import os, sys
import pickle
from dynesty import utils as dyutils
from allesfitter import config
from astropy.time import Time
import matplotlib.pyplot as plt
# base data path for the TESS known planet project
pathdata = os.environ['KNWN_DATA_PATH'] + '/'
pathdatapost = os.environ['KNWN_DATA_PATH'] + '/postproc/'
# name of the exoplanet
strgplan = sys.argv[1]
pathdataplan = pathdata + '%s/allesfit_global/allesfit_onlytess_full/' % strgplan
config.init(pathdataplan)
fileobjt = open(pathdataplan + 'results/save_ns.pickle', 'rb')
objtrest = pickle.load(fileobjt)
weig = np.exp(objtrest['logwt'] - objtrest['logz'][-1])
chan = dyutils.resample_equal(objtrest.samples, weig)
listkeys = config.BASEMENT.fitkeys
listlabl = config.BASEMENT.fitlabels
# get period and epoch posteriors
for k, labl in enumerate(listlabl):
if listkeys[k] == 'b_epoch':
postepoc = chan[:, k]
if listkeys[k] == 'b_period':
postperi = chan[:, k]
def plot_viol(pathdataoutp, pvalthrs=1e-3, boolonlytess=False):
strgtimestmp = datetime.datetime.now().strftime('%Y%m%d_%H%M%S')
print('allesfitter postprocessing violin plot started at %s...' %
strgtimestmp)
liststrgruns = ['woutTESS', 'alldata']
if boolonlytess:
liststrgruns.append(['TESS'])
numbruns = len(liststrgruns)
indxruns = np.arange(numbruns)
liststrgpara = [[] for i in indxruns]
listlablpara = [[] for i in indxruns]
listobjtalle = [[] for i in indxruns]
for i, strgruns in enumerate(liststrgruns):
pathdata = pathdataoutp + 'allesfits/allesfit_%s' % strgruns
print('Reading from %s...' % pathdata)
config.init(pathdata)
liststrgpara[i] = np.array(config.BASEMENT.fitkeys)
listlablpara[i] = np.array(config.BASEMENT.fitlabels)
# read the chain
listobjtalle[i] = allesfitter.allesclass(pathdata)
liststrgparaconc = np.concatenate(liststrgpara)
liststrgparaconc = np.unique(liststrgparaconc)
listlablparaconc = np.copy(liststrgparaconc)
for k, strgparaconc in enumerate(liststrgparaconc):
for i, strgruns in enumerate(liststrgruns):
if strgparaconc in liststrgpara[i]:
listlablparaconc[k] = listlablpara[i][np.where(
liststrgpara[i] == strgparaconc)[0][0]]
ticklabl = ['w/o TESS', 'w/ TESS']
if boolonlytess:
ticklabl.append(['only TESS'])
xpos = 0.6 * (np.arange(numbruns) + 1.)
for k, strgpara in enumerate(liststrgparaconc):
booltemp = True
for i in indxruns:
if not strgpara in liststrgpara[i]:
booltemp = False
if not booltemp:
continue
figr, axis = plt.subplots(figsize=(5, 4))
chanlist = []
for i in indxruns:
chanlist.append(
(listobjtalle[i].posterior_params[strgpara] -
np.mean(listobjtalle[i].posterior_params[strgpara])) * 24. *
60.)
axis.violinplot(chanlist, xpos, showmedians=True, showextrema=False)
axis.set_xticks(xpos)
axis.set_xticklabels(ticklabl)
if strgpara == 'b_period':
axis.set_ylabel('P [min]')
else:
axis.set_ylabel(listlablparaconc[k])
plt.tight_layout()
path = pathdataoutp + 'viol_%04d.svg' % (k)
print('Writing to %s...' % path)
figr.savefig(path)
plt.close()
listyear = [2021, 2023, 2025]
numbyear = len(listyear)
indxyear = np.arange(numbyear)
timejwst = [[[] for i in indxruns] for k in indxyear]
for k, year in enumerate(listyear):
epocjwst = astropy.time.Time('%d-01-01 00:00:00' % year,
format='iso').jd
for i in indxruns:
epoc = listobjtalle[i].posterior_params['b_epoch']
peri = listobjtalle[i].posterior_params['b_period']
indxtran = (epocjwst - epoc) / peri
indxtran = np.mean(np.rint(indxtran))
if indxtran.size != np.unique(indxtran).size:
raise Exception('')
timejwst[k][i] = epoc + peri * indxtran
timejwst[k][i] -= np.mean(timejwst[k][i])
timejwst[k][i] *= 24. * 60.
