For desired every capability add a bullet point, with sub-bullet points with references and where to steal code
-
Python Bindings
- All modules easily portable as standalone, for easy python bindings
-
One-executable
- For normal use and non-experts, use one executable with a input file
-
Improved parametrization for biomolecules and drugs
- Step one is reparametrization of available SQM methods using existing dataset of MP2 heat-of-formation for GDB7, GDB8 or GDB9 datasets. We aim to re-parametrize the following methods:
- MNDO - the baseline method
- MNDO/d - MNDO with d-orbitals
- MNDOC - MNDO with BWEN perturbation theory correlation term
- AM1 - MNDO with additional core-repulsion terms
- PM6 - Same as AM1, but pairwise core-repulsion terms http://openmopac.net/Manual/external.html
- OMx - Orthogonalization corrected method
- Modified OMx - Disable orthogonalization correction and compare effect of gaussian/ECP integrals with MNDO multipole integrals.
- Analytical implementation of overlap integrals
- Analytical implementation of MNDO two-electron integrals
- MiniAMBER code
- Fock-matrix assembly for NDDO, CNDO, MNDO, AM1/PM3/PM6/PM7, etc
- Should be easy to write (just to clarify, Andersx wrote that. Famous last words)
- RHF and UHF type wave functions
- Exact diagonalization (LAPACK) and pseudo-diagonalization
- Exact diagonalizer from MKL or LibElemental, pseudo diagonalizer from AMBER?
- Stable SCF convergers (Quadratic, DIIS, etc)
- ??
- Analytical first derivative (i.e. gradient)
- ??
- (Analytical Hessian?)
- ??
- Slater-Koster file parser
- AMBER
- Integral rotations
- AMBER
- Fock-matrix assembly
- Should be easy to write
- Spin-polarized/open-shell DFTB
- Mixers (Anderson/DIIS/Broyden)
- ??
- Analytical first derivative (i.e. gradient)
- ??
- (Analytical Hessian?)
- ??
- Dispersion D3
- Stefan Grimme
- H+
- Jimmy's kode
- Find out which optimizer is used in Gaussian, Gamess, etc
- ??
- TS optimizer:
- GSM
- Optimize to transition state (e.g. QST2/QST3)?
- Use Python optimizers (with fortran bindings) to test different optimizers
- PCM (which version?)
- ??
- COSMO/COSMO-RS
- ??
- Which scheme?
- FMO?
- Linear-scaling sparse matrix?
- Review of semiempirical approximations
- Construction of dataset from CCSD(T)
- Identify relevant molecules from ZINC database or GDB* database
- Heats-of-formation dataset
- Non-covalent interactions dataset
- Reaction barriers
- Rotational barriers