You must login before you can run this tool.
SMOOT (Stochastic Multi-Objective Optimization Tool) is an application developed in Rappture toolkit. The application runs optimization simulations using Gaussian process regression surrogate model and expected improvement over dominated hypervolume information acquisition function. The optimization algorithm has been written in Python.
The program consists of two main phases: initial data upload and sequential addition of observations. The uploaded data must be in CSV format. They consist of the following: observed data, lower and upper bounds of the inputs, and design points for the model. The additional data evaluated at design points suggested by the program can be uploaded after each simulation has been completed. The results of each iteration (model design) are displayed as a Pareto front and can also be downloaded as a CSV file. The program is designed to run iteratively for as long as the user is running experiments within available budget.
NanoHUB Team, Purdue University.
Special thanks to Professor Jitesh Panchal for providing test examples.
SURF Program 2016
Emmerich, M., & Klinkenberg, J. W. (2008). The computation of the expected improvement in dominated hypervolume of Pareto front approximations.Rapport technique, Leiden University.
Pandita, P., Bilionis, I., & Panchal, J. (2016). Extending Expected Improvement for High-dimensional Stochastic Optimization of Expensive Black-Box Functions. arXiv preprint arXiv:1604.01147.
Cite this work
Researchers should cite this work as follows:
(2016), "Stochastic Multi-Objective Optimization Tool", Juan Sebastian Martinez, Martin Figura, Ilias Bilionis, Piyush Pandita, Rohit Kaushal Tripathy, https://nanohub.org/resources/smoot.