listfigr = []
listaxis = []
## temporal evolution
figr, axis = plt.subplots(figsize=(5, 4))
listfigr.append(figr)
listaxis.append(axis)
axis.violinplot([timejwst[k][1] for k in indxyear], listyear)
axis.set_xlabel('Year')
axis.set_ylabel('Transit time residual [min]')
plt.tight_layout()
path = pathdataoutp + 'jwsttime.svg'
print('Writing to %s...' % path)
plt.savefig(path)
plt.close()
## without/with/only TESS prediction comparison
figr, axis = plt.subplots(figsize=(5, 4))
listfigr.append(figr)
listaxis.append(axis)
axis.violinplot(timejwst[1], xpos, points=2000)
axis.set_xticks(xpos)
axis.set_xticklabels(ticklabl)
axis.set_ylabel('Transit time residual in 2023 [min]')
#axis.set_ylim([-300, 300])
plt.tight_layout()
path = pathdataoutp + 'jwstcomp.svg'
print('Writing to %s...' % path)
plt.savefig(path)
plt.close()
return listfigr, listaxis
# all parameter summary
figr, axis = plt.subplots(figsize=(4, 3))
chanlist = []
axis.violinplot(chanlist, xpos, showmedians=True, showextrema=False)
axis.set_xticks(valutick)
axis.set_xticklabels(labltick)
axis.set_ylabel(lablparatemp)
plt.tight_layout()
path = pathdataoutp + 'para_%s.pdf'
print('Writing to %s...' % path)
figr.savefig(path)
plt.close()
# plot p values
## threshold p value to conclude significant difference between posteriors with and without TESS
if pvalthrs is None:
pvalthrs = 1e-6
lablparacomp = [[] for u in indxruns]
for u in indxruns:
lablparacomp[u] = list(
set(lablpara[indxrunsfrst[u]]).intersection(
lablpara[indxrunsseco[u]]))
# post-processing
## calculate the KS test statistic between the posteriors
numbparacomp = len(lablparacomp[u])
pval = np.empty(numbparacomp)
for j in range(numbparacomp):
kosm, pval[j] = scipy.stats.ks_2samp([indxrunsfrst[u]][:, j],
chan[indxrunsseco[u]][:, j])
kosm, pval[j] = scipy.stats.ks_2samp(chan[indxrunsfrst[u]][:, j],
chan[indxrunsseco[u]][:, j])
## find the list of parameters whose posterior with and without TESS are unlikely to be drawn from the same distribution
indxparagood = np.where(pval < pvalthrs)[0]
if indxparagood.size > 0:
figr, axis = plt.subplots(figsize=(12, 5))
indxparacomp = np.arange(numbparacomp)
axis.plot(indxparacomp, pval, ls='', marker='o')
axis.plot(indxparacomp[indxparagood],
pval[indxparagood],
ls='',
marker='o',
color='r')
axis.set_yscale('log')
axis.set_xticks(indxparacomp)
axis.set_xticklabels(lablparacomp[u])
if u == 0:
axis.set_title('Posteriors with TESS vs. without TESS')
if u == 1:
axis.set_title('Posteriors without TESS vs. only TESS')
if u == 2:
axis.set_title('Posteriors with TESS vs. only TESS')
axis.axhline(pvalthrs, ls='--', color='black', alpha=0.3)
plt.tight_layout()
path = pathdataoutp + 'kosm_com%d.pdf' % u
print('Writing to %s...' % path)
figr.savefig(path)
plt.close()
chan = [[] for i in indxrtyp]
lablpara = [[] for i in indxrtyp]
numbwalk = [[] for i in indxrtyp]
numbswep = [[] for i in indxrtyp]
numbsamp = [[] for i in indxrtyp]
numbburn = [[] for i in indxrtyp]
factthin = [[] for i in indxrtyp]
numbpara = [[] for i in indxrtyp]
pathdataplan = pathdata + strgplan + '/'
for i, strgrtyp in enumerate(liststrgrtyp):
pathdata = pathdataplan + 'allesfit_%s/allesfit_%stess_ns/' % (strgkind,
strgrtyp)
print('Reading from %s...' % pathdata)
config.init(pathdata)
lablpara[i] = config.BASEMENT.fitlabels
pathsave = pathdata + 'results/mcmc_save.h5'
if False and not os.path.exists(pathsave):
# sample from the posterior excluding the TESS data
print('Calling allesfitter to fit the data...')
