bayes_calibration
Bayesian calibration
Topics
bayesian_calibration, package_queso
Specification
Alias: nond_bayes_calibration
Arguments: None
Child Keywords:
Required/Optional 
Description of Group 
Dakota Keyword 
Dakota Keyword Description 

Required (Choose One) 
Bayesian Calibration Method 
Markov Chain Monte Carlo algorithms from the QUESO package 

(Experimental) Gaussian Process Models for Simulation Analysis (GPMSA) Bayesian calibration 

(Experimental Method) NonMCMC Bayesian inference using interval analysis 

DREAM (DiffeRential Evolution Adaptive Metropolis) 

Markov Chain Monte Carlo algorithms from the MUQ package 

Optional 
(Experimental) Adaptively select experimental designs for iterative Bayesian updating 

Optional 
Calibrate hyperparameter multipliers on the observation error covariance 

Optional 
Manually specify the burn in period for the MCMC chain. 

Optional 
Compute informationtheoretic metrics on posterior parameter distribution 

Optional 
Compute diagnostic metrics for Markov chain 

Optional 
Calculate model evidence (marginal likelihood of model) when using Bayesian methods 

Optional 
(Experimental) Postcalibration calculation of model discrepancy correction 

Optional 
Specify a subsampling of the MCMC chain 

Optional 
Specify probability levels at which to compute credible and prediction intervals 

Optional 
Stopping criterion based on objective function or statistics convergence 

Optional 
Number of iterations allowed for optimizers and adaptive UQ methods 

Optional 
Identifier for model block to be used by a method 

Optional 
Turn on scaling for variables, responses, and constraints 
Description
Bayesian calibration methods take prior information on parameter values (in the form of prior distributions) and observational data (e.g. from experiments) and infer posterior distributions on the parameter values. When the computational simulation is then executed with samples from the posterior parameter distributions, the results that are produced are consistent with (“agree with”) the experimental data. Calibrating parameters from a computational simulation model requires a likelihood function that specifies the likelihood of observing a particular observation given the model and its associated parameterization; Dakota assumes a Gaussian likelihood function. The algorithms that produce the posterior distributions on model parameters are most commonly Monte Carlo Markov Chain (MCMC) sampling algorithms. MCMC methods require many samples, often tens of thousands, so in the case of model calibration, often emulators of the computational simulation are used. For more details on the algorithms underlying the methods, see the Dakota User’s manual.
Dakota has four classes of Bayesian calibration methods: QUESO/DRAM, GPMSA, DREAM, and WASABI.
The QUESO methods use components from the QUESO library (Quantification of Uncertainty for Estimation, Simulation, and Optimization) developed at The University of Texas at Austin. Dakota uses its DRAM (Delayed Rejected Adaptive Metropolis) algorithm, and variants, for the MCMC sampling.
GPMSA (Gaussian Process Models for Simulation Analysis) is an approach developed at Los Alamos National Laboratory and Dakota currently uses the QUESO implementation. It constructs Gaussian process models to emulate the expensive computational simulation as well as model discrepancy. GPMSA also has extensive features for calibration, such as the capability to include a model discrepancy term and the capability to model functional data such as time series data. This is an experimental capability and not all features are available in Dakota yet.
DREAM (DiffeRential Evolution Adaptive Metropolis) is a method that runs multiple different chains simultaneously for global exploration, and automatically tunes the proposal covariance during the process by a selfadaptive randomized subspace sampling. [VtBD+09].
WASABI: NonMCMC Bayesian inference via interval analysis
Usage Tips
The Bayesian capabilities are under active development. At this stage, the QUESO methods in Dakota are the most advanced and robust, followed by DREAM, followed by GPMSA and WASABI which are not yet ready for production use.
The prior distribution is characterized by the properties of the
uncertain variables. Correlated priors are only supported for
unbounded normal, untruncated lognormal, uniform, exponential, gumbel,
frechet, and weibull distributions and require specification of
standardized_space
, for example, for QUESO
standardized_space
Note that as of Dakota 6.2, the field responses and associated field data may be used with QUESO and DREAM. That is, the user can specify field simulation data and field experiment data, and Dakota will interpolate and provide the proper residuals to the Bayesian calibration.