tkwant.onebody.kernels.PerturbationInterpolator

class tkwant.onebody.kernels.PerturbationInterpolator(syst, time_name, time_start, params=None, interpolation_type=Interpolation)

A class to extract and interpolate W(t).

This class can be used like tkwant.onebody.kernel.PerturbationExtractor, to calculate W(t) and the matrix-vector product W(t) * ket. Interpolation is done only for performance reasons, as W(t) from Kwant is slow to evaluate.

Parameters

  • syst (kwant.builder.FiniteSystem) – The system from which to extract the time-dependent perturbation.

  • time_name (str) – The name of the time argument. Only sites with Hamiltonian value functions with the argument name time_name are extracted into W(t).

  • time_start (float) – The initial time.

  • params (dict, optional) – Extra arguments to pass to the Hamiltonian of syst, excluding time.

  • interpolation_type (optional) – The interpolator class

Notes

By definition the perturbation is defined with respect to the initial time. The perturbation should be therefore zero for times t < time_start. This is not inforced by this routine however, but W(t) is always evaluated with the time argument it gets.

Methods

apply(self, double time, const double complex[:] ket, out=None)

Evaluate the matrix-vector product W(t) @ ket for the given time.

Parameters

  • time (int or float) – Time argument, must be equal or larger than time_start.

  • out (numpy.ndarray, optional) – Sparse matrix with to perform the operation in-place.

Returns

out – The product W(t) @ ket. If an out argument is given, the operation is performed in-place and the routine returns None.

Return type

numpy.ndarray or None

evaluate(self, double time, out=None)

Evaluate the W(t) matrix for the given time t.

Parameters

  • time (int or float) – Time argument, must be equal or larger than time_start.

  • out (coo_matrix, optional) – Sparse matrix with to perform the operation in-place.

Returns

out – The matrix W(t). The size of the W(t) matrix is size x size, containing nnz complex entries. If an out argument is given, the operation is performed in-place and the routine returns None.

Return type

coo_matrix or None

Attributes

nnz

The total number of time-dependent matrix elements.

Returns

nnz – The total number of time-dependent matrix elements, equal to len(W(t).data).

Return type

int

size

The size of the W(t) matrix.

Returns

size – The size of the W(t) matrix is size x size.

Return type

int