Explore saddle points - Githubissues

globus-labs / faster-molecular-hessians

Reduce the amount of computation needed for Hessians using machine learning

MIT License

0 stars 1 forks source link

Open WardLT opened 7 months ago

WardLT commented 7 months ago

See if we get negative frequencies in the proper directions for saddle points.

WardLT commented 7 months ago

Similarly, we could see if the approximate Hessians are able to identify the same "weak" bonds from the low frequencies (100 1/cm or less)