Exercise 7: How to run VARS/G-VARS in the ‘off-line’/external mode

Objectives:

This notebook runs a sensitivity analysis when the model is not directly coupled with VARS/G-VARS – for example, when the model is written in a different programming language. There are three steps:

(1) first run parts of VARS/G-VARS to generate all sample points (model input sets) based on the given distributional properties of interest and print them in a .csv file,

(2) independent of VARS-TOOL, read all those input sets from the .csv file, execute the external model for each row to obtain the respective outputs, and add those outputs as new column to the .csv file.

(3) read in the updated .csv file with model outputs and run the rest of VARS/G-VARS to generate all the sensitivity indices.

Example research question: Consider a research question similar to that of Exercise 2 or 3, with the only difference that the model you have to deal with is external to the Python environment; for example, it is an executable file.