Greedy nonlinear autoregression for multifidelity modeling. (Multi-fidelity Gaussian process regression with active learning based on a greedy approach)
MATLAB code for the paper, Greedy nonlinear autoregression for multifidelity computer models at different scales (https://www.sciencedirect.com/science/article/pii/S2666546820300124).
This repository provides a modification of the nonlinear autoregression method for a model sequential construction based on greedy approaches. It also provides the stochastic collocation method for multifidelity modeling and Gaussian process with sequential learning for comparison.
Please refer Demo_synthe_01.m and Demo_synthe_02.m for the usage of the code. Please add all subfolder to your path to start using the code.
The following figures (generated by running Demo_synthe_02.m) show the performance of greedyNAR and stochastic collocation for synthetic three-fidelity data.
Normal Gaussian process with only high-fidelity observations:
Stochastic collocation without low-fidelity observations:
Stochastic collocation with low-fidelity observations:
GreedyNAR without low-fidelity observations:
GreedyNAR with low-fidelity observations:
