def generate_eye_data(domain, field, scale=None):
"""
:param object domain: A domain
:param object field: Current field
:param double scale: Scale factor by which to divide power
:return: Temporal array and power matrix
:rtype: Tuple
Generates a temporal array and matrix containing eye traces. Each eye
trace consists of the power array calculated for successive bits.
"""
# Use the first bit range for the temporal axis:
t_eye = domain.t[0:domain.samples_per_bit]
# Calculate power array and normalise if requested:
P_t = temporal_power(field)
if(scale is not None):
P_t /= scale
# Reshape into matrix with total_bits rows and samples_per_bit columns:
P_eye = P_t.reshape(domain.total_bits, domain.samples_per_bit)
# If passed a matrix, pyplot will plot each column as a separate data
# series. Require each row to be plotted, so transpose the matrix:
return t_eye, P_eye.T
def calculate(self):
""" Actual calculation of metrics is performed in this function. """
maximum_absolute_difference = -1.0
sample_threshold = -1
maximum_Q = -1.0
sample_time = -1.0
TB = self.domain.total_bits
SPB = self.domain.samples_per_bit
for spb in range(SPB):
data = [temporal_power(self.field[tb * SPB + spb])
for tb in range(TB)]
threshold = np.sum(data) / TB
zeros = []
ones = []
for datum in data:
if(datum < threshold):
zeros.append(datum)
else:
ones.append(datum)
if((len(zeros) < 4) or (len(ones) < 4)):
print "Not enough ones and zeros to calculate Q"
#~raise Exception("Not enough ones and zeros to calculate Q")
absolute_difference = np.abs(np.mean(ones) - np.mean(zeros))
if(absolute_difference > maximum_absolute_difference):
maximum_absolute_difference = absolute_difference
maximum_Q = \
absolute_difference / (np.std(ones) + np.std(zeros))
sample_time = spb * self.domain.dt
sample_threshold = threshold
# Store ones and zeros arrays:
self.ones = ones
self.zeros = zeros
if(maximum_Q < 0.0):
raise Exception("Unable to calculate maximum Q!")
self.max_Q_dB = 10.0 * np.log10(maximum_Q)
self.sample_time = sample_time
self.sample_threshold = sample_threshold
self.mean_peak_power_ones = np.mean(self.ones)
self.mean_peak_power_zeros = np.mean(self.zeros)
delta_ones = np.max(self.ones) - np.min(self.ones)
AJ = 0.5 * delta_ones / np.mean(self.ones)
self.amplitude_jitter = AJ
ER = np.max(self.zeros) / (np.mean(self.ones) * (1.0 - AJ))
self.extinction_ratio = -10.0 * np.log10(ER)
