-
Notifications
You must be signed in to change notification settings - Fork 0
/
catalog_sampler.py
255 lines (222 loc) · 9.27 KB
/
catalog_sampler.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
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
from astropy.table import Table
import pandas as pd
import numpy as np
from mpi4py import MPI
import os
comm = MPI.COMM_WORLD
csize = comm.size
crank = comm.rank
cat_path = '/home/haoyanzhen_shao/project/simulation_work/data_dir/reference_catalog/CT1.5-N6397-GDR3-G26.0-rev_J2000.fits'
cat_raw = Table.read(cat_path)
cat_raw = cat_raw.to_pandas()
cat_raw = cat_raw[['RA','Dec','grav','teff','feh','app_sdss_g','pmra','pmdec','parallax','RV']]
# name_selection = ['Dist','grav','teff','feh','app_Gaia_G']
name_all = ['grav','teff','feh','app_sdss_g','pmra','pmdec','parallax','RV']
cat_raw_desc = cat_raw[name_all].describe()
stamp_nstar = '5e6'
nstar = int(5e6)
nstar = int(nstar/csize)
ra = np.random.rand(nstar) * 1.4 + 270.3
dec = np.random.rand(nstar) * 1.4 - 41.7
_hist,_edges = np.histogram(cat_raw['RV'],bins=90,range=(-280,440))
np.save('/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/histgram_rv.npy',np.array([_hist,_edges],dtype=object))
_hist = _hist/len(cat_raw['RV'])
_nedges = _edges + (_edges[1]-_edges[0])/2
_nedges = _nedges[:-1]
from scipy.interpolate import CubicSpline
interp = CubicSpline(_nedges,_hist)
a = np.random.rand(nstar)*(438+279) - 279
p = interp(a)
p[p<0] = 0
p = p/np.sum(p)
rv = np.random.choice(a,size=nstar,p=p)
del _hist,_edges,interp,_nedges,a,p
def stat(df):
df_selection_len = len(df.columns)
df_len = len(df)
df_desc = df.describe()
bins = np.array([np.arange(-1.6,0.9,0.1)**3+4.8,
np.arange(2000,8700,670),
np.arange(-3.72,0.54,0.426),
np.arange(-2.7,1.4,0.1)**3+25],dtype=object) # 5-27.8
range_ = np.array([df_desc.min()[:],df_desc.iloc[3:].max()[:]]).T
df_hist,_edges = np.histogramdd(df.to_numpy(),bins,range_)
np.save('/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/histgram_phy.npy',np.array([df_hist,_edges],dtype=object))
# hist调整
_lens = []
for i in range(4):
_lens.append(_edges[i][1:] - _edges[i][:-1])
_edges[i] = (_edges[i][:-1]+_edges[i][1:])/2
df_hist = df_hist/_lens[0][:,None,None,None]/_lens[1][None,:,None,None]/_lens[2][None,None,:,None]/_lens[3][None,None,None,:]
df_hist = df_hist/df_hist.sum()
# 概率分布函数插值
from scipy.interpolate import RegularGridInterpolator
interp = RegularGridInterpolator(_edges,df_hist,method='cubic')
return interp, _edges
