Evaluating Gaussian Process Metamodels and Sequential Designs for Noisy Level Set Estimation - Supplementary Material

This is the supplementary material for article "Evaluating Gaussian Process Metamodels and Sequential Designs for Noisy Level Set Estimation", including the source code and the simulation dataset for synthetic experiments (2D Modified Branin-Hoo function and 2D Michalewicz function) and case study (2D put option and 3D call option). 

The code is originally branched from the open source MATLAB library GPstuff  by Jarno Vanhatalo, Jaakko Riihimäki, Jouni Hartikainen, Pasi Jylänki, Ville Tolvanen, and Aki Vehtari (2013). GPstuff: Bayesian Modeling with Gaussian Processes. Journal of Machine Learning Research, 14(Apr):1175-1179. (Available at http://jmlr.csail.mit.edu/papers/v14/vanhatalo13a.html). Implementations of adaptive design with Gaussian Process and its application in Bermudan option is mainly included in folders "adaptive_design" and "bermudan_option_oracle", with other minor changes in the GPstuff source code according to the experiment setup in the article. Two demo files "adaptive_design/Demo.m" and "bermudan_option_oracle/bermudan_option" are included and can be used to generate some sample dataset for each case.