def testGetDist(self):
from getdist.command_line import getdist_command
os.chdir(self.tempdir)
res = getdist_command([self.root])
# Note this can fail if your local analysis defaults changes the default ignore_rows
self.assertTrue('-Ln(mean like) = 2.31' in res, res)
fname = 'testchain_pars.ini'
getdist_command(['--make_param_file', fname])
ini = IniFile(fname)
ini.params['no_plots'] = False
ini.params['plot_2D_num'] = 1
ini.params['plot1'] = 'x y'
ini.params['num_3D_plots'] = 1
ini.params['3D_plot1'] = 'x y x'
ini.params['triangle_params'] = '*[xy]*'
ini.saveFile(fname)
res = getdist_command([fname, self.root])
self.assertTrue('-Ln(mean like) = 2.31' in res)
def check_run():
for f in ['.py', '_2D.py', '_3D.py', '_tri.py']:
pyname = self.root + f
self.assertTrue(os.path.isfile(pyname))
subprocess.check_output(['python', pyname])
pdf = self.root + f.replace('py', 'pdf')
self.assertTrue(os.path.isfile(pdf))
os.remove(pdf)
os.remove(pyname)
check_run()
def testGetDist(self):
def callGetDist(args):
if os.getenv('TRAVIS', None):
return str(
subprocess.check_output(['GetDist.py'] + args,
env={'PATH': os.getenv('PATH')}))
else:
return str(
subprocess.check_output([
'python',
os.path.join(os.path.dirname(__file__), '..' +
os.sep, 'GetDist.py')
] + args))
os.chdir(self.tempdir)
res = callGetDist([self.root])
# Note this can fail if your local analysis defaults changes the default ignore_rows
self.assertTrue('-Ln(mean like) = 2.30' in res)
fname = 'testchain_pars.ini'
callGetDist(['--make_param_file', fname])
ini = IniFile(fname)
ini.params['no_plots'] = False
ini.params['plot_2D_num'] = 1
ini.params['plot1'] = 'x y'
ini.params['num_3D_plots'] = 1
ini.params['3D_plot1'] = 'x y x'
ini.params['plot_data_dir'] = ''
ini.params['triangle_params'] = '*[xy]*'
ini.saveFile(fname)
res = callGetDist([fname, self.root])
self.assertTrue('-Ln(mean like) = 2.30' in res)
self.assertFalse(
os.path.isfile(
os.path.join(self.tempdir, 'plot_data', 'testchain_2D_x_y')))
def checkRun():
for f in ['.py', '_2D.py', '_3D.py', '_tri.py']:
pyname = self.root + f
self.assertTrue(os.path.isfile(pyname))
subprocess.check_output(['python', pyname])
pdf = self.root + f.replace('py', 'pdf')
self.assertTrue(os.path.isfile(pdf))
os.remove(pdf)
os.remove(pyname)
checkRun()
ini.params['plot_data_dir'] = 'plot_data/'
ini.saveFile(fname)
callGetDist([fname, self.root])
self.assertTrue(
os.path.isfile(
os.path.join(self.tempdir, 'plot_data', 'testchain_2D_x_y')))
checkRun()
shutil.rmtree(os.path.join(self.tempdir, 'plot_data'))
def testGetDist(self):
def callGetDist(args):
if os.getenv('TRAVIS', None):
return str(subprocess.check_output(['GetDist.py'] + args, env={'PATH': os.getenv('PATH')}))
else:
return str(subprocess.check_output(
['python', os.path.join(os.path.dirname(__file__), '..' + os.sep, 'GetDist.py')] + args))
os.chdir(self.tempdir)
res = callGetDist([self.root])
# Note this can fail if your local analysis defaults changes the default ignore_rows
self.assertTrue('-Ln(mean like) = 2.30' in res)
fname = 'testchain_pars.ini'
callGetDist(['--make_param_file', fname])
ini = IniFile(fname)
ini.params['no_plots'] = False
ini.params['plot_2D_num'] = 1
