def default_exp_f_with_ss_and_rs(self, A, h, B):
" Use self-steepening and Raman-scattering within exponential term. " ""
term_spm = np.abs(A)**2
term_ss = (self.ss_factor / B) * ifft(self.omega * fft(term_spm * A))
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return np.exp(h * self.factor * (term_spm + term_ss + term_rs)) * B
def default_f_with_ss_and_rs(self, A, z):
""" Use self-steepening and Raman-scattering terms. """
term_spm = np.abs(A)**2
term_ss = self.ss_factor * ifft(self.omega * fft(term_spm * A))
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return self.factor * (term_spm + term_rs) * A + term_ss
def default_f_all(self, A, z):
""" Set all nonlinear terms. """
term_spm = np.abs(A)**2
convolution = ifft(self.h_R * fft(term_spm))
p = fft(A * (1.0 - self.f_R) * term_spm + self.f_R * A * convolution)
return ifft(self.factor * (1.0 + self.omega * self.ss_factor) * p)
def default_f_with_ss_and_rs(self, A, z):
""" Use self-steepening and Raman-scattering terms. """
term_spm = np.abs(A) ** 2
term_ss = self.ss_factor * ifft(self.omega * fft(term_spm * A))
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return self.factor * (term_spm + term_rs) * A + term_ss
def default_exp_f_with_ss_and_rs(self, A, h, B):
" Use self-steepening and Raman-scattering within exponential term. """
term_spm = np.abs(A) ** 2
term_ss = (self.ss_factor / B) * ifft(self.omega * fft(term_spm * A))
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return np.exp(h * self.factor * (term_spm + term_ss + term_rs)) * B
def default_f_all(self, A, z):
""" Set all nonlinear terms. """
term_spm = np.abs(A) ** 2
convolution = ifft(self.h_R * fft(term_spm))
p = fft(A * (1.0 - self.f_R) * term_spm + self.f_R * A * convolution)
return ifft(self.factor * (1.0 + self.omega * self.ss_factor) * p)
def __call__(self, domain, field):
"""
:param object domain: A domain
:param object field: Current field
:return: Field after modification by Gaussian filter
:rtype: Object
"""
# Convert field to spectral domain:
self.field = fft(field)
delta_nu = domain.nu - domain.centre_nu - self.offset_nu
factor = power(delta_nu / self.width_nu, (2 * self.m))
# Frequency values are in order, inverse shift to put in fft order:
self.shape = exp(-0.5 * ifftshift(factor))
if domain.channels > 1:
# Filter is applied only to one channel:
self.field[self.channel] *= self.shape
else:
self.field *= self.shape
# convert field back to temporal domain:
if self.type is "reflected":
return ifft(self.field)
else:
return field - ifft(self.field)
def default_exp_f_all(self, A, h, B):
""" Set all terms within an exponential factor. """
term_spm = np.abs(A) ** 2
convolution = ifft(self.h_R * fft(term_spm))
p = fft((1.0 - self.f_R) * term_spm + self.f_R * convolution)
return np.exp(h * ifft(self.factor *
(1.0 + self.omega * self.ss_factor) * p)) * B
def wdm_exp_f_cached(self, As, h):
if self.cached_factor is None:
print "Caching linear factor"
self.cached_factor = [np.exp(h * self.factor[0]),
np.exp(h * self.factor[1])]
return np.asarray([ifft(self.cached_factor[0] * fft(As[0])),
ifft(self.cached_factor[1] * fft(As[1]))])
def wdm_exp_f(self, As, h):
hf0 = h * self.factor[0]
hf1 = h * self.factor[1]
if self.phase_lim:
hf0 = self._limit_imag_part(hf0)
hf1 = self._limit_imag_part(hf1)
return np.asarray([ifft(np.exp(hf0) * fft(As[0])),
ifft(np.exp(hf1) * fft(As[1]))])
def default_exp_f_all(self, A, h, B):
""" Set all terms within an exponential factor. """
term_spm = np.abs(A)**2
convolution = ifft(self.h_R * fft(term_spm))
p = fft((1.0 - self.f_R) * term_spm + self.f_R * convolution)
return np.exp(h * ifft(self.factor *