allesfitter.mcmc_fit(pathdata)
# read the chain
## MCMC
#emceobjt = emcee.backends.HDFBackend(pathsave, read_only=True)
#chan[i] = emceobjt.get_chain()
#lpos[i] = emceobjt.get_log_prob()
## Nested sampling
def plot_publication_spots_from_posteriors(datadir,
Nsamples=20,
command='save',
mode='default'):
'''
command : str
'show', 'save', 'return', 'show and return', 'save and return'
mode: str
default : 5000 points, phase (-0.25,0.75), errorbars
zhan2019 : 100 points, phase (0,2), no errorbars
'''
fig, ax1, ax2, ax3 = setup_grid()
config.init(datadir)
posterior_samples = allesfitter.get_ns_posterior_samples(
datadir, Nsamples=Nsamples, as_type='2d_array')
for inst in config.BASEMENT.settings['inst_all']:
if config.BASEMENT.settings['host_N_spots_' + inst] > 0:
if mode == 'default':
xx = np.linspace(config.BASEMENT.data[inst]['time'][0],
config.BASEMENT.data[inst]['time'][-1], 5000)
elif mode == 'zhan2019':
xx = np.linspace(0, 2, 10000)
for i_sample, sample in tqdm(enumerate(posterior_samples)):
params = allesfitter.computer.update_params(sample)
spots = [[
params['host_spot_' + str(i) + '_long_' + inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' + inst] +
1)
],
[
params['host_spot_' + str(i) + '_lat_' + inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' +
inst] + 1)
],
[
params['host_spot_' + str(i) + '_size_' + inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' +
inst] + 1)
],
[
params['host_spot_' + str(i) + '_brightness_' +
inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' +
inst] + 1)
]]
model = allesfitter.computer.calculate_model(
params, inst, 'flux')
baseline = allesfitter.computer.calculate_baseline(
params, inst, 'flux')
model_xx = allesfitter.computer.calculate_model(
params, inst, 'flux', xx=xx) #evaluated on xx (!)
baseline_xx = allesfitter.computer.calculate_baseline(
params, inst, 'flux', xx=xx) #evaluated on xx (!)
if i_sample == 0:
if mode == 'default':
ax1 = axplot_data(ax1,
config.BASEMENT.data[inst]['time'],
config.BASEMENT.data[inst]['flux'],
flux_err=np.exp(
params['log_err_flux_' + inst]))
ax2 = axplot_residuals(
ax2,
config.BASEMENT.data[inst]['time'],
config.BASEMENT.data[inst]['flux'] - model -
baseline,
res_err=np.exp(params['log_err_flux_' + inst]))
ax3 = axplot_spots_2d(ax3, spots)
elif mode == 'zhan2019':
ax1 = axplot_data(
ax1,
np.concatenate(
(config.BASEMENT.data[inst]['time'],
config.BASEMENT.data[inst]['time'] + 1,
config.BASEMENT.data[inst]['time'] + 2)),
np.concatenate(
(config.BASEMENT.data[inst]['flux'],
config.BASEMENT.data[inst]['flux'],
config.BASEMENT.data[inst]['flux'])),
flux_err=None)
ax2 = axplot_residuals(
ax2,
np.concatenate(
(config.BASEMENT.data[inst]['time'],
config.BASEMENT.data[inst]['time'] + 1,
config.BASEMENT.data[inst]['time'] + 2)),
np.concatenate(
(config.BASEMENT.data[inst]['flux'] - model -
baseline, config.BASEMENT.data[inst]['flux'] -
model - baseline,
config.BASEMENT.data[inst]['flux'] - model -
baseline)),
res_err=None)
ax3 = axplot_spots_2d(ax3, spots)
ax1 = axplot_model(ax1, xx, model_xx + baseline_xx)
ax1.locator_params(axis='y', nbins=5)
if mode == 'zhan2019':
ax1.set(xlim=[0, 2])
ax2.set(xlim=[0, 2])
if 'save' in command:
pubdir = os.path.join(config.BASEMENT.outdir, 'pub')
