

In [4]:

    
%load_ext autoreload
%autoreload 2









    



The autoreload extension is already loaded. To reload it, use:
  %reload_ext autoreload



In [5]:

    
from pearce.emulator import OriginalRecipe, ExtraCrispy
from pearce.mocks import cat_dict
import numpy as np
from os import path



In [6]:

    
import matplotlib
#matplotlib.use('Agg')
from matplotlib import pyplot as plt
%matplotlib inline
import seaborn as sns
sns.set()



In [7]:

    
training_file = '/u/ki/swmclau2/des/PearceRedMagicXiChinchilla.hdf5'

em_method = 'nn'
split_method = 'random'



In [8]:

    
a = 1.0
z = 1.0/a - 1.0



In [ ]:

    
fixed_params = {'z':z}

n_leaves, n_overlap = 50, 1 emu = ExtraCrispy(training_file, n_leaves, n_overlap, split_method, method = em_method, fixed_params=fixed_params, custom_mean_function = None, downsample_factor = 1.0)



In [ ]:

    
emu = OriginalRecipe(training_file, method = em_method, fixed_params=fixed_params,
                    hyperparams = {'hidden_layer_sizes': (50, 25),
                                 'activation': 'relu', 'verbose': False, 
                                    'tol': 1e-8, 'learning_rate_init':1e-6,\
                                   'max_iter':5000, 'alpha':0, 'early_stopping':False, 'validation_fraction':0.3})

emu = OriginalRecipe(training_file, method = em_method, fixed_params=fixed_params, hyperparams = {'hidden_layer_sizes': (1000, 500, 300), 'activation': 'relu', 'verbose': True, 'tol': 1e-8, 'learning_rate_init':5e-5,\ 'max_iter':2000, 'alpha':0})



In [ ]:

    
np.abs(emu.goodness_of_fit(training_file, statistic = 'log_frac')).mean()



In [ ]:

    
np.abs(emu.goodness_of_fit(training_file, statistic = 'frac')).mean()



In [ ]:

    
fit_idxs = np.argsort(gof.mean(axis = 1))



In [ ]:

    
emu.goodness_of_fit(training_file).mean()#, statistic = 'log_frac')).mean()



In [ ]:

    
model = emu._emulator



In [ ]:

    
ypred = model.predict(emu.x)



In [ ]:

    
plt.hist( np.log10( (emu._y_std+1e-5)*np.abs(ypred-emu.y)/np.abs((emu._y_std+1e-5)*emu.y+emu._y_mean) ))



In [ ]:

    
( (emu._y_std+1e-5)*np.abs(ypred-emu.y)/np.abs((emu._y_std+1e-5)*emu.y+emu._y_mean) ).mean()



In [ ]:

    
emu._y_mean, emu._y_std



In [ ]:

    
for idx in fit_idxs[:10]:
    print gof[idx].mean()
    print (ypred[idx*emu.n_bins:(idx+1)*emu.n_bins]-emu.y[idx*emu.n_bins:(idx+1)*emu.n_bins])/emu.y[idx*emu.n_bins:(idx+1)*emu.n_bins]
    plt.plot(emu.scale_bin_centers, ypred[idx*emu.n_bins:(idx+1)*emu.n_bins], label = 'Emu')
    plt.plot(emu.scale_bin_centers, emu.y[idx*emu.n_bins:(idx+1)*emu.n_bins], label = 'True')
    plt.legend(loc='best')
    plt.xscale('log')
    plt.show()



In [ ]:

    
print dict(zip(emu.get_param_names(), emu.x[8*emu.n_bins, :]*emu._x_std+emu._x_mean))



In [ ]:

    
emu.get_param_names()

#print emu.x.shape #print emu.downsample_x.shape if hasattr(emu, "_emulators"): print emu._emulators[0]._x.shape else: print emu._emulator._x.shape



In [ ]:

    
emu._ordered_params

x, y, y_pred = emu.goodness_of_fit(training_file, statistic = 'log_frac')

x, y, y_pred

N = 25 for _y, yp in zip(y[:N], y_pred[:N]): #plt.plot(emu.scale_bin_centers , (_y - yp)/yp ,alpha = 0.3, color = 'b') plt.plot(emu.scale_bin_centers, _y, alpha = 0.3, color = 'b') plt.plot(emu.scale_bin_centers, yp, alpha = 0.3, color = 'r') plt.loglog();

%%timeit #truth_file = '/u/ki/swmclau2/des/PearceRedMagicWpCosmoTest.hdf5' gof = emu.goodness_of_fit(training_file, N = 100, statistic = 'log_frac')



In [ ]:

    
gof = emu.goodness_of_fit(training_file, statistic = 'frac')
print gof.mean()



In [ ]:

    
for row in gof:
    print row



In [ ]:

    
gof = emu.goodness_of_fit(training_file, statistic = 'frac')
print gof.mean()



In [ ]:

    
model = emu._emulator



In [ ]:

    
model.score(emu.x, emu.y)



In [ ]:

    
ypred = model.predict(emu.x)

np.mean(np.abs(ypred-emu.y)/emu.y)



In [ ]:

    
plt.plot(emu.scale_bin_centers, np.abs(gof.mean(axis = 0)) )
plt.plot(emu.scale_bin_centers, np.ones_like(emu.scale_bin_centers)*0.01)
plt.plot(emu.scale_bin_centers, np.ones_like(emu.scale_bin_centers)*0.05)
plt.plot(emu.scale_bin_centers, np.ones_like(emu.scale_bin_centers)*0.1)


plt.loglog();



In [ ]:

    
plt.plot(emu.scale_bin_centers, np.abs(gof.T),alpha = 0.1, color = 'b')
plt.plot(emu.scale_bin_centers, np.ones_like(emu.scale_bin_centers)*0.01, lw = 2, color = 'k')
plt.loglog();



In [ ]:

    
gof[:,i].shape



In [ ]:

    
for i in xrange(gof.shape[1]):
    plt.hist(np.log10(gof[:, i]), label = str(i), alpha = 0.2);
    
plt.legend(loc='best')
plt.show()



In [ ]: