- Compare to JERI.jl benchmarks
- Would be nice to be within a factor of 10 by release 0.1.0
- Determine whether the normalization constants are the same for all m-values corresponding to an L-value. They're NOT.
- Write out a standard integral record using this code
- Time for ethane
- Figure out right place to add normalization consts
- Working test chrr
- Host a webpage for MolecularIntegrals.jl
- Put badge-link to docs in README.md file
- Register julia package wherever I'm supposed to do this
- Post to discourse server
- Interpolate
- Gil, Johnson, Pople method
- Loop over basis sets, then use iindex
- Don't need to include every one in BSE
- Check against files in Data.jl
- Can replace SP with S,P if SP shells aren't working yet.
- Is this in a /data directory? How do I support it?
- vrr_test(ashell,cshell) code:
- Test all elements of vrr against ERI.jl code
- Random A,C, aex, cex? Or maybe fallback if A,C,aex,cex not specified
- Can use this as an excuse to write functions to generate random PGBFs in some distance/exp/powers
- hrr_test(ashell,bshell,cshell,dshell)
- Test all elements of hrr and hrr_dict against ERI.jl code
- Random A,B,C,D??? Or maybe fallback if A,B,C,D not specified
- Add comments to ERI code to the effect that it is slow code that is very likely to be correct that is kept around for reference and testing.
- Merge Low level OneInts tests with OneInts
- ERI tests should be against standard (large R) or pyquante values.
- vrr/hrr tests should be against ERI versions
- May have to redo indexing of shell_indices to make this work
- SP shells could be one of the things that is easier to do in the generated code (below)
- Look at what derivative code julia routines support.
- GAMESS team, psi4 team, etc.
- nocc, nclosed, nopen, nup, ndown, stoich, center!