def test_bounce_kernel(self):
amp = 0.1
tau = 10e-9
length = 100e-9
sampling_rate = 1e9
ker_real = kf.bounce_kernel(amp=amp,
time=tau,
length=length,
sampling_rate=sampling_rate)
ker_sampled = kf.bounce_kernel(amp=amp,
time=tau * sampling_rate,
length=length * sampling_rate,
sampling_rate=1)
np.testing.assert_almost_equal(ker_real, ker_sampled)
nr_samples = int(length * sampling_rate)
t_kernel = np.arange(nr_samples) / sampling_rate
bounce = kf.bounce(t_kernel,
amp=amp,
time=tau,
sampling_rate=sampling_rate)
y_corr0 = np.convolve(ker_real, bounce)
np.testing.assert_almost_equal(y_corr0[10:80], np.ones(70), decimal=2)
def bounce_correction(ysig,
tau: float,
amp: float,
sampling_rate: float = 1,
inverse: bool = False):
"""
Corrects for a bounce
Args:
ysig: the signal to be predistorted
tau: the time at which the bounce occurs
amp: the amplitude of the bounce correction
sampling_rate: the sampling rate of the signal
returns:
filtered_signal : the signal corrected for the bounce
"""
# kernel is cut of after 8*tau, this menas that it will only correct
# bounces up to 8th order, this is good for coefficients << 1
kern = bounce_kernel(amp,
time=tau,
length=8 * tau,
sampling_rate=sampling_rate)
if inverse:
raise NotImplemented
else:
filter_signal = np.convolve(ysig, kern, mode='full')
return filter_signal[:len(ysig)]
def test_bounce_kernel(self):
bl = 40e-9
ba = .2
bt = 12e-9
self.k0.bounce_amp_1(ba)
self.k0.bounce_tau_1(bt)
self.k0.bounce_length_1(bl)
kObj_bounce = self.k0.get_bounce_kernel_1()
kf_bounce = kf.bounce_kernel(amp=ba, time=bt * 1e9, length=bl * 1e9)
np.testing.assert_almost_equal(kObj_bounce, kf_bounce)
def get_bounce_kernel_2(self):
return kf.bounce_kernel(amp=self.bounce_amp_2(),
time=self.bounce_tau_2()*self.sampling_rate(),
length=self.bounce_length_2()*self.sampling_rate())
