- Dec 21, 2016
-
-
Before we used the include_dir argument to setup(), but this caused problems: kwant/kwant#48
-
-
-
-
Commit a8b0463e turned out to be a bad idea: unlike on Debian, on many platforms the numpy headers are not available in the default include path.
-
1. Now can be compiled without MUMPS (previously exception was thrown) 2. Gives sensible error message for Git install, but without installed Cython.
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
- Dec 20, 2016
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
- Dec 19, 2016
-
-
Pablo Piskunow authored
-
- Dec 16, 2016
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
- Dec 15, 2016
-
-
Pablo Piskunow authored
Add changes to notebook "kpm_random_hamiltonian" to check that peaks in SpectralDensity are located at the eigenvalues of a `kwant.rmt`
-
Pablo Piskunow authored
-
- Dec 14, 2016
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
- Dec 13, 2016
-
-
Pablo Piskunow authored
Different instances created with the same numpy random seed, used to create slight different `moments_list`. The randomness came from `scipy.sparse.linalg.eigh`, where a residual vector `v0` is chosen randomly, unless deternimend. Now `v0` is generated with `vector_factory`, so `when numpy.random.seed` is set to the same value between different creations of an instance of `SpectralDensity`, the moments and therefore, the densities are the same.
-
- Dec 08, 2016
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
- Nov 17, 2016
-
-
Pablo Piskunow authored
add docstring of `vector_factory` and deleted unused functions `_create_rand_vect_list` and `_sampling_points`.
-
- Nov 16, 2016
-
-
Pablo Piskunow authored
-
- Nov 11, 2016
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
default value of epsilon changed to 0.05 for improving the approximation near the edges of the spectrum
-
- Nov 10, 2016
-
-
Pablo Piskunow authored
fix updating coeficients `coef_cheb` for `scipy.polynomial.chebyshev` after calling `increase_accuracy`
-
- Nov 09, 2016
-
-
Pablo Piskunow authored
updates: better docstrings, __init__ can take kwant.operators, both operators and hamiltonians are converted to scipy.sparse.linalg.LinearOperator to perform matvec multiplications, function __call__ can take scalar or array arguments and returns the explicit evaluation of the polinomial, method average now uses scipy.integrate.simps, method set_energy_spacing changed to increase_energy_resolution.
-
- Nov 01, 2016
-
-
Pablo Piskunow authored
-
- Oct 31, 2016
-
-
Pablo Piskunow authored
Add tested version for the kpm module with methods for increasing accuracy, and a test notebook to try it
-
- Oct 28, 2016
-
-
Pablo Piskunow authored
-
Pablo Piskunow authored
-
- Oct 26, 2016
-
-
Christoph Groth authored
-
Christoph Groth authored
-