-
Notifications
You must be signed in to change notification settings - Fork 0
/
fit_foundation.py
98 lines (71 loc) · 3.48 KB
/
fit_foundation.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
from bayesn_model import SEDmodel
import numpy as np
import os.path
from numpyro.infer import init_to_value
from jax.random import PRNGKey, normal
import jax.numpy as jnp
import pandas as pd
model = SEDmodel(load_model='T21_model')
# dataset = 'Foundation_DR1'
# dataset = 'sim_nonzero_eps'
dataset = 'T21_training_set'
epsilons_on = True
# if os.path.exists("results/" + dataset):
# raise ValueError("It looks like this dataset has already been fit.")
# else:
# os.makedirs("results/" + dataset)
filt_map_dict = {'g': 'g_PS1', 'r': 'r_PS1', 'i': 'i_PS1', 'z': 'z_PS1'}
# foundation_file = open("data/lcs/Foundation_DR1/Foundation_DR1/Foundation_DR1.LIST","r")
# s = 'Foundation_DR1_ASASSN-15fa.txt\n'
# sn_names = [s[15:-5] for s in foundation_file.readlines()]
sn_list = pd.read_csv('data/lcs/tables/' + dataset + '.txt', comment='#', delim_whitespace=True, names=['sn', 'source', 'files'])
sn_names = (sn_list.sn.values)
best_ks = []
last_ks = []
best_samples_arr = []
last_samples_arr = []
def postprocess_add_mu(model, samples):
# num_sn = samples['Ds'].shape[0]
num_sn = 1
samples['Ds'] = samples['Ds'].reshape((num_sn,5000))
muhat = model.data[-3, 0, :]
# print(muhat.shape)
muhat_err = 10
Ds_err = jnp.sqrt(muhat_err * muhat_err + model.sigma0 * model.sigma0)
mu_mean = (np.squeeze(samples['Ds']) * jnp.power(muhat_err, 2) + muhat[...,None] * jnp.power(model.sigma0, 2)) / jnp.power(Ds_err, 2)
mu_sigma = jnp.sqrt((jnp.power(model.sigma0, 2) * jnp.power(muhat_err, 2)) / jnp.power(Ds_err, 2))
standard_normal_samples = normal(PRNGKey(123), shape=mu_mean.shape)
mu = mu_mean + standard_normal_samples * mu_sigma
# print(mu.shape)
delM = np.squeeze(samples['Ds']) - mu
samples['mu'] = mu
samples['delM'] = delM
return samples
for i, sn_name in enumerate(sn_names):
print(i, sn_name)
sn_list = [sn_name]
np.savetxt("temp_sn_list.txt", sn_list, fmt="%s", header="SNID", comments="")
# model.process_dataset('foundation', 'data/lcs/Foundation_DR1/Foundation_DR1/Foundation_DR1_' +sn_name + '.txt','data/lcs/Foundation_DR1/Foundation_DR1_' +sn_name + '.txt',
# filt_map_dict, data_mode='flux')
model.process_dataset('foundation', 'data/lcs/tables/' + dataset + '.txt', 'data/lcs/meta/' + dataset + '_meta.txt',
filt_map_dict, data_mode='flux', sn_list="temp_sn_list.txt")
# print("Fitting MCMC...")
# model.fit(250, 250, 4, str(dataset) + "/" + str(i) + '_mcmc',
# epsilons_on=epsilons_on, chain_method='parallel',
# init_strategy='median')
print("Fitting VI...")
# model.fit_with_vi(str(dataset) + "/" + str(i) + '_vi', init_strategy=init_to_value(values={'AV':jnp.array([0.01]), 'theta':jnp.array([1.]), 'Ds':jnp.array([35.])}))
# model.fit_with_vi(str(dataset) + "/" + str(i) + '_vi', init_strategy='median')
best_k, last_k, best_samples, last_samples = model.fit_zltn_get_ks(model.data, model.band_weights)
best_samples = postprocess_add_mu(model, best_samples)
last_samples = postprocess_add_mu(model, last_samples)
best_ks.append(best_k)
last_ks.append(last_k)
best_samples_arr.append(best_samples)
last_samples_arr.append(last_samples)
print(best_ks, last_ks)
# print(best_samples)
np.savetxt("foundation_results/best_ks_032824.txt", np.array(best_ks))
np.savetxt("foundation_results/last_ks_032824.txt", np.array(last_ks))
np.savez("foundation_results/best_samples_032824", np.array(best_samples_arr))
np.savez("foundation_results/last_samples_032824", np.array(last_samples_arr))