notebook.community

Edit and run



In [32]:

    
import numpy as np



In [33]:

    
pred_y = np.load('/u/ki/swmclau2/Git/pearce/bin/covmat/pred_y.npy')



In [34]:

    
data_y = np.load('/u/ki/swmclau2/Git/pearce/bin/covmat/data_y.npy')



In [35]:

    
np.sum(np.all(data_y == 0.0, axis = 0))









    Out[35]:





0



In [36]:

    
data_y.shape









    Out[36]:





(18, 34803)



In [37]:

    
from matplotlib import pyplot as plt
%matplotlib inline



In [47]:

    
R = (10**pred_y- 10**data_y)/(10**data_y)
##R = (pred_y - data_y)
#cov = R.dot(R.T)/(R.shape[1]-1)



In [48]:

    
plt.hist(R[0]);
#plt.xscale('log')
#plt.yscale('log');









    Out[48]:





(array([3.3165e+04, 1.2180e+03, 2.9000e+02, 8.4000e+01, 2.8000e+01,
        9.0000e+00, 5.0000e+00, 3.0000e+00, 0.0000e+00, 1.0000e+00]),
 array([-0.98433856,  1.60062725,  4.18559306,  6.77055888,  9.35552469,
        11.9404905 , 14.52545632, 17.11042213, 19.69538794, 22.28035376,
        24.86531957]),
 <a list of 10 Patch objects>)



In [54]:

    
10**data_y[0, R[0] > 1]









    Out[54]:





array([8221.49716631, 8052.12516699, 5576.45102775, ..., 6526.49171429,
       3330.29222323, 4929.39411412])



In [55]:

    
10**pred_y[0, R[0]>1]









    Out[55]:





array([16913.28512377, 16913.28512377, 16913.28512377, ...,
       19513.63976387, 12738.30977647, 12738.30977647])



In [13]:

    
idx = np.all(data_y==0.0, axis = 0)#)[0]



In [14]:

    
idx









    Out[14]:





array([False, False, False, ...,  True, False, False])



In [20]:

    
data_y[:, ~idx]









    Out[20]:





array([[ 4.11838003,  4.10221795,  4.11693687, ...,  4.98437113,
         4.91514731,  4.15238808],
       [ 3.75939547,  3.76115248,  3.75398315, ...,  4.67793847,
         4.61570999,  3.8030483 ],
       [ 3.40152412,  3.39552926,  3.39431375, ...,  4.3643377 ,
         4.30721256,  3.44421774],
       ...,
       [-0.17267683, -0.17037957, -0.17783609, ..., -0.23072924,
        -0.11767299,  0.08055333],
       [-0.44802075, -0.44671271, -0.45747991, ..., -0.5129851 ,
        -0.40135364, -0.22929061],
       [-0.75254936, -0.74923422, -0.76753095, ..., -0.83867383,
        -0.71219954, -0.55200283]])



In [21]:

    
pred_y[:,~idx]









    Out[21]:





array([[ 3.9150556 ,  3.9150556 ,  3.9150556 , ...,  4.26256363,
         4.26256363,  4.26256363],
       [ 3.64043547,  3.64043547,  3.64043547, ...,  3.98569088,
         3.98569088,  3.98569088],
       [ 3.3017141 ,  3.3017141 ,  3.3017141 , ...,  3.71863677,
         3.71863677,  3.71863677],
       ...,
       [-0.18634271, -0.18634271, -0.18634271, ...,  0.12577705,
         0.12577705,  0.12577705],
       [-0.46100042, -0.46100042, -0.46100042, ..., -0.13300802,
        -0.13300802, -0.13300802],
       [-0.76423197, -0.76423197, -0.76423197, ..., -0.44993651,
        -0.44993651, -0.44993651]])

R = (10**pred_y[:,~idx] - 10**data_y[:,~idx]) ##R = (pred_y - data_y) cov = R.dot(R.T)/(R.shape[1]-1)



In [17]:

    
print (np.abs(10**pred_y - 10**data_y)/(10**data_y)).mean(axis =1)









