def lift_test(train_path, Recovered_test0, Recovered_generate0):
num_levels = 3
num_lags = 0
kernE = kernels.SignatureKernel(base_kernel = gpflow.kernels.Exponential(), \
num_levels = num_levels, order = num_levels, num_lags = num_lags)
recover_path1 = np.array(Recovered_test0)
recover_path2 = np.array(Recovered_generate0)
X = tf.constant([leadlag(p) for p in train_path], 'float64')
Y = tf.constant([leadlag(p) for p in recover_path1], 'float64')
Z = tf.constant([leadlag(p) for p in recover_path2], 'float64')
H1_E = evaluation.Sig_TU2(X, Y, kernE)
H2_E = evaluation.Sig_TU2(X, Z, kernE)
return H1_E, H2_E
def randomise(self):
self.derivatives = np.array(
[self.random_derivative() for _ in range(self.n_points - 1)])
prices = np.r_[0., self.derivatives.cumsum()]
path = leadlag(prices)
self.set_path(path)
def mutate(self, prob=0.1):
for i in range(len(self.derivatives)):
if np.random.random() < prob:
self.derivatives[i] = self.random_derivative()
prices = np.r_[0., np.cumsum(self.derivatives)]
path = leadlag(prices)
self.set_path(path)
def _load_rough_bergomi(self, params):
grid_points_dict = {"M": 28, "W": 5, "Y": 252}
grid_points = grid_points_dict[self.freq]
params["T"] = grid_points / grid_points_dict["Y"]
paths = rough_bergomi(grid_points, **params)
self.windows = [leadlag(path) for path in paths]
def compare_ll_ta_plot(train_path, train_windows):
f, p = plt.subplots(1, 2, figsize=(16, 3))
batchsize = train_path.shape[0]
idx = np.random.randint(0, batchsize)
l1 = plot_helper_2dpath(p[0], leadlag(train_path[idx]), 'tab:blue')
l2 = plot_helper_2dpath(p[1], (train_windows[idx]), 'tab:orange')
p[0].legend((l1, ), ['lead-lag transformation'])
p[1].legend((l2, ), ['Time augmentation'])
for i in range(2):
p[i].grid()
p[i].set_xlabel('$X^{1}$')
p[i].set_ylabel('$X^{2}$')
plt.show()
def _load_data(self):
try:
self.data = pdr.get_data_yahoo(self.ticker, self.start,
self.end)["Close"]
except:
raise RuntimeError(
f"Could not download data for {self.ticker} from {self.start} to {self.end}."
)
self.windows = []
for _, window in self.data.resample(self.freq):
values = window.values # / window.values[0]
path = leadlag(values)
self.windows.append(path)
def _load_rough_bergomi(params, freq, ll=False):
grid_points_dict = {"M": 28, "W": 10, "Y": 252}
grid_points = grid_points_dict[freq]
N = grid_points - 1
params["T"] = grid_points / grid_points_dict["Y"]
paths = rough_bergomi(grid_points, **params)
if ll:
windows = np.array([leadlag(path) for path in paths])
time = 0
else:
time = np.linspace(0, 1, N + 1)
path_exp = paths[:, :, None]
path_exp = np.array([
np.concatenate([time[:, None], path0], axis=-1)
for path0 in path_exp
])
windows = path_exp
return windows, paths, time
def __add__(self, other):
"""Breed."""
derivatives = []
for derivative1, derivative2 in zip(self.derivatives,
other.derivatives):
if np.random.random() < 0.5:
derivatives.append(derivative1)
else:
derivatives.append(derivative2)
prices = np.r_[0., np.cumsum(derivatives)]
path = leadlag(prices)
o = Organism(self.n_points, self.pip, self.n_pips)
o.derivatives = derivatives
o.set_path(path)
return o
