def simpleVarveModel(signal, H, shape=1.5, mean=1, SNR=.25, seed=0):
''' Simple PSM for varves
Args:
signal (array): the signal matrix, each row is a time series
H (float): Hurst index, should be in (0, 1)
'''
# first, gammify the input
gamSig = gammify(signal, shape=shape, mean=mean, seed=seed)
# create gammafied autocorrelated fractal brownian motion series
# wa = Spectral.WaveletAnalysis()
# fBm_ts = wa.fBMsim(N=np.size(signal), H=H)
fBm_ts = fbm.fgn(np.size(signal), H)
gamNoise = gammify(fBm_ts, shape=shape, mean=mean, seed=seed)
# combine the signal with the noise, based on the SNR
varves = (gamSig * SNR + gamNoise * (1 / SNR)) / (SNR + 1 / SNR)
res = {
'signal': signal,
'gamSig': gamSig,
'fBm_ts': fBm_ts,
'gamNoise': gamNoise,
'varves': varves,
'H': H,
'shape': shape,
'mean': mean,
'SNR': SNR,
'seed': seed,
}
return res
def test_fgn(n_good, hurst_good, length_good, fbm_method_good):
fgn_sample = fgn(n_good, hurst_good, length_good, fbm_method_good)
assert isinstance(fgn_sample, np.ndarray)
assert len(fgn_sample) == n_good
Serie = fbm(n=Size, hurst=H)
a[seed],S[seed],_,O[seed] = evaluate_omega(Serie,1)
ax[0].plot(a,S,'o',mec='k',mfc='r',markersize=5,label=r'fBm',alpha=1)#,markersize=10,mfc='w',mec='r',mew=2,label=r'$\beta x$ Map',alpha=0.8)
ax[1].plot(a,O,'o',mec='k',mfc='r',markersize=5,alpha=1)#,markersize=10,mfc='w',mec='r',mew=2,label=r'$\beta x$ Map',alpha=0.8)
print('=>-------- 20%')
a = np.zeros(CI)
S = np.zeros(CI)
O = np.zeros(CI)
for seed in range(CI):
np.random.seed(seed)
H = 0.6*seed/CI + 0.2
Serie = fgn(n=Size, hurst=H)
a[seed],S[seed],_,O[seed] = evaluate_omega(Serie,1)
ax[0].plot(a,S,'s',mec='k',mfc='g',markersize=5,label=r'fGn',alpha=1)#,markersize=10,mfc='w',mec='r',mew=2,label=r'$\beta x$ Map',alpha=0.8)
ax[1].plot(a,O,'s',mec='k',mfc='g',markersize=5,alpha=1)#,markersize=10,mfc='w',mec='r',mew=2,label=r'$\beta x$ Map',alpha=0.8)
print('==>------- 30%')
#Cauchy noise
a = np.zeros(CI)
S = np.zeros(CI)
O = np.zeros(CI)
for seed in range(CI):
from fbm import fbm, fgn
import numpy as np
import csv
n_series = 100
n_size = 50000
fbm_matrix = []
fgn_matrix = []
for i in range(n_series):
fbm_sample = fbm(n=n_size, hurst=0.1, length=1, method='daviesharte')
fgn_sample = fgn(n=n_size, hurst=0.5, length=1, method='daviesharte')
fbm_matrix.append(fbm_sample)
fgn_matrix.append(fgn_sample)
print(i)
with open("../Data/fBm.csv", "w+") as my_csv:
csvWriter = csv.writer(my_csv, delimiter=',')
csvWriter.writerows(fbm_matrix)
with open("../Data/fGn.csv", "w+") as my_csv:
csvWriter = csv.writer(my_csv, delimiter=',')
csvWriter.writerows(fgn_matrix)