notebook.community

Edit and run



In [18]:

    
import cov_utils as mu



In [19]:

    
import numpy as np



In [20]:

    
def generateCov(dims, seed=None, low=-0.5, high=0.5):
    if seed is not None:
        np.random.seed(seed)
    x = np.random.uniform(low, high, size=dims)
    y = np.random.uniform(low, high,size=dims)
    m = np.outer(x, y)
    return np.dot(m, m.transpose())



In [21]:

    
cov = generateCov(dims=3, seed=42)



In [22]:

    
cov









    Out[22]:





array([[ 0.00387832, -0.01393287, -0.0071716 ],
       [-0.01393287,  0.05005381,  0.02576395],
       [-0.0071716 ,  0.02576395,  0.01326135]])



In [23]:

    
Cov = mu.covariance(cov, paramNames=['x0', 'x1', 'x2'])



In [24]:

    
Cov



In [26]:

    
mu.subcovariance(Cov, paramList=['x0', 'x2'])



In [27]:

    
Cov.ix[['x0', 'x1'],['x0', 'x1']].values









    Out[27]:





array([[ 0.00387832, -0.01393287],
       [-0.01393287,  0.05005381]])



In [28]:

    
dfc = mu.covariance(cov, paramNames=['x0', 'x1', 'x2']); dfc



In [29]:

    
Cov



In [31]:

    
mu.log_covariance(Cov, 'x0', 55.0)



In [32]:

    
isinstance('1', str)









    Out[32]:





True



In [33]:

    
Cov.iloc[0]









    Out[33]:





x0    0.000001
x1   -0.000253
x2   -0.000130
Name: x0, dtype: float64



In [34]:

    
Cov.values









    Out[34]:





array([[  1.28209084e-06,  -2.53324946e-04,  -1.30392694e-04],
       [ -2.53324946e-04,   5.00538076e-02,   2.57639484e-02],
       [ -1.30392694e-04,   2.57639484e-02,   1.32613495e-02]])



In [35]:

    
Cov.values.diagonal()









    Out[35]:





array([  1.28209084e-06,   5.00538076e-02,   1.32613495e-02])



In [ ]:

	x0	x1	x2
x0	0.003878	-0.013933	-0.007172
x1	-0.013933	0.050054	0.025764
x2	-0.007172	0.025764	0.013261

	x0	x1	x2
x0	0.000001	-0.000253	-0.000130
x1	-0.000253	0.050054	0.025764
x2	-0.000130	0.025764	0.013261