def rsync_ARTEMiS_alerts():
"""Syncrhonize alerts with ARTEMiS Signalmen
Returns:
--------
0 : download succeeded,
2 : download failed.
"""
# set I/O shell display
tcol, tun, tend, tit = "\033[0m\033[34m", "\033[0m\033[1m\033[34m", "\033[0m", "\033[0m\033[3m"
# rsync -azu [email protected]::FollowUp/FollowUpZ.signalmen .
printi(tcol + "Rsync ARTEMiS Signalmen" + tend)
# password in file
pswd = 'pswd_signalmen.txt'
# archive format: XKB180039I, with X=dataset, I=filter
proc = subprocess.Popen('rsync -az --password-file=' + pswd + ' [email protected]::FollowUp/FollowUpZ.signalmen ARTEMiS_alerts.txt', shell=True, executable='/bin/bash', stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
proc.wait()
# test whether data were downloaded, and if not treat as exception
stdout, stderr = proc.communicate()
if stderr:
printw(tun + "Failed to rsync ARTEMiS Signalmen\n" + tit + stderr + tend)
return 1
else:
printd(tit + " (list of alerts downloaded from ATERMiS Signalmen)" + tend)
return 0
def fill_webpage(inhtml, outhtml, fillwith):
"""Fill the template html page with new html content
Example
-------
>>> fillwpage('in.html', 'out.html', ('_keyword_', var))
>>> fillwpage('in.html', 'out.html', [('_keyword1_', var1), ('_keyword2_', var2)])
"""
# set I/O shell display
tcol, tun, tend, tit = "\033[0m\033[33m", "\033[0m\033[1m\033[33m", "\033[0m", "\033[0m\033[3m"
# important - add these lines between <header> tags --> mettre dans le code event.html
"""<link rel="stylesheet" href="http://cdn.pydata.org/bokeh/release/bokeh-0.12.9.min.css" type="text/css" /><script type="text/javascript" src="http://cdn.pydata.org/bokeh/release/bokeh-0.12.9.min.js"></script><!-- _SCRIPT_ -->"""
# fill html page
with open(inhtml, 'r') as f:
cont = f.read()
if type(fillwith) == tuple:
fillwith = [fillwith]
for fill in fillwith:
key, new = fill
cont = cont.replace(key, new)
f.close()
with open(outhtml, 'w') as f:
f.write(cont)
f.close()
# verbose
printi(tcol + "Fill document : " + tit + "'" + inhtml + "'" + tcol + " >>> " + tit + "'" + outhtml + "'" + tend)
for fill in fillwith:
printd(tit + " (keyword: " + fill[0] + ")" + tend)
def fit_subgrids(gridsprefix,
fitsprefix,
datasets,
gridlist,
init=None,
trange=None,
nprocs=1,
overwrite=False):
"""Fit light curve on magnification curves grids
IDEE : on pourra utiliser un random des 1000 pour faire
des maps a faible resolution !
"""
# set I/O shell display
tcol, tit, tend = "\033[0m\033[31m", "\033[0m\033[3m", "\033[0m"
# check whether input names does not contain extensions
if '.hdf5' in gridsprefix:
raise NameError("grid prefix should not contain .hdf5 extension")
if '.hdf5' in fitsprefix:
raise NameError("fit prefix should not contain .hdf5 extension")
# delete existing HDF5 files in fits/
if overwrite:
printd(tcol + "Removing previous HDF5 files from fits/" + tend)
proc = subprocess.Popen('rm -rf ' + fitsprefix + '*.hdf5',
shell=True,
executable='/bin/bash')
proc.wait()
# mutliprocessing: create grid list names
listmclibs, listlclibs = list(), list()
for gridi in gridlist:
mclib = gridsprefix + '_' + str(gridi) + '.hdf5'
lclib = fitsprefix + '_' + str(gridi) + '.hdf5'
listmclibs.append(mclib)
listlclibs.append(lclib)
# mutliprocessing: create arguments of _process_fits, and create workers pool
printi(tcol + "Starting manager with PID " + tit + str(os.getpid()) +
tcol + " running " + tit + str(nprocs) + tcol + " process(es)" +
tend)
listargs = zip(listmclibs, listlclibs, repeat(datasets), repeat(trange),
repeat(init))
pool = Pool(processes=nprocs)
pool.imap_unordered(_process_fits, listargs)
# collect results
pool.close()
pool.join()
def rsync_ARTEMiS_data(event, syncdir):
"""Syncrhonize data with ARTEMiS
Returns:
--------
0 : synchronization succeeded and
local data were updated,
1 : synchronization succeeded but
local data were already up to date,
2 : synchronization failed.
