About Input Parameters and Data Sampling
For most input parameters in an uncertainty quantification study, the parameter name, distribution type, specifications that define the distribution, lower and upper CDF or bounds of the distribution, and parameter units are needed to specify how the inputs are sampled. For screening, the Morris sampling method is used, and for other UQ study types, Latin hypercube sampling is used. Additional details about the input distribution and data sampling are provided in the theory section (see
Data sampling methods
).
In addition to generating input parameter data from analytic probability distributions, specified values and data column from result table can be used for sensitivity analysis, uncertainty propagation, and reliability analysis.
For verification and surrogate-based Monte Carlo analysis, the input parameters are usually the same as the input parameters used to generate quantities of interest through COMSOL model evaluations.