Home Learning resources Workshops Research papers

In this section you will find some resources on the history matching with emulation (HME) framework, and how to implement the method using the hmer package.

Resources on HME

The resources below can be addressed independently of each other:

• An introduction to HME: a short tutorial which demonstrates the main concepts of HME using a simple one-dimensional example.

• Bayes Linear Emulation and History Matching: a presentation on the HME methodology by Ian Vernon.

Resources on hmer

The hmer package is available on CRAN, together with

• a reference manual;
• an introductory vignette on the main functionality of the package;
• a vignette showing how to use the package with one and two-dimensional models;
• a vignette on the calibration of stochastic and bimodal models;
• a history matching with emulation handbook, which presents some of the issues that arise most frequently in emulation and history matching and explains how to circumvent them.

In addition to those resources, we created the following tutorials, meant to be addressed in the proposed ordered:

1. Deterministic tutorial: a general introduction to hmer’s functionalities for the calibration of deterministic models. In Section 2 of this tutorial, we show how to use the full_wave function, which allows to perform a wave of the history matching with emulation process just with one command.

2. Deterministic practical tutorial: a practical, interactive introduction to hmer’s functionalities for the calibration of deterministic models. You can work through this practical tutorial by running the R script without solutions line by line. Note that the tutorial has tasks that require you to write your own code. Solutions to these tasks can be found both in the html file and in the R script with solutions. A shorter version of this tutorial can be found here.

3. Stochastic practical tutorial: a practical introduction to hmer’s functionalities for the calibration of models presenting stochasticity and/or bimodality. You can work through this practical tutorial by running the R script without solutions line by line. Note that the tutorial has tasks that require you to write your own code. Solutions to these tasks can be found both in the html file and in the R script with solutions.

Relationships between parameters (e.g. param_1 < param_2) can be taken into consideration in the history matching process by the customisation of the implausibility measure. A brief guide on how to customise implausibility can be found here.

Template to set up HME on your model

This repository contains an R-script Deterministic_Template_hmer_script that will guide you through setting up history matching with emulation on your deterministic model of interest. If instead you would like to calibrate a stochastic model, you can use the Stochastic_Template_hmer_script.