1Blurb:: 2Samples variables on full factorial grid of study points 3 4Description:: 5Dakota's multidimensional parameter study computes response data sets 6for an n-dimensional grid of points. Each continuous and discrete 7range variable is partitioned into equally spaced intervals between 8its upper and lower bounds, each discrete set variable is partitioned 9into equally spaced index intervals. The partition boundaries in 10n-dimensional space define a grid of points, and every point is 11evaluated. 12 13<b> Default Behavior </b> 14 15By default, the multidimensional parameter study operates over all 16types of variables. 17 18<b> Expected Outputs </b> 19 20A multidimensional parameter study produces a set of responses for 21each parameter set that is generated. 22 23<b> Usage Tips </b> 24 25Since the initial values from the variables specification will not be 26used, they need not be specified. 27 28Topics:: parameter_studies 29Examples:: 30This example is taken from the Users Manual and is a good comparison to the examples on 31\ref method-centered_parameter_study and \ref method-vector_parameter_study. 32\verbatim 33# tested on Dakota 6.0 on 140501 34environment 35 tabular_data 36 tabular_data_file = 'rosen_multidim.dat' 37 38method 39 multidim_parameter_study 40 partitions = 10 8 41 42model 43 single 44 45variables 46 continuous_design = 2 47 lower_bounds -2.0 -2.0 48 upper_bounds 2.0 2.0 49 descriptors 'x1' "x2" 50 51interface 52 analysis_driver = 'rosenbrock' 53 fork 54 55responses 56 response_functions = 1 57 no_gradients 58 no_hessians 59 60\endverbatim 61 62This example illustrates the full factorial combinations of parameter values created 63by the multidim_parameter_study. With 10 and 8 partitions, there are actually 6411 and 9 values for each variable. This means that \f$ 11 \times 9 = 99 \f$ 65function evaluations will be required. 66 67Theory:: 68Faq:: 69See_Also:: method-centered_parameter_study, method-list_parameter_study, method-vector_parameter_study 70