS in the Dakota input file, the lhs keyword must appear in conjunction with the sample_type keyword.

Usage Tips

Latin Hypercube Sampling is very robust and can be applied to any problem. It is fairly effective at estimating the mean of model responses and linear correlations with a reasonably small number of samples relative to the number of variables.


    sample_type lhs
    samples = 20