Class for the Dempster-Shafer Evidence Theory methods within DAKOTA/UQ. More...
Public Member Functions | |
NonDLocalEvidence (Model &model) | |
constructor | |
~NonDLocalEvidence () | |
destructor | |
Protected Member Functions | |
void | initialize () |
perform any required initialization | |
void | set_cell_bounds () |
set the optimization variable bounds for each cell | |
void | truncate_to_cell_bounds (RealVector &initial_pt) |
truncate initial_pt to respect current cell lower/upper bounds | |
void | post_process_cell_results (bool minimize) |
post-process a cell minimization/maximization result | |
void | post_process_response_fn_results () |
post-process the interval computed for a response function | |
void | post_process_final_results () |
perform final post-processing |
Class for the Dempster-Shafer Evidence Theory methods within DAKOTA/UQ.
The NonDEvidence class implements the propagation of epistemic uncertainty using Dempster-Shafer theory of evidence. In this approach, one assigns a set of basic probability assignments (BPA) to intervals defined for the uncertain variables. Input interval combinations are calculated, along with their BPA. Currently, the response function is evaluated at a set of sample points, then a response surface is constructed which is sampled extensively to find the minimum and maximum within each input interval cell, corresponding to the belief and plausibility within that cell, respectively. This data is then aggregated to calculate cumulative distribution functions for belief and plausibility.