PAOFLOW.hamiltonian.do_double_grid

Functions

do_double_grid(data_controller)

Interpolate the Hamiltonian onto a finer k-grid via zero-padding in real space.

Module Contents

PAOFLOW.hamiltonian.do_double_grid.do_double_grid(data_controller)

Interpolate the Hamiltonian onto a finer k-grid via zero-padding in real space.

Parameters:

data_controller (DataController) – Object providing data_arrays and data_attributes. Required array: HRs (shape (nawf, nawf, nk1, nk2, nk3, nspin)). Required attributes: nawf, nk1, nk2, nk3, nspin, nfft1, nfft2, nfft3, npool.

Returns:

Adds or updates the following entries in data_controller.data_arrays and data_controller.data_attributes:

• Hksp : np.ndarray, shape (snawf, nfft1, nfft2, nfft3, nspin), complex — the k-space Hamiltonian on the extended, interpolated grid, distributed over MPI pools.

Updates attributes: nk1 = nfft1, nk2 = nfft2, nk3 = nfft3, nkpnts = nfft1 * nfft2 * nfft3.

Return type:

None

Notes

Fourier interpolation is achieved by inserting zeros in the frequency domain before the inverse transform. Specifically, the real-space Hamiltonian \(H(\mathbf{R})\) (HRs) is zero-padded from (nk1, nk2, nk3) to (nfft1, nfft2, nfft3) using zero_pad(), which correctly preserves Hermitian symmetry (\(H(-\mathbf{R}) = H^\dagger(\mathbf{R})\)) for both even and odd grid sizes. A forward FFT then yields the interpolated \(H(\mathbf{k})\) on the dense grid.

The reshaped HRs array (nawf**2 rows) is scattered across MPI pools before processing; only rank 0 performs the initial reshape.