Susi Lehtola

I bet this is because of the calls to `exit()` in libpsio, combined with the lack of flushing of the error messages to the output. I have replaced the calls...

I've actually been thinking about Fermi smearing recently, as well, but didn't come to a conclusion which code to put it into; Psi4 was and is one of the options....

> @susilehtola, I've switched the build-from-src libxc target to v6.1 somewhere, and I'll make sure that makes it into the psi v1.8 release. c-f packaging is already at v6.1. Is...

Actually, it sounds like there might be an issue in the code, although I am not sure if I fully understand stability analysis. > The instability analysis gave us the...

Right, I guess I was a bit tired. The Taylor expansion of the energy in the orbital rotation $$ E(\boldsymbol{\theta})=E_{0}+\boldsymbol{\theta}\cdot\boldsymbol{g}+\frac{1}{2}\boldsymbol{\theta}\cdot\boldsymbol{H}\cdot\boldsymbol{\theta} + \mathcal{O}(\boldsymbol{\theta}^3) $$ where $$ \boldsymbol{g} = \nabla_{\boldsymbol{\theta}}E\Big|_{\boldsymbol{\theta}=\boldsymbol{0}} $$ is...

It also might be an issue with the functional. In my experience with finite element calculations, the B97 functionals are not very well-behaved numerically. I would not be surprised if...

Note that the SAD guess in Psi4 is not the best one possible, as it is not symmetry aware. Such a SAD guess is available in PySCF, and it is...

If you call Psi4 with `maxiter=0` it builds the Fock matrix from the SAD density and diagonalizes it to give you orbitals and then builds a new density from these...

> would you have any thoughts/concerns on doing the SAD guessing in `PySCF` and then doing the row/col re-ordering? What do you need the guesses for? PySCF can also be...

> I'm trying to build a ML model that can predict the converged density matrix. To verify whether the model is of any value, I wish to plug the predictions...