CosmoLayer is a package implementing differentiable COSMO-type activity coefficient calculation layers for neural network models.
CosmoLayer leverages automatic differentiation and GPU acceleration to enable efficient computation and gradient-based optimization of COSMO model parameters.
CosmoLayer is available as a conda package on the mdtools channel. To install it, run:
conda install -c conda-forge -c mdtools cosmolayerOr:
mamba install -c mdtools cosmolayerTo use CosmoLayer in your own Python script or Jupyter notebook, simply import it as follows:
import cosmolayerDocumentation for the latest CosmoLayer version is available at Github Pages.
Copyright (c) 2026 Charlles Abreu