def groupbywtm(waves,
types,
modes,
freqs,
values,
dvalues=None,
keepnans=True):
"""group outputs from dispersion by wave, type, modes
:param waves: iterable, "R" or "L", Rayleigh or Love
:param types: iterable, "C" or "U", phase or group
:param modes: iterable, ints, mode numbers
:param freqs: iterable, floats, frequencies, Hz
:param values: iterable, floats, velocity values, km/s
:param dvalues: None or iterable, floats, uncertainties of values, km/s
:param keepnans: bool, by default the nans are removed from the arrays
(this occurs especially for overtones below the cut-off period)
all arrays must be the same length
groupbywtm demultiplexes these arrays to produce, for each mode
w (wave key, scalar), t (type key, scalar), m (mode number, scalar), freqs (array), values (array), dvalues (array)
"""
freqs = np.array(freqs, float)
types = np.array(list(types), "|S1")
waves = np.array(list(waves), "|S1")
modes = np.array(modes, int)
assert len(waves) == len(types) == len(modes) == len(freqs)
if dvalues is not None:
dvalues = np.array(dvalues, float)
assert len(waves) == len(dvalues)
w, t, m = munique(waves, types, modes)
I = np.lexsort((m, t, w))
for w, t, m in zip(w[I], t[I], m[I]):
J = (waves == w) & (types == t) & (modes == m)
if keepnans:
K = np.ones(len(values[J]), bool)
else:
K = ~np.isnan(values[J])
L = np.argsort(freqs[J][K])
if dvalues is None:
yield w, t, m, freqs[J][K][L], values[J][K][L]
else:
yield w, t, m, freqs[J][K][L], values[J][K][L], dvalues[J][K][L]
def wtm(self):
for w, t, m in zip(*munique(self.data['WAVE'], self.data['TYPE'], self.data['MODE'])):
yield w, t, m
def get_all(self):
for WAVE, TYPE, FLAG, MODE in \
zip(*munique(self.data['WAVE'], self.data['TYPE'], self.data['FLAG'], self.data['MODE'])):
yield self.get(mode=MODE, wave=WAVE, type=TYPE, flag=FLAG)