ini.params['plot1'] = 'x y'
ini.params['num_3D_plots'] = 1
ini.params['3D_plot1'] = 'x y x'
ini.params['plot_data_dir'] = ''
ini.params['triangle_params'] = '*[xy]*'
ini.saveFile(fname)
res = callGetDist([fname, self.root])
self.assertTrue('-Ln(mean like) = 2.30' in res)
self.assertFalse(os.path.isfile(os.path.join(self.tempdir, 'plot_data', 'testchain_2D_x_y')))
def checkRun():
for f in ['.py', '_2D.py', '_3D.py', '_tri.py']:
pyname = self.root + f
self.assertTrue(os.path.isfile(pyname))
subprocess.check_output(['python', pyname])
pdf = self.root + f.replace('py', 'pdf')
self.assertTrue(os.path.isfile(pdf))
os.remove(pdf)
os.remove(pyname)
checkRun()
ini.params['plot_data_dir'] = 'plot_data/'
ini.saveFile(fname)
callGetDist([fname, self.root])
self.assertTrue(os.path.isfile(os.path.join(self.tempdir, 'plot_data', 'testchain_2D_x_y')))
checkRun()
shutil.rmtree(os.path.join(self.tempdir, 'plot_data'))
# if (l >= 2): SN += (2 * l + 1) * fsky * (d[l - 2, 1] / (NoiseCl + d[l - 2, 1])) ** 2
# print 'Number of modes: ', SN
outfile = open(outDir + outRoot + '_Noise.dat', 'w')
for l in range(2, lmax + 1):
outfile.write("%d " % l + " ".join("%E" % elem for elem in NoiseOut[l]) + "\n")
outfile.close()
dataset['fullsky_exact_fksy'] = fsky
dataset['name'] = outRoot
dataset['dataset_format'] = 'CMBLike2'
dataset['like_approx'] = 'exact'
dataset['cl_lmin'] = lmin
dataset['cl_lmax'] = lmax
dataset['binned'] = False
dataset['cl_hat_includes_noise'] = False
dataset['cl_hat_file'] = outPath + outRoot + '.dat'
dataset['cl_hat_order'] = dataCols
dataset['cl_noise_file '] = outPath + outRoot + '_Noise.dat'
dataset['cl_noise_order'] = 'TT EE BB'
ini.saveFile(outDir + outRoot + '.dataset')
shutil.copy(lensedTotClFileRoot, outDir + outRoot + '.dat')
def makeGrid(batchPath,
settingName=None,
settings=None,
readOnly=False,
interactive=False):
batchPath = os.path.abspath(batchPath) + os.sep
# 0: chains, 1: importance sampling, 2: best-fit, 3: best-fit and Hessian
cosmomcAction = 0
if not settings:
if not settingName:
if not pathIsGrid(batchPath):
raise Exception(
'Need to give name of setting file if batchPath/config does not exist'
)
readOnly = True
sys.path.insert(0, batchPath + 'config')
sys.modules['batchJob'] = batchjob # old name
settings = __import__(
IniFile(batchPath +
'config/config.ini').params['setting_file'].replace(
'.py', ''))
else:
settings = __import__(settingName, fromlist=['dummy'])
batch = batchjob.batchJob(batchPath, settings.ini_dir)
if hasattr(settings, 'skip'): batch.skip = settings.skip
batch.makeItems(settings, messages=not readOnly)
if readOnly:
for jobItem in [b for b in batch.jobItems]:
if not jobItem.chainExists():
batch.jobItems.remove(jobItem)
batch.save()
print('OK, configured grid with %u existing chains' %
(len(batch.jobItems)))
return batch
else:
batch.makeDirectories(settings.__file__)
batch.save()
# priors and widths for parameters which are varied
start_at_bestfit = getattr(settings, 'start_at_bestfit', False)
params = getattr(settings, 'params', default_params)
param_extra = getattr(settings, 'param_extra_opts',
default_param_extra_opts)
for jobItem in batch.items(wantSubItems=False):
jobItem.makeChainPath()