(1.0 + self.omega * self.ss_factor) * p)) * B
def __call__(self, domain, field):
# Convert field to spectral domain:
self.field = fft(field[self.numb_ch - 1])
self.field *= self.ampl_main.sqrt_G
if self.numb_ch > 1:
return [field[0], ifft(self.field)]
else:
return [ifft(self.field), field[1]]
def wdm_exp_f_cached(self, As, h):
if self.cached_factor is None:
print "Caching linear factor"
self.cached_factor = [
np.exp(h * self.factor[0]),
np.exp(h * self.factor[1])
]
return np.asarray([
ifft(self.cached_factor[0] * fft(As[0])),
ifft(self.cached_factor[1] * fft(As[1]))
])
def __call__(self, domain, field):
if self.ind is None:
self.ind = max(np.where(domain.nu < self.cutoff_nu)[0])
if self.field_pump is not None:
self.field_pump = ifftshift(fft(self.field_pump))
if self.dir_to_up is True:
self.field_pump[self.ind:] = 0.0
else:
self.field_pump[:self.ind] = 0.0
else:
self.field_pump = np.zeros(field.size, dtype=field.dtype)
if not (len(self.field_pump) == len(field)):
raise DiffArraysError(
"arrays of fields is different lengths")
self.field = ifftshift(fft(field))
if self.dir_to_up is True:
self.field[:self.ind] = 0.0
else:
self.field[self.ind:] = 0.0
return ifft(fftshift(self.field + self.field_pump))
def __call__(self, domain, field):
self.field = fft(field[0])
factor = 1.0 / (np.exp(self.slope *
(domain.nu / self.centre_nu - 1.0)) + 1.0)
self.field *= factor
self.field = ifft(self.field)
return [field[0] - self.field, self.field]
def __call__(self, domain, field):
# Convert field to spectral domain:
self.field = fft(field)
# Calculate linear gain from logarithmic gain (G_dB -> G_linear)
G = power(10, 0.1 * self.gain)
if self.E_sat is not None:
E = energy(field, domain.t)
G = G / (1.0 + E / self.E_sat)
sqrt_G = sqrt(G)
if domain.channels > 1:
self.field[0] *= sqrt_G
self.field[1] *= sqrt_G
else:
self.field *= sqrt_G
# convert field back to temporal domain:
return ifft(self.field)
def __call__(self, domain, field):
# Convert field to spectral domain:
self.field = fft(field)
if self.gain is not None:
# Calculate linear gain from logarithmic gain (G_dB -> G_linear)
G = power(10, 0.1 * self.gain)
sqrt_G = sqrt(G)
if domain.channels > 1:
self.field[0] *= sqrt_G
self.field[1] *= sqrt_G
else:
self.field *= sqrt_G
if self.power is not None:
pass
# convert field back to temporal domain:
return ifft(self.field)
def __call__(self, domain, field):
# Convert field to spectral domain:
self.field = fft(field)
if self.gain is not None:
# Calculate linear gain from logarithmic gain (G_dB -> G_linear)
G = power(10, 0.1 * self.gain)
sqrt_G = sqrt(G)
if domain.channels > 1:
self.field[0] *= sqrt_G
self.field[1] *= sqrt_G
else:
self.field *= sqrt_G
if self.power is not None:
pass
# convert field back to temporal domain:
return ifft(self.field)
def __call__(self, domain, field):
"""
:param object domain: A domain
:param object field: Current field
:return: Field after modification by Gaussian filter
:rtype: Object
"""
# Convert field to spectral domain:
self.field = fft(field)
delta_nu = domain.nu - domain.centre_nu - self.offset_nu
factor = power(delta_nu / self.width_nu, (2 * self.m))
# Frequency values are in order, inverse shift to put in fft order:
self.shape = exp(-0.5 * ifftshift(factor))
if domain.channels > 1:
# Filter is applied only to one channel:
self.field[self.channel] *= self.shape
else:
self.field *= self.shape
# convert field back to temporal domain:
return ifft(self.field)
def default_f_with_rs(self, A, z):
""" Use Raman-scattering term. """
term_spm = np.abs(A) ** 2
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return self.factor * (term_spm + term_rs) * A
