We announce the release of the GPdoemd open-source python package
https://github.com/cog-imperial/GPdoemd by Simon Olofsson.
GPdoemd considers a setting with multiple rival mechanistic models and insufficient data to discriminate between them, i.e. to discard inaccurate models. GPdoemd performs model discrimination by suggesting the maximally informative next experimental design.
GPdoemd extends analytic methods from classical literature by introducing Gaussian process surrogates to approximate the mechanistic models.
For more information:
http://proceedings.mlr.press/v80/olofsson18a.html