"""
# set I/O shell display
tcol, tun, tend, tit = "\033[0m\033[34m", "\033[0m\033[1m\033[34m", "\033[0m", "\033[0m\033[3m"
# check event name format
if len(event) != 8:
raise ValueError("wrong format for event name: correct is 'xByyzzzz'")
# get microlensing season
year = '20' + event[2:4]
# rsync -azu [email protected]::Data2018/OOB180088I.dat .
printi(tcol + "Rsync ARTEMiS database" + tend)
# password in file
pswd = 'pswd_artemis.txt'
# archive format: XKB180039I, with X=dataset, I=filter
artemisevent = event.replace('M', 'K')
proc = subprocess.Popen('rsync -avzu -L --password-file=' + pswd + ' [email protected]::Data' + year + '/*' + artemisevent + '*.dat ' + syncdir + ' --stats | grep "files transferred:"', shell=True, executable='/bin/bash', stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
proc.wait()
# get std output and error
stdout, stderr = proc.communicate()
# treat rsync outputs
if stderr:
printw(tun + "Failed to rsync ARTEMiS\n" + tit + stderr + tend)
return 2
else:
# check if files were updated
if ' 0' in stdout:
printd(tit + " (files have not changed on ARTEMiS server)" + tend)
return 1
else:
printd(tit + " (local files have been updated from ATERMiS server)" + tend)
return 0
def retrieve(event, crossrefs=None, download=True):
"""Create or update microlensing event data files
Calling
=======
retrieve(event)
Parameters
----------
event : string
Name of microlensing event: 'xByyzzzz', where:
x is O (OGLE alert) or B (MOA alert)
yy is the year
zzzz is the event ID
Example
-------
>>> datasets = retrieve('OB180222')
"""
# set I/O shell display
tcol, tun, tend, tit = "\033[0m\033[34m", "\033[0m\033[1m\033[34m", "\033[0m", "\033[0m\033[3m"
# check event name format
if len(event) != 8:
raise ValueError("wrong format for event name: correct is 'xByyzzzz'")
# get microlensing season
year = '20' + event[2:4]
# directories arborescence
datapath = event + '/data/'
syncdir = datapath + 'sync/'
fitsdir = event + '/fits/'
if not os.path.isdir(event):
# create dir: ./event/
printi(tcol + "Create new event folder " + tit + event + "/" + tend)
proc = subprocess.Popen('mkdir ' + event, shell=True, executable='/bin/bash')
proc.wait()
# else:
# printi(tcol + "Update event " + tit + event + tend)
if not os.path.isdir(datapath):
# create dir: ./event/data/
proc = subprocess.Popen('mkdir ' + datapath, shell=True, executable='/bin/bash')
proc.wait()
if not os.path.isdir(syncdir):
# create dir: ./event/data/sync/
proc = subprocess.Popen('mkdir ' + syncdir, shell=True, executable='/bin/bash')
proc.wait()
if not os.path.isdir(fitsdir):
# create dir: ./event/fits/
proc = subprocess.Popen('mkdir ' + fitsdir, shell=True, executable='/bin/bash')
proc.wait()
# check whether to get new data, loca data or stop
if download:
# if crossfrefs file is given
newkmt = []
if crossrefs:
newsync = 3
refs = getcrossrefs(crossrefs, event)
# newkmt = []
for ref in refs.split():
if 'K' in ref:
newkmt = addkmt(event, ref)
elif newsync != 0:
newsync = rsync_ARTEMiS_data(ref, syncdir)
# if crossfrefs file not given:
else:
if 'K' in event:
newkmt = addkmt(event, event)
else:
newsync = rsync_ARTEMiS_data(event, syncdir)