# def stat2(df):
# df_len = len(df)
# df_desc = df.describe()
# bins = np.array([np.arange(-8,8,0.5)**3,
# np.arange(-8,8,0.5)**3,
# np.float_power(10,np.arange(-8,2,0.25))],dtype=object)
# range_ = np.array([df_desc.min()[:],df_desc.iloc[3:].max()[:]]).T
# df_hist,_edges = np.histogramdd(df.to_numpy(),bins,range_)
# np.save('/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/histgram_ast.npy',np.array([df_hist,_edges],dtype=object))
# # hist调整
# _lens = []
# for i in range(3):
# _lens.append(_edges[i][1:] - _edges[i][:-1])
# # _edges[i] = (_edges[i][:-1]+_edges[i][1:])/2
# _edges[i] = _edges[i][:-1]
# df_hist = df_hist/_lens[0][:,None,None]/_lens[1][None,:,None]/_lens[2][None,None,:]
# df_hist = df_hist/df_hist.sum()
# # 概率分布函数插值
# from scipy.interpolate import RegularGridInterpolator
# interp = RegularGridInterpolator(_edges,df_hist,method='nearest')
# return interp, _edges
name_groups = []
name_groups.append(name_all[:4])
name_groups.append(name_all[4:-1])
bins = np.array([40,20,20,100,20,20,20])
interp_phy, _edge_phy = stat(cat_raw[name_groups[0]])
# interp_ast, _edge_ast = stat2(cat_raw[name_groups[1]])
def starcount_phy_log(x):
ndim = 4
try:
p = interp_phy(x)
if p <= 0:
p = 1e-20
return np.log10(p)
except Exception:
p = 0
for i in range(ndim):
if x[i] > _edge_phy[i][-1]:
p += -30 - 30*(x[i]-_edge_phy[i][-1])/(_edge_phy[i][-1]-_edge_phy[i][0])
elif x[i] < _edge_phy[i][0]:
p += -30 - 30*(_edge_phy[i][0]-x[i])/(_edge_phy[i][-1]-_edge_phy[i][0])
if p > -10:
p -= 20
return p
# def starcount_ast_log(x):
# ndim = 3
# try:
# p = interp_ast(x)
# if p <= 0:
# p = 1e-20
# return np.log10(p)
# except Exception:
# p = 0
# for i in range(ndim):
# if x[i] > _edge_ast[i][-1]:
# p += -30 - 100*(x[i]-_edge_ast[i][-1])/(_edge_ast[i][-1]-_edge_ast[i][0])
# elif x[i] < _edge_ast[i][0]:
# p += -30 - 100*(_edge_ast[i][0]-x[i])/(_edge_ast[i][-1]-_edge_ast[i][0])
# if p > -20:
# p -= 20
# return p
from emcee import EnsembleSampler
nwalkers = 100
ndim = 4
psteps = 100
nsteps = int(nstar/nwalkers)
# x[grav,teff,feh,mag_g]
# [ 3.76236096e-01, 5.35259056e+00],
# [ 6.57588529e+02, 8.69925293e+03],
# [-3.72083282e+00, 5.36125898e-01],
# [ 1.68997382e+00, 2.74582329e+01]
x0 = np.empty((nwalkers,ndim))
x0[:,0] = np.random.rand(nwalkers) * 1.6 + 3.7
x0[:,1] = np.random.rand(nwalkers) * 8000 + 660
x0[:,2] = np.random.rand(nwalkers) * 4 - 3.7
x0[:,3] = np.random.rand(nwalkers) * 25 + 2
sampler = EnsembleSampler(nwalkers,ndim,starcount_phy_log)
x = sampler.run_mcmc(x0, psteps, progress=True)
sampler.reset()
sampler.run_mcmc(x, nsteps, progress=True)
samples_phy = sampler.get_chain().reshape((-1,ndim))
# nwalkers = 100
# ndim = 3
# psteps = 1000
# nsteps = int(nstar/nwalkers)
# # x[mag_g,pmra,pmdec,parallax]
# # [ 5.21828318e+00, 2.74582329e+01],
# # [-3.84212126e+02, 1.68569655e+02],
# # [-4.81517687e+02, 1.01266962e+02],
# # [ 8.70646100e-04, 7.40291565e+01]
# x0 = np.empty((nwalkers,ndim))
# x0[:,0] = np.random.rand(nwalkers) * 500 - 380
# x0[:,1] = np.random.rand(nwalkers) * 550 - 480
# x0[:,2] = np.random.rand(nwalkers) * 70 + 1
# # 不确定mag对pm和pa产生影响的原因，可能是基于星系动力学，由于采样较难，目前假设为均匀分布情况下无影响。
# # x_mag = samples_phy[:,-1][:,None]
# sampler = EnsembleSampler(nwalkers,ndim,starcount_ast_log)
# sampler.reset()
# # samplel = []
# # for i in range(nsamples):
# # x0 = sampler.sample(x0)
# # samplel.append(x0)
# x = sampler.run_mcmc(x0, psteps, progress=True)
# sampler.reset()
# sampler.run_mcmc(x, nsteps, progress=True)
# samples_ast = sampler.get_chain().reshape((-1,ndim))
grav = samples_phy[:,0].copy()