if not os.path.exists(pubdir): os.makedirs(pubdir)
if mode == 'default':
fig.savefig(os.path.join(pubdir,
'host_spots_' + inst + '.pdf'),
bbox_inches='tight')
elif mode == 'zhan2019':
fig.savefig(os.path.join(pubdir,
'host_spots_' + inst + '_zz.pdf'),
bbox_inches='tight')
plt.close(fig)
if 'show' in command:
plt.show()
if 'return' in command:
return fig, ax1, ax2, ax3
def plot_spots_from_posteriors(datadir, Nsamples=10, command='return'):
if command == 'show':
Nsamples = 1 #overwrite user input and only show 1 sample if command=='show'
config.init(datadir)
posterior_samples_dic = allesfitter.get_ns_posterior_samples(
datadir, Nsamples=Nsamples)
for sample in tqdm(range(Nsamples)):
params = {}
for key in posterior_samples_dic:
params[key] = posterior_samples_dic[key][sample]
for inst in config.BASEMENT.settings['inst_all']:
if config.BASEMENT.settings['host_N_spots_' + inst] > 0:
spots = [[
params['host_spot_' + str(i) + '_long_' + inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' + inst] +
1)
],
[
params['host_spot_' + str(i) + '_lat_' + inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' +
inst] + 1)
],
[
params['host_spot_' + str(i) + '_size_' + inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' +
inst] + 1)
],
[
params['host_spot_' + str(i) + '_brightness_' +
inst]
for i in range(
1, config.BASEMENT.settings['host_N_spots_' +
inst] + 1)
]]
if command == 'return':
fig, ax, ax2 = plot_spots(spots, command='return')
plt.suptitle('sample ' + str(sample))
spotsdir = os.path.join(config.BASEMENT.outdir, 'spotmaps')
if not os.path.exists(spotsdir): os.makedirs(spotsdir)
fig.savefig(
os.path.join(
spotsdir, 'host_spots_' + inst +
'_posterior_sample_' + str(sample)))
plt.close(fig)
elif command == 'show':
plot_spots(spots, command='show')
for companion in config.BASEMENT.settings['companions_all']:
for inst in config.BASEMENT.settings['inst_all']:
if config.BASEMENT.settings[companion + '_N_spots_' +
inst] > 0:
spots = [[
params[companion + '_spot_' + str(i) + '_long_' + inst]
for i in range(
1, config.BASEMENT.settings[companion +
'_N_spots_' + inst] +
1)
],
[
params[companion + '_spot_' + str(i) +
'_lat_' + inst]
for i in range(
1, config.BASEMENT.settings[companion +
'_N_spots_' +
inst] + 1)
],
[
params[companion + '_spot_' + str(i) +
'_size_' + inst]
for i in range(
1, config.BASEMENT.settings[companion +
'_N_spots_' +
inst] + 1)
],
[
params[companion + '_spot_' + str(i) +
'_brightness_' + inst]
for i in range(
1, config.BASEMENT.settings[companion +
'_N_spots_' +
inst] + 1)
]]
if command == 'return':
fig, ax, ax2 = plot_spots(spots, command='return')
plt.suptitle('sample ' + str(sample))
spotsdir = os.path.join(config.BASEMENT.outdir,
'spotmaps')
if not os.path.exists(spotsdir): os.makedirs(spotsdir)
fig.savefig(
os.path.join(
spotsdir, companion + '_spots_' + inst +
'_posterior_sample_' + str(sample)))
plt.close(fig)
elif command == 'show':
plot_spots(spots, command='show')
def plot_viol(pathbase, liststrgstar, liststrgruns, lablstrgruns, pathimag, pvalthrs=1e-3):
strgtimestmp = datetime.datetime.now().strftime('%Y%m%d_%H%M%S')
print('allesfitter postprocessing violin plot started at %s...' % strgtimestmp)
# construct global object
gdat = gdatstrt()