# check if proceeds
if not (newsync == 0 or newkmt):
printi(tun + "No new data available for " + tit + event + tun + " (nothing to do)" + tend)
sys.exit(0)
if newkmt:
printi(tun + "New KMT data available for " + tit + event + tend)
if newsync == 0:
printi(tun + "New SYNC data available for " + tit + event + tend)
else:
printi(tcol + "Use local datasets" + tend)
# create final list od datasets
# printi(tcol + "Create final list of datasets" + tend)
# kevent = event
# syncset = fnmatch.filter(os.listdir(syncdir), '*' + event.replace('M', 'K') + '*.dat')
syncset = fnmatch.filter(os.listdir(syncdir), '*.dat')
for ds in syncset:
reformat(syncdir + ds, datapath + ds.replace('K', 'M'), ['mag', 'errmag', 'hjd'])
#
# if 'K' in ds:
# proc = subprocess.Popen('cp ' + syncdir + ds + ' ' + syncdir + ds.replace('K', 'M'), shell=True, executable='/bin/bash')
# proc.wait()
# syncset = [ds.replace('K', 'M') for ds in syncset]
# fetch other datasets in data/
othersets = fnmatch.filter(os.listdir(datapath), '*.dat')
if os.path.isfile(datapath + 'params.dat'):
othersets.remove('params.dat')
# remove duplicates
datasets = list(np.unique(syncset + othersets))
# discard symblinks (useful only if rsync is used without -L option)
datasets = [dataset for dataset in datasets if not os.path.islink(syncdir + dataset)]
# get OGLE first, if not MOA first, otherwise keep order
datasets = orderlist(datasets)
## reformat only sync/ datasets ## A FAIRE PLUS HAUT !!!
# for dataseti in datasets:
# if os.path.isfile(syncdir + dataseti):
# reformat(syncdir + dataseti, datapath + dataseti, ['mag', 'errmag', 'hjd'])
# add relative path to datasets names
for i in range(len(datasets)):
datasets[i] = datapath + datasets[i]
printi(tcol + "Found " + tit + str(len(datasets)) + tcol + " dataset(s)" + tend)
printd(tit + " " + str(datasets) + tend)
return datasets
def _process_fits((mclib, lclib, datasets, trange, init)):
"""Process of fit_subgrids"""
# set I/O shell display
tfil, tun, tcol, tit, tend = "\033[0m\033[1;35m", "\033[0m\033[1;31m", "\033[0m\033[31m", "\033[0m\033[3m", "\033[0m"
# check that mclib(.hdf5) exists
if not os.path.isfile(mclib):
raise IOError("file '" + mclib + "' is missing")
# verbose
printi(tcol + "Launching " + tit + "'" + mclib + "'" + tcol +
" grid with PID " + tit + str(os.getpid()) + tend)
with h5py.File(mclib, 'r') as fmclib:
# NEW read datasets and time range
mc = MagnificationCurve()
lc = LightCurve(datasets, trange=trange)
# global subgrid attributes
Ns = fmclib.attrs['Ns']
Nq = fmclib.attrs['Nq']
# prepare grid
grids = np.empty(Ns * Nq, dtype=np.float_)
gridq = np.empty(Ns * Nq, dtype=np.float_)
gridu0 = np.empty(Ns * Nq, dtype=np.float_)
gridalpha = np.empty(Ns * Nq, dtype=np.float_)
gridtE = np.empty(Ns * Nq, dtype=np.float_)
gridt0 = np.empty(Ns * Nq, dtype=np.float_)
gridrho = np.empty(Ns * Nq, dtype=np.float_)
gridchidof = np.empty(Ns * Nq, dtype=np.float_)
gridchi = np.empty(Ns * Nq, dtype=np.float_)
bestmc = np.empty(Ns * Nq, dtype=np.dtype('a128'))
for nsq in range(len(fmclib.keys())):
sqlib = fmclib[str(nsq)]
grids[nsq] = sqlib.attrs['s']