teff = samples_phy[:,1].copy()
feh = samples_phy[:,2].copy()
gaia_g = samples_phy[:,3].copy()
# pmra = samples_ast[:,0]
# pmdec = samples_ast[:,1]
# parallax = samples_ast[:,2]
del interp_phy,_edge_phy,sampler,samples_phy
from scipy.optimize import curve_fit
def norm(x,mu,var):
return 1/((2*np.pi)**0.5 * var) * np.exp(-(x-mu)**2/var**2)
x = cat_raw['pmra']
bins = np.arange(-400,400,0.01)
_hist, _edge = np.histogram(x,bins)
# popt,pcov,infodict,mesg,flg = curve_fit(norm,_edge[:-1],_hist)
popt,pcov = curve_fit(norm,_edge[:-1],_hist)
pmra = np.random.normal(loc=popt[0],scale=popt[1],size=nstar)
x = cat_raw['pmdec']
_hist, _edge = np.histogram(x,bins)
popt,pcov = curve_fit(norm,_edge[:-1],_hist)
pmdec = np.random.normal(loc=popt[0],scale=popt[1],size=nstar)
def exp(x,k,e):
return k * 2 ** (-e*x)
x = cat_raw['parallax']
bins = np.arange(0,40,0.01)
_hist, _edge = np.histogram(x,bins)
popt,pcov,info,mesg,flg= curve_fit(exp,_edge[:-1],_hist,full_output=True)
a = np.random.rand(nstar)*10
p = exp(a,popt[0],popt[1])
p /= p.sum()
parallax = np.random.choice(a,size=nstar,p=p)
del a,p
cat_len = nstar
for _data in [ra,dec,gaia_g,teff,grav,feh,pmra,pmdec,rv,parallax]:
cat_len = (len(_data) if len(_data)<cat_len else cat_len)
for _data in [ra,dec,gaia_g,teff,grav,feh,pmra,pmdec,rv,parallax]:
_data = _data[:cat_len]
cat_new = Table([ra,dec,gaia_g,teff,grav,feh,pmra,pmdec,rv,parallax],
names=('RA','Dec','app_sdss_g','teff','grav','feh','pmra','pmdec','RV','parallax'))
cat_new.write('/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/from_N6397_%s_%s.fits'%(stamp_nstar,crank), format='fits',overwrite=True)
ssss=0
comm.gather(ssss,root=0)
if crank != 0:
pass
else:
from astropy.table import vstack
for i in range(csize-1):
i+=1
cati = Table.read('/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/from_N6397_%s_%s.fits'%(stamp_nstar,i))
cat_new = vstack((cat_new,cati))
del cati
cat_new = cat_new.to_pandas()
cat_new = cat_new.drop(cat_new[(cat_new['grav']<1.3)|(cat_new['grav']>5.5)|
(cat_new['teff']<2000)|(cat_new['teff']>10000)|
(cat_new['feh']<-4)|(cat_new['feh']>0.5)].index)
cat_new = cat_new.dropna(axis=0)
cat_new = Table.from_pandas(cat_new)
cat_new.write('/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/from_N6397_%s.fits'%stamp_nstar, format='fits',overwrite=True)
from self_test.cat2hdf5 import fits_to_hdf5 as f2h
f2h("/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/from_N6397_%s.fits"%stamp_nstar,
"/home/haoyanzhen_shao/project/simulation_work/data_dir/catalog/crowdedField_sampler0627/from_N6397_%s.hdf5"%stamp_nstar)