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


Ensures that the samples of discrete variables with finite support are unique

Specification

Alias: none

Argument(s): none

Description

Traditional LHS can generate replicate samples when applied to discrete variables. This keyword enforces uniqueness, which is determined only over the set of discrete variables with finite support. This allows one to generate LHS for a mixed set of continuous and discrete variables whilst still enforcing that the set of discrete LHS components of all the samples are unique.

Default Behavior

Uniqueness of samples over discrete variables is not enforced.

Usage Tips

Uniqueness can be useful when applying discrete LHS to simulations without noise.

Examples

method,
        sampling
          samples = 12
          seed = 123456 
          sample_type lhs backfill

variables,
        active all
        uniform_uncertain = 1
          lower_bounds =  0.
          upper_bounds =  1.
          descriptors  = 'continuous-uniform'

        discrete_uncertain_set
          integer = 1
           elements_per_variable = 4
            elements  1 3 5 7
            descriptors =    'design-set-int'
          real = 1
            initial_point = 0.50
            set_values  =  0.25 0.50 0.75 1.00
            descriptors =  'design-set-real'

interface,
        direct analysis_driver = 'text_book'

responses,
        response_functions = 3
        no_gradients
        no_hessians

See Also

These keywords may also be of interest: