Dakota Reference Manual
Version 6.4
LargeScale Engineering Optimization and Uncertainty Analysis

This page discusses discrete design, uncertain, and state variables (which have discrete
in their keyword name) as they have similar specifications. These include:
In addition, some aleatory uncertain variables, e.g., binomial_uncertain, are discrete integervalued random variables specified using parameters. These are described on their individual keyword pages.
Sets
Sets of integers, reals, and strings have similar specifications, though different value types.
The variables are specified using three keywords:
elements_per_variable
 a list of positive integers specifying how many set members each variable admitselements
 a list of the permissible integer values in ALL sets, concatenated together.elements_per_variable
, or an integer multiple of number of variableselements_per_variable
, and each partition is assigned to a variable.elements_per_variable
, and the partitions of elements
must match the strings from descriptors
For string variables, each string element value must be quoted and may contain alphanumeric, dash, underscore, and colon. White space, quote characters, and backslash/metacharacters are not permitted.
Examples are given on the pages:
Range
For discrete variables defined by range(s), the lower_bounds
and upper_bounds
restrict the permisible values. For design variables, this constrains the feasible design space and is frequently used to prevent nonphysical designs. This is a discrete interval variable that may take any integer value within bounds (e.g., [1, 4], allowing values of 1, 2, 3, or 4). For some variable types, each variable is can be defined by multiple ranges.
Examples are given on the pages: