def __init__(self,
a1=0.3,
a2=0.7,
tau=64,
sigma=1.5,
roessler_kwargs={},
return_input=False):
self._a1 = a1
self._a2 = a2
self._tau = tau
self._sigma = sigma
self._roessler_kwargs = roessler_kwargs
self._return_input = return_input #wether also the Roessler-timeseries should be returned
self._tsr = TwoStochasticRoessler(**roessler_kwargs)
def __init__(self,a1=0.3,a2=0.7,tau=64,sigma=1.5,roessler_kwargs={},return_input=False):
self._a1 = a1
self._a2 = a2
self._tau = tau
self._sigma = sigma
self._roessler_kwargs = roessler_kwargs
self._return_input = return_input #wether also the Roessler-timeseries should be returned
self._tsr = TwoStochasticRoessler(**roessler_kwargs)
class Winterhalder(object):
"""Implements a autoregressive (thus lowpass-filtered) and time-delayed
propagated signal. It uses a system of two diffusively coupled stochastic
Roessler oscillators.
cp. Winterhalder et al., PLA A 356 (2006), 26-34
Equation:
u(t) = a_1 u(t) + a_2 x(t-tau) + sigma n(t)
"""
def __init__(self,
a1=0.3,
a2=0.7,
tau=64,
sigma=1.5,
roessler_kwargs={},
return_input=False):
self._a1 = a1
self._a2 = a2
self._tau = tau
self._sigma = sigma
self._roessler_kwargs = roessler_kwargs
self._return_input = return_input #wether also the Roessler-timeseries should be returned
self._tsr = TwoStochasticRoessler(**roessler_kwargs)
def integrate(self, ts):
"""Integrate the underlying system and propagate and dilate it."""
ts = np.array(ts)
timestep = ts[1] - ts[
0] # Assume equally spaced times, needed for TwoStochasticRoessler anyways
ts += 100
ts = np.concatenate(
[np.arange(ts[0] - 1000 * timestep, ts[0], timestep),
ts]) #iteriere 500 Schritte vor
print ts, np.diff(ts).min(), np.diff(ts).max(), np.median(
np.diff(ts)), ts[998:1002]
roe_ts = self._tsr.integrate(ts)
#Define Shorthands
a1 = self._a1
a2 = self._a2
tau = self._tau
sigma = self._sigma
rv = roe_ts[:, 0].copy()
for i in range(tau, roe_ts.shape[0]):
if i % 100 == 0:
print i
rv[i] = a1 * rv[i - 1] + a2 * roe_ts[
i - tau, 0] + sigma * np.random.normal(0, 1)
if self._return_input:
return rv[1000:], roe_ts[1000:, :]
else:
return rv[1000:]
class Winterhalder(object):
"""Implements a autoregressive (thus lowpass-filtered) and time-delayed
propagated signal. It uses a system of two diffusively coupled stochastic
Roessler oscillators.
cp. Winterhalder et al., PLA A 356 (2006), 26-34
Equation:
u(t) = a_1 u(t) + a_2 x(t-tau) + sigma n(t)
"""
def __init__(self,a1=0.3,a2=0.7,tau=64,sigma=1.5,roessler_kwargs={},return_input=False):
self._a1 = a1
self._a2 = a2
self._tau = tau
self._sigma = sigma
self._roessler_kwargs = roessler_kwargs
self._return_input = return_input #wether also the Roessler-timeseries should be returned
self._tsr = TwoStochasticRoessler(**roessler_kwargs)
def integrate(self,ts):
"""Integrate the underlying system and propagate and dilate it."""
ts = np.array(ts)
timestep = ts[1]-ts[0] # Assume equally spaced times, needed for TwoStochasticRoessler anyways
ts += 100
ts = np.concatenate([np.arange(ts[0]-1000*timestep,ts[0],timestep),ts]) #iteriere 500 Schritte vor
print ts, np.diff(ts).min(),np.diff(ts).max(),np.median(np.diff(ts)),ts[998:1002]
roe_ts = self._tsr.integrate(ts)
#Define Shorthands
a1 = self._a1
a2 = self._a2
tau = self._tau
sigma = self._sigma
rv = roe_ts[:,0].copy()
for i in range(tau,roe_ts.shape[0]):
if i%100==0:
print i
rv[i] = a1*rv[i-1] + a2*roe_ts[i-tau,0] + sigma*np.random.normal(0,1)
if self._return_input:
return rv[1000:], roe_ts[1000:,:]
else:
return rv[1000:]
