params_dict=params_dict,
cov_matrix=None,
init_params=init_params,
epoch=epoch,
num_train=num_train,
local_samples=None,
stepStop_n=stepStop_n)
predictor.train(path='SNe_BAO')
chain_ann = predictor.chain_ann
predictor.plot_steps()
predictor.plot_contours(fill_contours=False, show_titles=True)
predictor.save_steps()
predictor.save_contours()
predictor.eco.plot_loss()
#%%
labels = cosmic_params.ParamsProperty(param_names,
params_dict=params_dict).labels
plc.Contours(chain_ann).plot(bins=100,
smooth=5,
labels=labels,
fill_contours=False,
show_titles=True,
best_values=fid_params,
show_best_value_lines=True)
plt.show()
# %% estimate parameters using ECoPANN
randn_num = ; steps_n = 4
predictor = ann.RePredict(union, cov_matrix=None, path='union2.1_fwCDM',
randn_num=randn_num, steps_n=steps_n,
params_dict=simulator.params_dict)
predictor.from_chain()
# predictor.from_net()
chain_ann = predictor.chain_ann
predictor.plot_steps()
predictor.plot_contours(fill_contours=False, show_titles=True)
predictor.save_steps()
predictor.save_contours()
predictor.eco.plot_loss()
#%%
chain_mcmc = np.load('data/MCMC_chains/chain_fwCDM_2params.npy')
chain_all = [chain_ann, chain_mcmc]
param_names = ['w', 'omm']
plc.Contours(chain_all).plot(labels=cosmic_params.ParamsProperty(param_names).labels,smooth=5,
fill_contours=False,show_titles=True,line_width=2,layout_adjust=[0.0,0.0],
lims=None,legend=True,legend_labels=['ANN', 'MCMC'])
#%%
plt.show()
"""
===============================
Triangle plot with one data set
===============================
Plot triangle plot with one data set using some default settings.
"""
import coplot.plot_contours as plc
import matplotlib.pyplot as plt
import numpy as np
test1 = np.random.randn(50000, 3)
# If you want to use some default settings, you can use it like this:
fig = plc.Contours(test1).plot(labels=[r'$a$', r'$b$', r'$c$'], legend=True, subplots_adjust=False)
# and you can save the figure like this:
# pl.savefig('test_triangle_1.pdf', fig)
plt.show()
"""
==========================================================
Triangle plot without 1D contours using multiple data sets
==========================================================
Plot good figures with multiple data sets using more settings.
"""
import coplot.plot_contours as plc
import matplotlib.pyplot as plt
import numpy as np
test1 = np.random.randn(50000, 3)
test2 = test1 + 1
# If you want to plot good figures, you should use it like this:
fig = plc.Contours([test1,test2]).plot_2d(bins=150,labels=[r'$a$', r'$b$', r'$c$'],\
colors=['r','g'],line_styles=['-','--'],fill_contours=False, show_titles=True,
smooth=3,best_values=[[0,0,0],[1,1,1]],show_best_value_lines=False,ticks_size=8,
legend=True,legend_labels=['Data 1', 'Data 2'], subplots_adjust=False)
# and you can save the figure like this:
# pl.savefig('test_triangle_3.pdf', fig)
plt.show()
param_names,
params_dict=params_dict,
cov_matrix=None,
init_params=init_params,
epoch=epoch,
num_train=num_train,
local_samples=None,
stepStop_n=stepStop_n)
predictor.train(path='linear')
chain_ann = predictor.chain_ann
predictor.plot_steps()
predictor.plot_contours(bins=50, fill_contours=False, show_titles=True)
predictor.save_steps()
predictor.save_contours()
predictor.eco.plot_loss()
#%%
labels = cosmic_params.ParamsProperty(param_names,
params_dict=params_dict).labels
plc.Contours(chain_ann).plot(bins=50,
labels=labels,
fill_contours=False,
show_titles=True,
best_values=[a_fid, b_fid],
show_best_value_lines=True)
plt.show()
