Skip to content

Latest commit

 

History

History
69 lines (54 loc) · 2.44 KB

TODO.md

File metadata and controls

69 lines (54 loc) · 2.44 KB

MolecularIntegrals.jl tasks and roadmap.

Timing and benchmarks

  • Compare to JERI.jl benchmarks
  • Would be nice to be within a factor of 10 by release 0.1.0

Contracted routines

  • 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

More documentation

  • Host a webpage for MolecularIntegrals.jl
  • Put badge-link to docs in README.md file

~~~~~~~~~~~~~ Release version 0.1.0 ~~~~~~~~~~~~~~~~~~~~

  • Register julia package wherever I'm supposed to do this
  • Post to discourse server

Post help message to Julia discourse channel

Additional improvements to Boys

  • Interpolate
  • Gil, Johnson, Pople method

Speed eri_fetcher

  • Loop over basis sets, then use iindex

Include basis sets in g94 format

  • 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?

Write fuzz-like test functions in runtests.jl

  • 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

Bump version to 0.2

Get SP shells working

  • 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)

Someday changes

Derivatives??

  • Look at what derivative code julia routines support.

Reach out to julia package maintainers to use MolecularIntegrals?

  • GAMESS team, psi4 team, etc.

Implement other molecule methods:

  • nocc, nclosed, nopen, nup, ndown, stoich, center!