dace

Design and Analysis of Computer Experiments

Topics

Specification

  • Alias: None

  • Arguments: None

  • JSON argument key: None

Child Keywords:

Required/Optional

Description of Group

JSON Group Key

Dakota Keyword

Dakota Keyword Description

Optional

model_pointer

Identifier for model block to be used by a method

Optional

variance_based_decomp

Activates global sensitivity analysis based on decomposition of response variance into contributions from variables

Optional

seed

Seed of the random number generator

Optional

fixed_seed

Reuses the same seed value for multiple random sampling sets

Optional

id_method

Name the method block; helpful when there are multiple

Optional

output

Control how much method information is written to the screen and output file

Optional

final_solutions

Number of designs returned as the best solutions

Required (Choose One)

Sub Method

sub_method

grid

Grid Sampling

random

Uses purely random Monte Carlo sampling to sample variables

oas

Orthogonal Array Sampling

lhs

Uses Latin Hypercube Sampling (LHS) to sample variables

oa_lhs

Orthogonal Array Latin Hypercube Sampling

box_behnken

Box-Behnken Design

central_composite

Central Composite Design

Optional

samples

Number of samples for sampling-based methods

Optional

main_effects

ANOVA

Optional

quality_metrics

Calculate metrics to assess the quality of quasi-Monte Carlo samples

Optional

symbols

Number of replications in the sample set

Description

The Distributed Design and Analysis of Computer Experiments (DDACE) library provides the following DACE techniques:

  1. grid sampling ( grid)

  2. pure random sampling ( random)

  3. orthogonal array sampling ( oas)

  4. latin hypercube sampling ( lhs)

  5. orthogonal array latin hypercube sampling ( oa_lhs)

  6. Box-Behnken ( box_behnken)

  7. central composite design ( central_composite)

These methods all generate point sets that may be used to drive a set of computer experiments. Note that all of the DACE methods generated randomized designs, except for Box-Behnken and Central composite which are classical designs. That is, the grid sampling will generate a randomized grid, not what one typically thinks of as a grid of uniformly spaced points over a rectangular grid. Similar, the orthogonal array is a randomized version of an orthogonal array: it does not generate discrete, fixed levels.

In addition to the selection of the method, there are keywords that affect the method outputs:

  1. main_effects

  2. quality_metrics

  3. variance_based_decomp

And keywords that affect the sampling:

  1. fixed_seed

  2. symbols

  3. samples

  4. seed