1. Informal TODO list¶

1.1. General¶

Never ending: write more documentation.
Link with ASE.
More strict unit checking in the parameter files. Change ‘e’ to ‘electron’.
Literature reference system, so users know what papers to cite when they run a simulation.
ATSELECT
- Add support for @N feature to ATSELECT.
- Add caching to the ATSELECT compiler.
System IO stuff:
- Read system from pdb, hdf5 file.
- Write system to xyz, pdf file.
- Optional link to openbabel (if installed, not a mandatory dependency).
- More compact chk file.
Add topological analysis to System class.
See TODO comments in code.

FF Generator
- Add support for scopes to the Generator classes, through sections in the parameter file, or by requiring a different file for every scope.
- Make an FF for methanol, and a methanol-water system to facilitate the testing.
- scope:ffatype and scope:number combinations in the parameter.
FF models
- Gaussian charge distributions.
- ACKS2.
- Dielectric background for fixed charge models.
Parallel force evaluation.
Convert from C to C++.
Smooth truncation of electrostatic interactions in reciprocal space to obtain a perfectly differentiable PES.
Cell lists.
NlogN scaling of the electrostatics.
Correctly treat the periodic boundary conditions in very skewed cells. The current implementation of the minimum image convention is, just like in most MD codes, rather naive. Check the answers on the scicomp faq.
Allow one to override the default names of the ForcePart* instances, e.g. to allow two different valence parts.

Restart functionality.
Constraints and restraints.
An optional select argument for the iterative algorithms, e.g. to easily freeze the remaining atoms during an MD or optimization. Similarly, one should also be able to construct a partial force field model.
RefTraj derivative of the Iterative class.
Parallel sampling methods.