1Blurb::
2Evidence theory with evidence measures computed with global optimization methods
3Description::
4\c global_evidence allows the user to specify
5several global approaches for calculating the belief and plausibility functions:
6\li \c lhs - note: this takes the minimum and maximum of the samples as the bounds per
7"interval cell combination."
8\li \c ego - uses Efficient Global Optimization which is based on an adaptive Gaussian process surrogate.
9\li \c sbo - uses a Gaussian process surrogate (non-adaptive) within an optimization process.
10\li \c ea - uses an evolutionary algorithm.  This can be expensive as the ea will be run for each interval cell combination.
11
12
13Note that to calculate the plausibility and belief cumulative
14distribution functions, one has to look at all combinations of
15intervals for the uncertain variables. In terms of implementation, if
16one is using LHS sampling as outlined above, this method creates a
17large sample over the response surface, then examines each cell to
18determine the minimum and maximum sample values within each cell. To
19do this, one needs to set the number of samples relatively high: the
20default is 10,000 and we recommend at least that number. If the model
21you are running is a simulation that is computationally quite
22expensive, we recommend that you set up a surrogate model within the
23%Dakota input file so that \c global_evidence performs its sampling and
24calculations on the surrogate and not on the original model. If one
25uses optimization methods instead to find the minimum and maximum
26sample values within each cell, this can also be computationally
27expensive.
28
29<b> Additional Resources </b>
30
31See the topic page \ref topic-evidence_theory for important
32background information and usage notes.
33
34Refer to \ref topic-variable_support for information on supported
35variable types.
36
37
38Topics:: epistemic_uncertainty_quantification_methods, evidence_theory
39Examples::
40Theory::
41The basic idea is that one specifies an "evidence structure"
42on uncertain inputs and propagates that to obtain belief and
43plausibility functions on the response functions. The inputs are
44defined by sets of intervals and Basic Probability Assignments (BPAs).
45Evidence propagation is computationally expensive, since the minimum
46and maximum function value must be calculated for each "interval cell
47combination." These bounds are aggregated into belief and
48plausibility.
49Faq::
50See_Also::	method-global_interval_est, method-local_evidence, method-local_interval_est
51