8. Known Bugs/Limitations
1. The unit cell in the final output image is not exactly of the correct dimension, and non 90-degree axes are not corrected for. Within semper, this is easy to do later. Note: for p3 and p6 groups a rectangular unit cell is output which is equivalent to two full unit cells with the origin in the middle of the left-hand side. This is done deliberately to reduce astigmatism.
2. For cm and the p3/p6 groups origin definition by restricting the range is not done automaticaly. No enantiomorph selection is included, since the way the phases are permuted will tend to select one enantiomorph in most cases.
3. The program peaks2D could do with some more work.
4. The shell codes all require UNIX and some work is required to convert them all the fortran (c ?) for PC's or MAC's.
5. Peaks2D should do a more rigorous refinement, ideally inclusive of Debye-Waller factors suitable for a final publication.
6. The current method of calculating the atomic structure factors should be tidied up.
7. Channelling values and/or a multislice refinement could be added - not simple.
8. If the structure is almost symmetric (e.g. Si (111) 7x7 which is really p3m1 but approximately p6mm) it is better to solve first in the higher symmetry group. In the lower symmetry the algorithm tends to give too asymmetric results at least in two-dimensions with surface data. Additional information about the symmetry (e.g. from rel-rods) is very important.