ini = IniFile()
for param in jobItem.param_set:
ini.params['param[' + param + ']'] = params[param]
if param_extra is not None and param in param_extra:
ini.params.update(param_extra[param])
if hasattr(settings, 'extra_opts'):
ini.params.update(settings.extra_opts)
ini.params['file_root'] = jobItem.chainRoot
cov_dir_name = getattr(settings, 'cov_dir', 'planck_covmats')
covdir = os.path.join(batch.basePath, cov_dir_name)
covmat = os.path.join(covdir, jobItem.name + '.covmat')
if not os.path.exists(covmat):
covNameMappings = getattr(settings, 'covNameMappings', None)
mapped_name_norm = jobItem.makeNormedName(covNameMappings)[0]
covmat_normed = os.path.join(covdir, mapped_name_norm + '.covmat')
covmat = covmat_normed
if not os.path.exists(covmat) and hasattr(jobItem.data_set,
'covmat'):
covmat = batch.basePath + jobItem.data_set.covmat
if not os.path.exists(covmat) and hasattr(settings, 'covmat'):
covmat = batch.basePath + settings.covmat
else:
covNameMappings = None
if os.path.exists(covmat):
ini.params['propose_matrix'] = covmat
if getattr(settings, 'newCovmats', True):
ini.params['MPI_Max_R_ProposeUpdate'] = 20
else:
hasCov = False
ini.params['MPI_Max_R_ProposeUpdate'] = 20
covmat_try = []
if 'covRenamer' in dir(settings):
covmat_try += settings.covRenamer(jobItem.name)
covmat_try += settings.covRenamer(mapped_name_norm)
if hasattr(settings, 'covrenames'):
for aname in [jobItem.name, mapped_name_norm]:
covmat_try += [
aname.replace(old, new, 1)
for old, new in settings.covrenames if old in aname
]
for new1, old1 in settings.covrenames:
if old1 in aname:
name = aname.replace(old1, new1, 1)
covmat_try += [
name.replace(old, new, 1)
for old, new in settings.covrenames
if old in name
]
if 'covWithoutNameOrder' in dir(settings):
if covNameMappings:
removes = copy.deepcopy(covNameMappings)
else:
removes = dict()
for name in settings.covWithoutNameOrder:
if name in jobItem.data_set.names:
removes[name] = ''
covmat_try += [jobItem.makeNormedName(removes)[0]]
covdir2 = os.path.join(
batch.basePath,
getattr(settings, 'cov_dir_fallback', cov_dir_name))
for name in covmat_try:
covmat = os.path.join(batch.basePath, covdir2,
name + '.covmat')
if os.path.exists(covmat):
ini.params['propose_matrix'] = covmat
print('covmat ' + jobItem.name + ' -> ' + name)
hasCov = True
break
if not hasCov:
print('WARNING: no matching specific covmat for ' +
jobItem.name)
ini.params['start_at_bestfit'] = start_at_bestfit
updateIniParams(ini, jobItem.data_set.params, batch.commonPath)
for deffile in settings.defaults:
ini.defaults.append(batch.commonPath + deffile)
if hasattr(settings, 'override_defaults'):
ini.defaults = [
batch.commonPath + deffile
for deffile in settings.override_defaults
] + ini.defaults
ini.params['action'] = cosmomcAction
ini.saveFile(jobItem.iniFile())
if not start_at_bestfit:
setMinimize(jobItem, ini)
variant = '_minimize'
ini.saveFile(jobItem.iniFile(variant))
# add ini files for importance sampling runs
for imp in jobItem.importanceJobs():
if not getattr(jobItem, 'importanceFilter', None): continue
if batch.hasName(imp.name.replace('_post', '')):
raise Exception(
'importance sampling something you already have?')