# copy unnamed inputs to the global object
#for attr, valu in locals().iter():
for attr, valu in locals().items():
if '__' not in attr and attr != 'gdat':
setattr(gdat, attr, valu)
# runs to be compared for each star
gdat.numbruns = len(liststrgruns)
gdat.indxruns = np.arange(gdat.numbruns)
gdat.pathimag = pathimag
gdat.liststrgstar = liststrgstar
# stars
numbstar = len(liststrgstar)
gdat.indxstar = np.arange(numbstar)
# plotting
gdat.strgplotextn = 'png'
# read parameter keys, labels and posterior from allesfitter output
liststrgpara = [[] for i in gdat.indxruns]
listlablpara = [[] for i in gdat.indxruns]
gdat.listobjtalle = [[[] for m in gdat.indxstar] for i in gdat.indxruns]
for i in gdat.indxruns:
for m in gdat.indxstar:
pathalle = pathbase + '%s/allesfits/allesfit_%s/' % (gdat.liststrgstar[m], gdat.liststrgruns[i])
print('Reading from %s...' % pathalle)
config.init(pathalle)
liststrgpara[i] = np.array(config.BASEMENT.fitkeys)
listlablpara[i] = np.array(config.BASEMENT.fitlabels)
# read the chain
print('pathalle')
print(pathalle)
gdat.listobjtalle[i][m] = allesfitter.allesclass(pathalle)
# concatenate the keys, labels from different runs
gdat.liststrgparaconc = np.concatenate(liststrgpara)
gdat.liststrgparaconc = np.unique(gdat.liststrgparaconc)
gdat.listlablparaconc = np.copy(gdat.liststrgparaconc)
for k, strgparaconc in enumerate(gdat.liststrgparaconc):
for i, strgruns in enumerate(liststrgruns):
if strgparaconc in liststrgpara[i]:
gdat.listlablparaconc[k] = listlablpara[i][np.where(liststrgpara[i] == strgparaconc)[0][0]]
gdat.numbparaconc = len(gdat.liststrgparaconc)
gdat.indxparaconc = np.arange(gdat.numbparaconc)
for k, strgpara in enumerate(gdat.liststrgparaconc):
booltemp = True
for i in gdat.indxruns:
if not strgpara in liststrgpara[i]:
booltemp = False
if not booltemp:
continue
## violin plot
## mid-transit time prediction
plot(gdat, gdat.indxstar, indxpara=np.array([k]), strgtype='paracomp')
## per-star
#for m in gdat.indxstar:
# plot(gdat, indxstar=np.array([m]), indxpara=k, strgtype='paracomp')
# calculate the future evolution of epoch
gdat.listyear = [2021, 2023, 2025]
numbyear = len(gdat.listyear)
gdat.indxyear = np.arange(numbyear)
gdat.timejwst = [[[[] for m in gdat.indxstar] for i in gdat.indxruns] for k in gdat.indxyear]
for k, year in enumerate(gdat.listyear):
epocjwst = astropy.time.Time('%d-01-01 00:00:00' % year, format='iso').jd
for i in gdat.indxruns:
for m in gdat.indxstar:
epoc = gdat.listobjtalle[i][m].posterior_params['b_epoch']
peri = gdat.listobjtalle[i][m].posterior_params['b_period']
indxtran = (epocjwst - epoc) / peri
indxtran = np.mean(np.rint(indxtran))
if indxtran.size != np.unique(indxtran).size:
raise Exception('')
gdat.timejwst[k][i][m] = epoc + peri * indxtran
gdat.timejwst[k][i][m] -= np.mean(gdat.timejwst[k][i][m])
gdat.timejwst[k][i][m] *= 24. * 60.
listfigr = []
listaxis = []
# temporal evolution of mid-transit time prediction
plot(gdat, gdat.indxstar, strgtype='epocevol')
## per-star
#for m in gdat.indxstar:
# plot(gdat, indxstar=np.array([m]), strgtype='epocevol')
## mid-transit time prediction
plot(gdat, gdat.indxstar, strgtype='jwstcomp')
## per-star
#for m in gdat.indxstar:
# plot(gdat, indxstar=np.array([m]), strgtype='jwstcomp')
return listfigr, listaxis