gridq[nsq] = sqlib.attrs['q']
fits = list()
usefit = 0
for id in sqlib:
# read mc
f_u0, f_alpha, f_tE, f_t0, f_rho, f_chidof, f_chi = [],[],[],[],[],[],[]
mc.read(mclib, str(nsq) + '/' + id)
# fit only if ∆mag(th) > ∆mag(exp)
dmag = 2.5 * np.log10(
fmclib[str(nsq) + '/' + id].attrs['mumax'])
if dmag < lc.content['dmag']:
printi(tfil + "Model delta(mag) too low : skipping" + tend)
printd(tit + " (delta(mag_th) = " + str(dmag) + " < " +
str(lc.content['dmag']) + tend)
else:
usefit += 1
# read reference croi in mc
croi = fmclib[str(nsq) + '/' + id].attrs['refcroi']
# fit for 0 < alpha < π/2
lc.fit(mc, croi, '+', init=init)
f_u0.append(lc.params['u0'])
f_alpha.append(lc.params['alpha'])
f_tE.append(lc.params['tE'])
f_t0.append(lc.params['t0'])
f_rho.append(lc.params['rho'])
f_chidof.append(lc.content['chi2'][0] /
lc.content['chi2'][1])
f_chi.append(lc.content['chi2'][0])
# fit for π < alpha < 3π/2)
lc.fit(mc, croi, '-', init=init)
f_u0.append(lc.params['u0'])
f_alpha.append(lc.params['alpha'])
f_tE.append(lc.params['tE'])
f_t0.append(lc.params['t0'])
f_rho.append(lc.params['rho'])
f_chidof.append(lc.content['chi2'][0] /
lc.content['chi2'][1])
f_chi.append(lc.content['chi2'][0])
# add fit to list if chi2 is not inf
if not np.all(np.isinf(f_chidof)):
arg = np.argmin(f_chidof)
u0 = f_u0[arg]
alpha = f_alpha[arg]
tE = f_tE[arg]
t0 = f_t0[arg]
rho = f_rho[arg]
chidof = f_chidof[arg]
chi = f_chi[arg]
fits.append([id, u0, alpha, tE, t0, rho, chidof, chi])
# verbose
printd(tcol +
"Percentage of useful magnification curves is about " +
tit + "{0:.0f}".format(100. * float(usefit) / float(id)) +
"%" + tend)
if fits:
# sort fits by increasing chi2 and get parameters
fits = np.array(fits)
arg = np.argsort(np.array(fits[:, 6], dtype=np.float_))
mcs = np.array(fits[arg, 0], dtype=np.int_)
u0 = np.array(fits[arg, 1], dtype=np.float_)
alpha = np.array(fits[arg, 2], dtype=np.float_)
tE = np.array(fits[arg, 3], dtype=np.float_)
t0 = np.array(fits[arg, 4], dtype=np.float_)
rho = np.array(fits[arg, 5], dtype=np.float_)
chidof = np.array(fits[arg, 6], dtype=np.float_)
chi = np.array(fits[arg, 7], dtype=np.float_)
# save best-fit parameters and chi2/dof
gridu0[nsq] = u0[0]
gridalpha[nsq] = alpha[0]
gridtE[nsq] = tE[0]
gridt0[nsq] = t0[0]
gridrho[nsq] = rho[0]
gridchidof[nsq] = chidof[0]
gridchi[nsq] = chi[0]
bestmc[nsq] = str(nsq) + '/' + str(mcs[0])
# verbose
printi(tcol + "Best-fit model at grid point " + tit + "'" +
str(nsq) + "'" + tcol + " in file " + tit + mclib +
tcol + " is " + tit + "'" + str(mcs[0]) + "'" + tcol +
" with " + tit + "chi2/dof={:.3e}".format(chidof[0]) +
tend)
else:
gridchidof[nsq] = np.inf
gridchi[nsq] = np.inf
# save log(X^2) map in HDF5 file: overwrite existing file
with h5py.File(lclib, 'w') as fitres:
gS = np.unique(grids)
gQ = np.unique(gridq)
gs, gq = np.meshgrid(gS, gQ)
fitres.create_dataset('s', data=gs)
fitres.create_dataset('q', data=gq)
fitres.create_dataset('u0', data=gridu0.reshape(Ns, Nq).T)
fitres.create_dataset('alpha', data=gridalpha.reshape(Ns, Nq).T)
fitres.create_dataset('tE', data=gridtE.reshape(Ns, Nq).T)
fitres.create_dataset('t0', data=gridt0.reshape(Ns, Nq).T)
