Lori A. Burns

Psi4 is really only safe to install with a defaults base, not a c-f base, on Linux. If you added `defaults` to the channel list or `anaconda::intel-openmp`, it'd likely solve,...

I'm surprised at the libxc problem, as that build number is at the top of the list here, https://anaconda.org/psi4/libxc/files, and you've got psi4 in the channels list. Perhaps you've got...

Any chance you're building CMAKE_BUILD_TYPE=Debug? That's been seen, and healed upon switch to Release.

Not sure what happened to the `C`, but the remainder of the valid string probably got mistaken for a comment and removed. While Psithon + preprocessing is very capable for...

I’ve hit that myself. Add to the cmake command -DENABLE_simint=OFF. I’ll have another go at the simint ppl to get them to release a tarball with simint am8 so we...

oh, that reminds me that I ought to increase the AM of the conda L2 for Linux. For better and worse, there's more AM knobs to turn for L2 than...

Bumping up just the eri3 for energy to 8 would likely be manageable, but I don't know about 10 or 12. I'm curious what EFV thinks at your query https://github.com/evaleev/libint/issues/250

I've built a 10 once for Susi. Compilation time isn't too bad when it's only energy ints -- Hessians are painful. The real obstacle is storage space for the conda...

I got a high AM version generated. It built all its files (some indiv files take 10h to compile) and then it failed on the link step. I've still got...

Don't turn off validate. That's giving a generic per atom ``` oxy = qcelemental.models.Molecule(geometry=[0,0,0], symbols=["O"], masses=[12.999], validate=False) #mtol=1.e-6) print(oxy.mass_numbers) # returns [16], wrong! ``` What is the use case here?...