This release introduces many API changes and a much more stream-lined
interface for interacting with sisl.
You are heavily encouraged to update your distribution.
Here is a compressed list of changes:
- sdata is now an input AND output dependent command.
It first reads the input and output files, in a first run, then
it determines the options for the given set of files.
Secondly, the sdata command uses "position dependent" options.
This means that changing the order of options may change the output.
- tbtncSile
* Correct vector currents (for xsf files)
* bug-fix for Gamma-only calculations
* returned DOS is now correctly in 1/eV (older versions returned 1/Ry)
* fixed sdata atomic[orbital] ranges such that, e.g. `--atom [1-2][3-5]`
(for atom 1 and 2 and only orbitals 3, 4 and 5 on those atoms.)
* DOS queries now has an extra argument (E) which returns only for the
given energy.
* When storing tables in sdata this now adds information regarding
each column at the top (instead of at the bottom).
Furthermore, the information is more descriptive
- Changed all `square` named arguments to `orthogonal`
- Added nsc field to xyz files (to retain number of supercells)
- Added `move` function for geometry (same as translate)
- Added `prepend` function, equivalent to `append`, but adding the
atoms in the beginning instead of the end
- Fixed many bugs related to the use of Python-ranges (as opposed to numpy ranges)
- SparseCSR now enables operations:
a = SparseCSR(...)
a = a * 2 + 2
is now viable. This enables easy scaling, translation etc. using the
sparse matrix format (very handy for magnetic fields).
- Enabled `del` for SparseCSR, i.e. `del SparseCSR(..)[0, 1]` will
remove the element, completely.
- Enabled reading of the TSHS file from SIESTA 4.1, now we may easily interact
with SIESTA.
- Moved version.py to info.py
- Moved scripts to `entry_points`, this makes scripts intrinsic in the module
and one may import and use the commands as their command-line equivalents.
- Hamiltonian.construct now takes a single argument which is the function
for the inner loop.
The old behaviour may be achieved by doing either:
>>> func = Hamiltonian.create_construct(R, param)
>>> Hamiltonian.construct(func)
or
>>> Hamiltonian.construct((R, param))
- The atoms contained in the Geometry are now not duplicated in case of many
similar Atom objects. This should reduce overhead and increase throughput.
However, the efficiency is not optimal yet.
- Added many more tests, thus further stabilizing sisl
I would really like help with creating more tests!
Please help if you can!
<!--
Local Variables:
mode: text
comment-column: 0
tab-width: 2
End:
-->