fitres.create_dataset('rho', data=gridrho.reshape(Ns, Nq).T)
fitres.create_dataset('chidof', data=gridchidof.reshape(Ns, Nq).T)
fitres.create_dataset('chi', data=gridchi.reshape(Ns, Nq).T)
fitres.flush()
fitres.close()
fmclib.flush()
fmclib.close()
# verbose
printi(tun + "Light curve grid " + tit + "'" + lclib + "'" + tun +
" complete" + tend)
def compute_subgrids(gridsprefix, gridlist, nprocs=1, f_rcroi=2.):
"""Create magnification curve sub-grids (HDF5 files)"""
# set I/O shell display
tcol, tit, tend = "\033[0m\033[31m", "\033[0m\033[3m", "\033[0m"
# check whether input names does not contain extensions
if '.hdf5' in gridsprefix:
raise NameError("grid prefix should not contain .hdf5 extension")
# create name
grid = gridsprefix + '.hdf5'
# check weather library exists
if not os.path.isfile(grid):
raise IOError("file '" + grid + "' is missing")
# verbose
printd(tcol + "Grid " + tit + "'" + grid + "'" + tcol + " chosen" + tend)
# mutliprocessing: create grid list names
listmclibs, listgrids = list(), list()
with h5py.File(grid, 'r') as fgrid:
Ngs = fgrid.attrs['Ngs']
Ngq = fgrid.attrs['Ngq']
nmc = fgrid.attrs['nmc']
pcaus = fgrid.attrs['pcaus']
k = 0
for j in range(Ngq):
for i in range(Ngs):
if k in gridlist:
# list of mc libraries to process
mclibk = gridsprefix + '_' + str(k) + '.hdf5'
# # if file exist, abort --> non on complete maintenant
# if os.path.isfile(mclibk):
# raise IOError("file '" + mclibk + "' already exists")
# add mc library to to-process list
listmclibs.append(mclibk)
# list of corresponding s,q values
gridi = fgrid[str(i) + ' ' + str(j)]
listgrids.append((gridi['s'][:], gridi['q'][:]))
k += 1
fgrid.flush()
fgrid.close()
# mutliprocessing: create arguments of _process_grids, and create workers pool
printi(tcol + "Starting manager with PID " + tit + str(os.getpid()) +
tcol + " running " + tit + str(nprocs) + tcol + " process(es)" +
tend)
listargs = zip(listmclibs, listgrids, repeat(nmc), repeat(pcaus),
repeat(f_rcroi))
pool = Pool(processes=nprocs)
pool.imap_unordered(_process_grids, listargs)
# collect results
pool.close()
pool.join()
def create_maingrid(gridsprefix,
srange,
qrange,
majgrid,
mingrid,
nmc,
pcaus,
axis=[0.1, 10., 8e-6, 1.2]):
"""Generate an HDF5 file with definition of sub- s,q-grids
Parameters
----------
gridsprefix : str
Name (without any extension) of output HDF5 library of magnifications
curves s,q-grids and corresponding PDF map of grids.
srange : tuple
Global range in s. Default is: srange=(0.2, 5.0)
qrange : tuple
Global range in q. Default is: qrange=(1e-5, 1.)
majgrid : tuple
Number of sub- s,q-grids. Default is: majgrid=(12, 5)
mingrid : tuple
Size of sub s,q-grids. Default is: mingrid=(7, 7)
axis : float 1-D array, optional
Plot limits. Usage: axis=[xmin, xmax, ymin, ymax].
Default is: axis=[0.1, 10., 8e-6, 1.].
plot : bool, optional
If True, display grid map on screen with pylab.show()
and produce an output PDF file. Default is: False.
Returns
-------
out : HDF5 file
File containing arrays of s and q for each of the sub-grids.
out : pylab.show() and PDF file
If plot=True, display grid map on screen and produce a PDF file.
