Example: Analytical function

This example shows a surrogate-assisted Bayesian calibration of a time dependent analytical function. A detailed explanation of this example is provided in the TUTORIAL.

Model and Data

Pylink model

Property	Setting
Model type	Function
Number of input parameters	Set to either 2 or 10
Number of output parameters	1
Time- or space- dependency	Time-dependent output, 10 steps in time
MC reference	Yes

Priors

Parameter	Distribution
0-10	Uniform in (-5,5)

Discrepancy

Property	Setting
Distribution type	Gaussian
Characteristic value	variance: data^2

Surrogate

MetaModel settings

Property	Setting
surrogate-type	aPCE
associated model	‘analytical function’
degree choices	max degree 12, q-norm truncation 0.85 or 0.5 depending on number of inputs
regression	FastARD with bootstrapping

Training choices

Property	Setting
Static sampling method	latin-hypercube
Number of static samples	3*#inputparams
Number of total samples	150
Number of samples per AL iteration	1
AL tradeoff scheme	None
AL exploration method	random, n_candidates=1000, n_cand_groups=4
AL exploitation method	Bayesian Active Design (BAL) with DKL