def process_diffraction(self, dz, n_jobs=MAX_NUMBER_OF_CPUS):
"""
:param dz: current step along evolutionary coordinate z
:param n_jobs: number of threads for parallelization
:return: None
"""
# calculation of current linear phase shift
current_lin_phase = 0.5j * dz / self._beam.medium.k_0
# forward parallel fast Fourier transform
fft_obj = fft2(self._beam._field, threads=n_jobs)
# linear phase increment
field_fft = self.__phase_increment(fft_obj(), self._beam.n_x,
self._beam.n_y,
array(self._beam.k_xs),
array(self._beam.k_ys),
current_lin_phase)
# backward parallel fast Fourier transform
ifft_obj = ifft2(field_fft, threads=n_jobs)
# field initialization with updated values
self._beam._field = ifft_obj()
def eval_ifft2(data, num_threads=12):
'''
Evaluate rfft2
'''
ift = ifft2(data,
threads=num_threads,
planner_effort="FFTW_ESTIMATE",
auto_align_input=True,
auto_contiguous=True)
# Return rfft2 result
return ift(data.copy(), np.zeros(ift.output_shape,
dtype=ift.output_dtype)).copy()
def process(self):
"""Noise and autocorr functions generation"""
# noise field generation
proto = self.__generate_protoarray(self._n_x, self._n_y,
self._variance_expected,
self._r_corr_in_points)
proto_shifted = fftshift(proto, axes=(0, 1))
proto_fft_obj = ifft2(proto_shifted)
proto_fft_normalized = self.__normalize_after_fft(proto_fft_obj())
self._noise_field = proto_fft_normalized
# initialization of arrays for real and imaginary parts
self._noise_field_real, self._noise_field_imag = \
self._initialize_noise_arrays(self._noise_field, self._n_x, self._n_y)
# autocorrelation functions calculation
self._calculate_autocorrelations()