Examples
--------
>>> grs = GridSearch()
>>> grs = gengridlibs('gridlib-0')
>>> grs.gengridlibs('gridlib-1', srange=(0.8, 1.25), qrange=(0.001, 0.1), majgrid=(3, 3), mingrid=(6, 6), axis=[0.7, 1.4, 0.0006, 0.2], plot=True)
"""
# set I/O shell display
tit, tcol, tend = "\033[0m\033[3m", "\033[0m\033[35m", "\033[0m"
# check whether the grid name does not contain extensions
if '.hdf5' in gridsprefix:
raise NameError("name should not contain extension")
# create file names
pdfname = gridsprefix + '.pdf'
libname = gridsprefix + '.hdf5'
# check weather grid already exists
if os.path.isfile(libname):
raise IOError("file '" + libname + "' already exists")
# define sub-grids
smin, smax = srange[0], srange[1]
qmin, qmax = qrange[0], qrange[1]
Ngs, Ngq = majgrid[0], majgrid[1]
Ns, Nq = mingrid[0], mingrid[1]
S = np.empty([Ngs, Ns], dtype=np.float_)
Q = np.empty([Ngq, Nq], dtype=np.float_)
fullS = np.geomspace(smin, smax, Ngs * Ns, endpoint=True)
fullQ = np.geomspace(qmin, qmax, Ngq * Nq, endpoint=True)
for i in range(Ngs):
S[i, ] = fullS[i * Ns:(i + 1) * Ns]
for j in range(Ngq):
Q[j, ] = fullQ[j * Nq:(j + 1) * Nq]
# verbose
printi(tcol + "Create grid " + tit + "'" + libname + "'" + tcol +
" (view configuration:" + tit + "'" + pdfname + "'" + tcol + ")" +
tend)
printd(tit + " (" + str(smin) + " ≤ s ≤ " + str(smax) + ", " + str(qmin) +
" ≤ q ≤ " + str(qmax) + ", " + str(Ngs) + " x " + str(Ngq) +
" sub-grids, each of size " + str(Ns) + " x " + str(Nq) + ")" +
tend)
# create individual grids and store in HDF5 file
grids = list()
with h5py.File(libname, 'w') as gridlib:
gridlib.attrs['Ngs'] = np.shape(S)[0]
gridlib.attrs['Ngq'] = np.shape(Q)[0]
gridlib.attrs['nmc'] = nmc
gridlib.attrs['pcaus'] = pcaus
for j in range(np.shape(Q)[0]):
for i in range(np.shape(S)[0]):
grids.append((S[i, ], Q[j, ]))
gridlibk = gridlib.create_group(u'' + str(i) + ' ' + str(j))
gridlibk.create_dataset('s', data=S[i, ])
gridlibk.create_dataset('q', data=Q[j, ])
gridlib.flush()
gridlib.close()
# plot template grid
plt.rc('font', size=14)
plt.close('all')
fig, MAP = plt.subplots(1, figsize=(8, 6))
MAP.set_xscale('log')
MAP.set_yscale('log')
MAP.set_xlim([axis[0], axis[1]])
MAP.set_ylim([axis[2], axis[3]])
plt.subplots_adjust(left=0.16,
bottom=0.13,
right=0.94,
top=0.94,
wspace=None,
hspace=None)
MAP.set_title(r'Binary-lens search map template')
MAP.set_xlabel(r'$s$')
MAP.set_ylabel(r'$q$')
for k in range(len(grids)):
SP, QP = np.meshgrid(grids[k][0], grids[k][1])
MAP.scatter(SP, QP, marker='*', s=12)
plt.text(grids[k][0][0], grids[k][1][0], str(k), fontsize=15)
plt.xticks(np.array([0.2, 0.3, 0.5, 0.7, 1, 2, 3, 4, 5]),
np.array(['0.2', '0.3', '0.5', '0.7', '1', '2', '3', '4', '5']))
plt.savefig(pdfname)
def process_results(gridsprefix, fitsprefix, nmod=9):
"""Process fit results
nmod : int
Maximal number of output models.
"""
# set I/O shell display
tcol, tun, tend, tit = "\033[0m\033[35m", "\033[0m\033[1m\033[35m", "\033[0m", "\033[0m\033[3m"
# check that names do not have extension
if '.hdf5' in gridsprefix:
raise NameError("grid prefix should not contain .hdf5 extension")
if '.hdf5' in fitsprefix:
raise NameError("fit prefix should not contain .hdf5 extension")
# check that mcgrid(.hdf5) exists
grid = gridsprefix + '.hdf5'
if not os.path.isfile(grid):
raise IOError("file '" + grid + "' is missing")
