Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
CAS PES crashing #758
I'm trying to calculate a potential energy surface for H2O double dissociation in a (4e,4o) active space using the input
Judging from output.dat, the run appears to crash in DiskJK for the third step.
Somewhat oddly, if I look at the occupation numbers for the second step, I see
whereas in the first step they're still sane
What's going on here?!
This might be due to a problem we noticed when computing PES with CASSCF. Guess orbitals are assumed to be orthogonal, which is not true if the geometry is changed. We have developed some python scripts to manually orthogonalize the guess orbitals with respect to the current overlap matrix. @lcyyork may be able to help.
Here is a script can orthogonalize orbitals between different geometries. However it does not consider frozen orbitals. They can be simply added following the comments inside the script.
To use it, you can put the following to the input:
I do not worry about efficiency usually, but you can probably make it faster using slicing when filling in the data to matrix.