A collection of Jupyter notebooks that illustrate how to use the dfi app for
data-free inference.

This example uses a special form of data-free inference to calibrate an example
quadratic model given summary statistics (a measurement value and an associated
measurement error). This example contains 5 different Jupyter notebooks that
each highlight a different aspect of the problem, as per the descriptions below.
The Jupyter notebooks call the dfi app in UQTk using the Python `subprocess`
module.

================================================================================
Requirements

Please make sure the environment variable "UQTK_INS" is set, e.g.,

export UQTK_INS=/path/to/uqtk/install/dir

================================================================================
Files

- utils.py: helper file to specify the location of the different programs, adapt
  as needed

- 1_quadratic_one_data_set.ipynb: a first simple example illustrating the use of
  the dfi app

- 2_quadratic_two_data_sets.ipynb: a second example illustrating how to use the
  dfi app with two different data sets

- 3_quadratic_two_data_sets_disagree: a third example illustrating how to use
  the dfi app with two disagreeing data sets, and showing the effect of using
  a weighted likelihood

- 4_quadratic_two_data_sets_different_nb_of_points.ipynb: a fourth example
  illustrating how to use the dfi app with two data sets with a different
  number of measurements each, and showing the effect of using a weighting
  factor for each data set proportional to the number of measurements in each
  data set

- 5_quadratic_with_optimal_beta.ipynb: a final example illustrating how to use
  the dfi app when calibrating the model against both the measurement values
  and associated error bars