README

Generalised Likelihood Uncertainty Estimation (GLUE) Framework Joost Delsman, Deltares, 2011

Generic Python framework to conduct GLUE analyses NOTE: Package is still under construction and undocumented

Necessary:

1. model parameters, that:

• hold the statistical properties of the a priori parameter space
• hold the statistical properties of the a posteriori parameter space
• can return a random value based on the a priori parameter space
• can accept an evaluated parameter + behavioural / non-behavioural statement from the evaluator

2. a model, that:

• returns a result set based on a parameter set
• can be called successively to explore the parameter space
• resides outside GLUE
• (can even be outside python)
• function or wrapper must return a dict with parameters and their values, recognised by the evaluator

3. a model evaluator, that:

• evaluates the model result by a prescribed set of rules
• assigns behavioural / non-behavioural

4. a framework that:

• initializes a monte-carlo sequence
• for each run:
  • get parameter values
  • run the model
  • evaluate the model
  • store behavioural runs