    



[156.69064996  81.58986949  42.67205311  21.62070892  11.61350012
   6.06503616   3.37484636   1.96848642   1.13942661   0.69239594
   0.44652251   0.34090538   0.34407876   0.44513515   0.76723441
   0.45432373   0.39043311   0.38987021]



In [19]:

    
print (np.abs(10**pred_y[:,~idx] - 10**data_y[:,~idx])/(10**data_y[:,~idx])).mean(axis =1)









    



[0.63758194 0.63463386 0.64438464 0.65487216 0.71850737 0.78757685
 0.88191457 0.83872168 0.67990476 0.51379105 0.3750293  0.30920634
 0.32954652 0.43979423 0.76848754 0.45482544 0.38908898 0.38743915]



In [23]:

    
1000*16%88









    Out[23]:





72



In [27]:

    
output = np.load('/nfs/slac/g/ki/ki18/des/swmclau2/xi_gg_corrabzheng07_test_v2/output_0200.npy')



In [28]:

    
np.any(np.all(output == 0.0, axis = 1) )









    Out[28]:





True



In [29]:

    
np.all(output==0.0, axis = 1)









    Out[29]:





array([False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False, False, False, False,
       False, False, False, False, False, False,  True])



In [42]:

    
output[264%88]









    Out[42]:





array([ 5.05407834e+00,  4.74243783e+00,  4.41788886e+00,  4.07753464e+00,
        3.72948666e+00,  3.36533425e+00,  2.99406083e+00,  2.60826370e+00,
        2.18848191e+00,  1.72897029e+00,  1.30115833e+00,  9.90957820e-01,
        7.45103642e-01,  5.10272667e-01,  2.69493097e-01,  2.03637922e-03,
       -2.84013291e-01, -5.91268435e-01])



In [43]:

    
import h5py



In [44]:

    
f = h5py.File('/nfs/slac/g/ki/ki18/des/swmclau2/xi_gg_corrabzheng07_test_v2/PearceXiggCosmoCorrABTest.hdf5', 'r')



In [45]:

    
f['cosmo_no_00']['a_1.000']['obs'].value[265]









    Out[45]:





array([ 4.58452852,  4.23025209,  3.86812086,  3.49986129,  3.1170586 ,
        2.72898277,  2.32850633,  1.95072324,  1.62719941,  1.3721967 ,
        1.16226295,  0.96732309,  0.76413756,  0.560868  ,  0.35742791,
        0.08080695, -0.22537171, -0.54293441])



In [46]:

    
from pearce.emulator import NashvilleHot
#from GPy.kern import *
import numpy as np
from os import path

#training_file = '/home/users/swmclau2/scratch/xi_gg_zheng07_cosmo_v4/PearceXiggCosmo.hdf5'
#test_file = '/home/users/swmclau2/scratch/xi_gg_zheng07_cosmo_test_v4/PearceXiggCosmoTest.hdf5'
training_file = '/nfs/slac/g/ki/ki18/des/swmclau2/xi_gg_corrabzheng07_v2/PearceXiggCosmoCorrAB.hdf5'
test_file = '/nfs/slac/g/ki/ki18/des/swmclau2/xi_gg_corrabzheng07_test_v2/PearceXiggCosmoCorrABTest.hdf5'


em_method = 'gp'
fixed_params = {'z':0.0}
hyperparams = {}

emu = NashvilleHot(training_file, hyperparams=hyperparams,fixed_params = fixed_params, downsample_factor = 0.1)









    



 /u/ki/swmclau2/.local/lib/python2.7/site-packages/pearce/emulator/emu.py:2100: UserWarning:WARNING: NaN detected. Skipped 226 points in training data.
 /u/ki/swmclau2/.local/lib/python2.7/site-packages/paramz/transformations.py:111: RuntimeWarning:overflow encountered in expm1



In [ ]:

    
pred_y, data_y = emu.goodness_of_fit(test_file, statistic = None)



In [ ]: