- Added last set of unit tests.
- Added an option to print the steps taken to reduce the cell.
- Updated the floating point tolerance to better reflect the original paper.
- Changed the algorithm so that it now calculates the niggli G vector from the updated (partially reduced) lattice after each step to improve stability.
- Changed steps 3 and 4 to work of integer values rather than the actual floats to improve stability.
- Re-organized some of the condition checks to improve stability.
- Added a printing of the original lattice vectors to the debug raises.
- Changed _find_C3 and _find_C4 to static methods that take integers rather than floats.
- Fixed issue reported in issue #5.
- Updated the floating point tolerance to depend on cell volume.
- Increased the number of allowed iterations to allow for larger cells.
- Improved the tolerance on floating point comparisons in the Niggli condition checker.
- Added Niggli condition checks to the subroutine. The class now issues a RuntimeError if the Niggli conditions aren't met.
- Changed some of the comparisons to be np.allclose.
-Fixed the niggli reduction so that the values of the angles are used to contsruct the transformation matrix before they are reassigned. -Fixed condition 7 which was changing the value of C instead of B.
-Fixed niggli so that the columns and not the rows are used to colculate the constants. -Shortened the calls in the inequalities slightly.
-Fixed the call to swap in niggli.py.
-Created main niggli class and the subroutines needed to perform the niggli reduction.
-Created repo.