-
Notifications
You must be signed in to change notification settings - Fork 0
/
dio_fit.py
152 lines (115 loc) · 4.27 KB
/
dio_fit.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
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
#!/usr/bin/env python3
import disk_in_orbit as dio
import matplotlib as mpl
mpl.use("Agg")
import numpy as np
import matplotlib.pyplot as plt
import emcee
import corner
from astropy.io.fits import getdata
import os
os.environ["OMP_NUM_THREADS"] = "1"
from multiprocessing import Pool
from sys import argv
import pickle
nproc = int(argv[1])
def inv_var_weight_avg(x, ex):
# compute inverse variance weighted average and error
w = 1./(ex**2) # weights
x_wav = np.nansum(w * x, axis=0) / np.nansum(w, axis=0)
σ_wav = 1./np.sqrt(np.nansum(w, axis=0))
return x_wav, σ_wav
def make_flux(mags, mag_errors, baseline_mag):
fluxes = 10**(0.4*(baseline_mag- mags))
flux_errors = fluxes * (10**((2*mag_errors)/5) - 1)
return fluxes, flux_errors
def priors(x):
if any((x<0)|(x>1)) or (x[4]>x[3]):
return -np.inf
else:
return 0.0
def map01(val, lower, upper):
return (val * (upper-lower)) + lower
def mapXs(x):
_x = x.copy()
if len(_x.shape)==1:
_x = _x[np.newaxis,:]
_x[:,0] = t0_est + map01(_x[:,0], -10.0, 10.0) # t0
_x[:,1] = map01(_x[:,1], 0.0, 1.0) # relative velocity
_x[:,2] = map01(_x[:,2], 0.0, 10.0) # minimum impact parameter
_x[:,3] = map01(_x[:,3], 0.0, 10.0) # semimajor axis of disk
_x[:,4] = map01(_x[:,4], 0.0, 10.0) # semiminor axis of disk
_x[:,5] = map01(_x[:,5], 0.0, np.pi) # angle of disk relative to direction of orbital motion
_x[:,6] = map01(_x[:,6], 0.0, 0.04) # transmissivity of disk (0-4%)
if len(x.shape)==1:
return _x[0,:]
return _x
def lnprob(x,I,xp,yp,tp,f,ef):
"""The log-likelihood function."""
prior = priors(x) # all x between 0 and 1
if prior<0:
return prior
_x = mapXs(x)
# generate light curve
_f = dio.get_model_lc_disk(I, xp, yp, tp, *_x)
# calc likelihood
residual = ((f-_f)/ef)**2
_lnprob = -(np.sum(residual)/2) + prior
#print(_lnprob)
return _lnprob
# Load I data
I = np.genfromtxt('data/OGLE.I.dat', dtype=['f8','f4','f4'], names=['tobs','mag','emag'])
imags = I["mag"]
eimags = I["emag"]
imjdobs = I["tobs"]+50000.
# Load V data
V = np.genfromtxt('data/OGLE.V.dat', dtype=['f8','f4','f4'], names=['tobs','mag','emag'])
vmags = V["mag"]
evmags = V["emag"]
vmjdobs = V["tobs"]+50000.
# Load Ks data
Ks = getdata("data/virac2_data.fits", -1, view=np.recarray)
Ks = Ks[(Ks["tileloc"]!=0) & (Ks["filter"].astype("U2")=="Ks")]
kmags = Ks["hfad_mag"]
ekmags = Ks["hfad_emag"]
kmjdobs = Ks["mjdobs"]
# estimated baseline mag
t0_est = int(imjdobs[np.argmax(imags)])
m0_i_est, em0_i_est = inv_var_weight_avg(imags[np.abs(imjdobs-t0_est)>365],
eimags[np.abs(imjdobs-t0_est)>365])
m0_v_est, em0_v_est = inv_var_weight_avg(vmags[np.abs(vmjdobs-t0_est)>365],
evmags[np.abs(vmjdobs-t0_est)>365])
m0_k_est, em0_k_est = inv_var_weight_avg(kmags[np.abs(kmjdobs-t0_est)>365],
ekmags[np.abs(kmjdobs-t0_est)>365])
# fluxes, errors
iflux, iflux_error = make_flux(imags, eimags, m0_i_est)
vflux, vflux_error = make_flux(vmags, evmags, m0_v_est)
kflux, kflux_error = make_flux(kmags, ekmags, m0_k_est)
# mjdobs, fluxes, flux errors to use for fitting
fit_mjdobs = np.concatenate((
#vmjdobs[np.abs(vmjdobs-t0_est)<365],
imjdobs[np.abs(imjdobs-t0_est)<200],
#kmjdobs[np.abs(kmjdobs-t0_est)<365],
))
fit_fluxes = np.concatenate((
#vflux[np.abs(vmjdobs-t0_est)<365],
iflux[np.abs(imjdobs-t0_est)<200],
#kflux[np.abs(kmjdobs-t0_est)<365],
))
fit_flux_errors = np.concatenate((
#vflux_error[np.abs(vmjdobs-t0_est)<365],
iflux_error[np.abs(imjdobs-t0_est)<200],
#kflux_error[np.abs(kmjdobs-t0_est)<365],
))
# get giant intensity map
I,xp,yp = dio.get_I(step=0.02) # step=0.02 means max model error is ~0.35 * data error
ndim, nwalkers = 7, 4096 #int(2**15)
p0 = np.random.rand(ndim * nwalkers).reshape((nwalkers, ndim))
nSteps = 1024 #int(2**12)
# run the sampler
with Pool(processes=nproc) as pool:
sampler = emcee.EnsembleSampler(nwalkers, ndim, lnprob, args=[I,xp,yp,fit_mjdobs,fit_fluxes,fit_flux_errors], pool=pool)
pos, prob, state = sampler.run_mcmc(p0,nSteps)
samples = sampler.chain.reshape((-1, ndim))
# save output
pickle.dump([pos,prob,state,samples], open("dio_fit_op.p", "wb"))