Science Advances

Supplementary Materials

This PDF file includes:

  • Fig. S1. Typical learning curves for training the AIMNet model.
  • Table S1. Descriptive statistics for ANI-1x dataset, the extension with molecules containing S, F, or Cl atoms, and resulted combined extended dataset ANI-1x-SFCl.
  • Table S2. Descriptive statistics for COMP6 dataset, extension containing molecules with S, F, or Cl atoms, and resulted combined extended dataset COMP6-SFCl.
  • Table S3. Self-energies for atoms used to compute normalized atomization energies.
  • Table S4. Neural network architectures for each AIMNet blocks.
  • Table S5. Performance of ANI-1x and AIMNet on COMP6-SFCl dataset.

Download PDF

Files in this Data Supplement: