- Feb 26, 2020
-
-
Christoph Groth authored
-
- Dec 24, 2019
-
-
Joseph Weston authored
First steps in ND vectorized systems Closes #326 See merge request !341
-
- Dec 20, 2019
-
-
Joseph Weston authored
Following a request by @cwg.
-
Joseph Weston authored
Previously infinite vectorized systems explicitly stored the sites of the previous unit cell, and one of the tests and 'hamiltonian' relied upon this fact. We therefore need to change these places now that infinite vectorized systems store only the sites of the FD.
-
Joseph Weston authored
Because 'sites' and 'id_by_site' are required for backwards compatibility they must know about the sites of the previous unit cell. In principle we could compute these as-needed, but explicitly storing them allows the implementations of _Site and _IdBySite to remain simple.
-
Joseph Weston authored
Previously we checked whether the subgraphs of a term used site arrays corresponding to sites >= cell_size, now we can directly check that the symmetry element of a term is not the identity (zero).
-
Joseph Weston authored
+ Store sites in the fundamental domain only + Compute and store the symmetry element with each term Here we continue to build the hopping terms from the graph, however when extending to handle >1D translational symmetry we will need to query the Builder directly. We could in principle already do that at this stage, and it would mean that less has to change later.
-
Joseph Weston authored
-
Joseph Weston authored
-
Joseph Weston authored
Even though we currently import 'Symmetry' and 'NoSymmetry' in 'builder' (to maintain backwards compatibility), we should use the canonical location within Kwant itself.
-
Joseph Weston authored
N-dimensional systems will need symmetry elements, so the symmetry definition needs to be in the system module.
-
Joseph Weston authored
Previously we required an array of group elements to act on a SiteArray (1 element per site), however often we want to act with a single group element on *all* the sites in a SiteArray. For TranslationalSymmetry we implement this using array broadcasting.
-
Joseph Weston authored
hamiltonian_submatrix is not meaningful for infinite systems, and (unlike un-vectorized systems) is not even used for constructing cell_hamiltonian and inter_cell_hopping in the vectorized case.
-
Joseph Weston authored
The docstring refers to site orderings, and explicit references to the graph; these are things that vectorized systems keep for backwards compatibility, but we wish to move away from. For this reason we move the docstring to the (unvectorized) infinite system class and add a more specific docstring to the vectorize infinite system class.
-
Joseph Weston authored
We now construct the site arrays to pass to 'val' in a single place (this is more readable).
-
Joseph Weston authored
We had forgotten to pass the 'vectorized' flag when constructing a Builder with a symmetry
-
Joseph Weston authored
This is a pure refactor.
-
- Dec 18, 2019
-
-
Joseph Weston authored
add plotly plotter See merge request kwant/kwant!294
-
Joseph Weston authored
-
Joseph Weston authored
-
- Dec 16, 2019
-
-
Kelvin Loh authored
-
- Dec 15, 2019
-
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
Raises a RuntimeError when incompatible inputs are provided to the plot function using the plotly backend.
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
Called once during setting of plotly backend.
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
Kelvin Loh authored
-
-
- Dec 13, 2019
-
-
Joseph Weston authored
make _normalize_matrix_blocks check number of orbitals Closes #346 See merge request kwant/kwant!349
-
- Dec 12, 2019
-
-
Joseph Weston authored
-