Simultaneous identification of models and parameters of scientific simulators

Paper

The paper is accepted at ICML 2024, for the time beeing here is a link to the preprint: https://arxiv.org/abs/2305.15174

Code

• The implementation for the (mixture of) grassmann distribution is a standalone package, which can be found here: https://github.com/mackelab/grassmann_binary_distribution
• The Gaussian MDN with marginalization during training time and its training is implemented in the SBI toolbox: https://github.com/sbi-dev/sbi/tree/cornelius_SBMI
• The gaussian prior can be found in this repository (sbmi/graph_prior)
• A short tutorial can be found here: https://github.com/sbi-dev/sbi/blob/cornelius_SBMI/tutorials/18_model_inference.ipynb