Likelihood System

Overview

For performing Bayesian inference using MCMC techniques, a likelihood function needs to be defined for measuring the quality of model predictions with reference to the experiments. The likelihood functions can be defined in the [Likelihood] block.

Creating a Likelihood Function

A likelihood function is created by inheriting from LikelihoodFunctionBase and ReporterInterface and overriding the function method in the base class. See the Gaussian class for an example.

Example Input File Syntax

[Likelihood]
  [gaussian]
    type = Gaussian
    noise = 'noise_specified/noise_specified'
    file_name = 'exp1.csv'
    log_likelihood = true
  []
[]

Available Objects

Stochastic Tools App
ExtremeValueGeneralized extreme value likelihood function evaluating the model goodness against experiments.
GaussianGaussian likelihood function evaluating the model goodness against experiments.
TruncatedGaussianTruncatedGaussian likelihood function evaluating the model goodness against experiments.

Available Actions

Stochastic Tools App
AddLikelihoodActionAdds Likelihood objects.

(modules/stochastic_tools/test/tests/likelihoods/gaussian_derived/main.i)

[StochasticTools]
[]

[Distributions]
  [mu1]
    type = Normal
    mean = 0.0
    standard_deviation = 0.5
  []
[]

[Samplers]
  [sample]
    type = MonteCarlo
    distributions = 'mu1 mu1'
    num_rows = 3
    seed = 2547
  []
[]

[MultiApps]
  [sub]
    type = SamplerFullSolveMultiApp
    input_files = sub.i
    sampler = sample
  []
[]

[Transfers]
  [param]
    type = SamplerParameterTransfer
    to_multi_app = sub
    sampler = sample
    parameters = 'BCs/left/value BCs/right/value'
    # to_control = 'stochastic'
  []
  [reporter_transfer]
    type = SamplerReporterTransfer
    from_reporter = 'average/value'
    stochastic_reporter = 'constant'
    from_multi_app = sub
    sampler = sample
  []
[]

[Reporters]
  [constant]
    type = StochasticReporter
  []
  [noise_specified]
    type = ConstantReporter
    real_names = 'noise_specified'
    real_values = '0.2'
  []
  [likelihoodtest]
    type = TestLikelihood
    likelihoods = 'gaussian'
    model_pred = constant/reporter_transfer:average:value
    sampler = sample
  []
[]

[Likelihood]
  [gaussian]
    type = Gaussian
    noise = 'noise_specified/noise_specified'
    file_name = 'exp1.csv'
    log_likelihood=true
  []
[]

[Executioner]
  type = Steady
[]

[Outputs]
  file_base ='loglikelihood_scalar'
  [out]
    type = JSON
    execute_system_information_on = NONE
  []
[]

Overview
Creating a Likelihood Function
Example Input File Syntax
Available Objects
Available Actions