# verbose
printd(tcol + "Grid file " + tit + "'" + grid + "'" + tend)
# collect fit files and fill missing ones
missingfits, missing = [], []
with h5py.File(grid, 'r') as fgrid:
Ngs = fgrid.attrs['Ngs']
Ngq = fgrid.attrs['Ngq']
k = 0
Cgs, Cgq, Cgu0, Cgalpha, CgtE, Cgt0, Cgrho, Cgchidof, Cgchi = [],[],[],[],[],[],[],[],[]
for j in range(Ngq):
Lgs, Lgq, Lgu0, Lgalpha, LgtE, Lgt0, Lgrho, Lgchidof, Lgchi = [],[],[],[],[],[],[],[],[]
for i in range(Ngs):
gridfitsk = fitsprefix + '_' + str(k) + '.hdf5'
if os.path.isfile(gridfitsk):
# fit file exists
with h5py.File(gridfitsk, 'r') as fgridfitsk:
Lgs.append(fgridfitsk['s'][:])
Lgq.append(fgridfitsk['q'][:])
Lgu0.append(fgridfitsk['u0'][:])
Lgalpha.append(fgridfitsk['alpha'][:])
LgtE.append(fgridfitsk['tE'][:])
Lgt0.append(fgridfitsk['t0'][:])
Lgrho.append(fgridfitsk['rho'][:])
Lgchidof.append(fgridfitsk['chidof'][:])
Lgchi.append(fgridfitsk['chi'][:])
fgridfitsk.flush()
fgridfitsk.close()
else:
# fit file is missing
default = fgrid[str(i) + ' ' + str(j)]
meshs, meshq = np.meshgrid(default['s'][:],
default['q'][:])
Lgs.append(meshs)
Lgq.append(meshq)
fails = np.full_like(meshs, np.inf)
Lgu0.append(fails)
Lgalpha.append(fails)
LgtE.append(fails)
Lgt0.append(fails)
Lgrho.append(fails)
Lgchidof.append(fails)
Lgchi.append(fails)
missingfits.append(gridfitsk)
missing.append((default['s'][:], default['q'][:]))
k += 1
Cgs.append(np.concatenate(Lgs, axis=1))
Cgq.append(np.concatenate(Lgq, axis=1))
Cgu0.append(np.concatenate(Lgu0, axis=1))
Cgalpha.append(np.concatenate(Lgalpha, axis=1))
CgtE.append(np.concatenate(LgtE, axis=1))
Cgt0.append(np.concatenate(Lgt0, axis=1))
Cgrho.append(np.concatenate(Lgrho, axis=1))
Cgchidof.append(np.concatenate(Lgchidof, axis=1))
Cgchi.append(np.concatenate(Lgchi, axis=1))
fgrid.flush()
fgrid.close()
s = np.concatenate(Cgs, axis=0)
q = np.concatenate(Cgq, axis=0)
u0 = np.concatenate(Cgu0, axis=0)
alpha = np.concatenate(Cgalpha, axis=0)
tE = np.concatenate(CgtE, axis=0)
t0 = np.concatenate(Cgt0, axis=0)
rho = np.concatenate(Cgrho, axis=0)
chidof = np.concatenate(Cgchidof, axis=0)
chi = np.concatenate(Cgchi, axis=0)
search_map = [s, q, chidof, chi, missing]
# verbose
if missingfits:
printi(tcol + "Fit crashed for " + tit + str(len(missingfits)) + tcol +
" sub-grids" + tend)
for mi in missingfits:
printd(tit + " ('" + mi + "')" + tend)
# order models by X^2
ind = np.unravel_index(np.argsort(chidof, axis=None), chidof.shape)
models = list()
for i in range(nmod):
params = [
u0[ind][i], alpha[ind][i], tE[ind][i], t0[ind][i], rho[ind][i]
]
if np.any(np.isinf(params)):
nmod = i
break
models.append({
's': s[ind][i],
'q': q[ind][i],
'u0': u0[ind][i],
'alpha': alpha[ind][i],
'tE': tE[ind][i],
't0': t0[ind][i],
'rho': rho[ind][i]
})
# list best-fit parameters
befi = " {0:<2s} {1:<9s} {2:<11s} {3:<12s} {4:<10s} {5:<10s} {6:<12s} {7:<11s} {8:<10s} {9:<12s}\n".format(
'', 's', 'q', 'u0', 'alpha', 'tE', 't0', 'rho', 'X^2/dof', 'X^2')
for i in range(nmod):
befi += " {0:<2d} {1:<9.6f} {2:<11.4e} {3:<+12.4e} {4:<10f} {5:<10f} {6:<12f} {7:<11.4e} {8:<10f} {9:<12.1f}\n".format(
i + 1, s[ind][i], q[ind][i], u0[ind][i], alpha[ind][i], tE[ind][i],
t0[ind][i], rho[ind][i], chidof[ind][i], chi[ind][i])
# create _rank.txt output file
f = open(fitsprefix + '_rank.txt', 'w')
f.write(befi)
f.close()
# verbose
printi(tcol + "Best-fit models ranking:\n" + tend + befi)
return search_map, models