def __call__(self, domain, field):
factor = self.linearity(domain)
return ifft(np.exp(factor) * fft(field))
def default_f(self, A, z):
return ifft(self.factor * fft(A))
def __call__(self, domain, field):
self.field = fft(field)
self.field = ifft(self.matrix_oper(self.field[0], self.field[1]))
return self.field
def default_exp_f_with_rs(self, A, h, B):
""" Use Raman-scattering in exponential term. """
term_spm = np.abs(A)**2
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return np.exp(h * self.factor * (term_spm + term_rs)) * B
def default_f(self, A, z):
return ifft(self.factor * fft(A))
def default_exp_f_with_ss(self, A, h, B):
""" Use self-steepening in exponential term. """
term_spm = np.abs(A)**2
term_ss = (self.ss_factor / B) * ifft(self.omega * fft(term_spm * A))
return np.exp(h * self.factor * (term_spm + term_ss)) * B
def wdm_exp_f(self, As, h):
return np.asarray([ifft(np.exp(h * self.factor[0]) * fft(As[0])),
ifft(np.exp(h * self.factor[1]) * fft(As[1]))])
def default_exp_f(self, A, h):
return ifft(np.exp(h * self.factor) * fft(A))
def wdm_f(self, As, z):
return np.asarray([ifft(self.factor[0] * fft(As[0])),
ifft(self.factor[1] * fft(As[1]))])
def default_exp_f_cached(self, A, h):
if self.cached_factor is None:
print "Caching linear factor"
self.cached_factor = np.exp(h * self.factor)
return ifft(self.cached_factor * fft(A))
def default_exp_f(self, A, h):
return ifft(np.exp(h * self.factor) * fft(A))
def default_exp_f_with_ss(self, A, h, B):
""" Use self-steepening in exponential term. """
term_spm = np.abs(A) ** 2
term_ss = (self.ss_factor / B) * ifft(self.omega * fft(term_spm * A))
return np.exp(h * self.factor * (term_spm + term_ss)) * B
def default_exp_f(self, A, h):
hf = self.factor * h
if self.phase_lim:
hf = self._limit_imag_part(hf)
return ifft(np.exp(hf) * fft(A))
def default_f_with_ss(self, A, z):
""" Use self-steepening only. """
term_spm = np.abs(A)**2 * A
term_ss = self.ss_factor * ifft(self.omega * fft(term_spm))
return self.factor * (term_spm + term_ss)
def default_exp_f_cached(self, A, h):
if self.cached_factor is None:
self.cache(h)
return ifft(self.cached_factor * fft(A))
def default_f_with_rs(self, A, z):
""" Use Raman-scattering term. """
term_spm = np.abs(A)**2
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return self.factor * (term_spm + term_rs) * A
def wdm_f(self, As, z):
return np.asarray([ifft(self.factor[0] * fft(As[0])),
ifft(self.factor[1] * fft(As[1]))])
def default_f_with_ss(self, A, z):
""" Use self-steepening only. """
term_spm = np.abs(A) ** 2 * A
term_ss = self.ss_factor * ifft(self.omega * fft(term_spm))
return self.factor * (term_spm + term_ss)
def default_exp_f_with_rs(self, A, h, B):
""" Use Raman-scattering in exponential term. """
term_spm = np.abs(A) ** 2
term_rs = self.rs_factor * ifft(1j * self.omega * fft(term_spm))
return np.exp(h * self.factor * (term_spm + term_rs)) * B
def __call__(self, domain, field):
self.Domega = domain.omega - domain.centre_omega
self.factor = 1j * fftshift(self.time * self.Domega)
return ifft(np.exp(self.factor) * fft(field))
def wdm_exp_f_cached(self, As, h):
if self.cached_factor is None:
self.cache(h)
return np.asarray([ifft(self.cached_factor[0] * fft(As[0])),
ifft(self.cached_factor[1] * fft(As[1]))])
def default_exp_f_cached(self, A, h):
if self.cached_factor is None:
print "Caching linear factor"
self.cached_factor = np.exp(h * self.factor)
return ifft(self.cached_factor * fft(A))
def wdm_exp_f(self, As, h):
return np.asarray([
ifft(np.exp(h * self.factor[0]) * fft(As[0])),
ifft(np.exp(h * self.factor[1]) * fft(As[1]))
])