for minimize in (False, True):
if minimize and not getattr(imp, 'want_minimize', True):
continue
ini = IniFile()
updateIniParams(ini, imp.importanceSettings, batch.commonPath)
if cosmomcAction == 0 and not minimize:
for deffile in settings.importanceDefaults:
ini.defaults.append(batch.commonPath + deffile)
ini.params['redo_outroot'] = imp.chainRoot
ini.params['action'] = 1
else:
ini.params['file_root'] = imp.chainRoot
if minimize:
setMinimize(jobItem, ini)
variant = '_minimize'
else:
variant = ''
ini.defaults.append(jobItem.iniFile())
ini.saveFile(imp.iniFile(variant))
if cosmomcAction != 0: break
if not interactive: return batch
print('Done... to run do: python python/runbatch.py ' + batchPath)
if not start_at_bestfit:
print('....... for best fits: python python/runbatch.py ' + batchPath +
' --minimize')
print('')
print('for importance sampled: python python/runbatch.py ' + batchPath +
' --importance')
print('for best-fit for importance sampled: python python/runbatch.py ' +
batchPath + ' --importance_minimize')
elif os.path.exists(batch.commonPath + args.base_ini):
ini.defaults.append(batch.commonPath + args.base_ini)
else:
raise ValueError("base_ini file not found")
if hasattr(batch, 'getdist_options'):
ini.params.update(batch.getdist_options)
tag = ''
if jobItem.isImportanceJob or args.burn_removed or jobItem.isBurnRemoved():
ini.params['ignore_rows'] = 0
elif args.burn_remove is not None:
ini.params['ignore_rows'] = args.burn_remove
if jobItem.isImportanceJob:
ini.params['compare_num'] = 1
ini.params['compare1'] = jobItem.parent.chainRoot
if args.no_plots: ini.params['no_plots'] = True
if args.make_plots: ini.params['make_plots'] = True
fname = ini_dir + jobItem.name + tag + '.ini'
ini.params.update(jobItem.dist_settings)
ini.saveFile(fname)
if not args.norun and (not args.notexist or not jobItem.getDistExists()) and (
not args.update_only or jobItem.getDistNeedsUpdate()):
if jobItem.chainExists():
print("running: " + fname)
processes.add(subprocess.Popen([args.command] + args.command_params + [fname]))
while len(processes) >= args.procs:
time.sleep(.1)
processes.difference_update([p for p in processes if p.poll() is not None])
else:
if not args.exist: print("Chains do not exist yet: " + jobItem.chainRoot)
bf = types.BestFit(root + '.minimum',
setParamNameFile=root + '.paramnames',
want_fixed=True)
for camb, cosmomc in list(pars.items()):
par = bf.parWithName(cosmomc)
if par is not None: ini.params[camb] = par.best_fit
ini.params['scalar_amp(1)'] = float(ini.params['scalar_amp(1)']) / 1e9
nmassive = 1
neffstandard = 3.046 / 3
ini.params['massless_neutrinos'] = float(
ini.params['massless_neutrinos']) - neffstandard * nmassive
ini.params['massive_neutrinos'] = int(round(neffstandard * nmassive))
ini.params['nu_mass_degeneracies'] = neffstandard * nmassive
ini.params['share_delta_neff'] = False
ini.params['tensor_spectral_index(1)'] = -float(
ini.params['initial_ratio(1)']) / 8
inPars = IniFile(root + '.inputparams')
if inPars.bool('use_nonlinear_lensing', True):
ini.params['do_nonlinear'] = 3
# Note, if you want accurate spectrun on small scales, may need to increase accuracy
ini.saveFile(sys.argv[2])
print(
'OK, though note this does not support all parameter extensions from LCDM')
def make_forecast_cmb_dataset(fiducial_Cl,
output_root,
output_dir=None,
noise_muK_arcmin_T=None,
noise_muK_arcmin_P=None,
NoiseVar=None,
ENoiseFac=2,
fwhm_arcmin=None,
lmin=2,
lmax=None,
fsky=1.0,
lens_recon_noise=None,
cl_dict_lmin=0): # pragma: no cover
"""
Make a simulated .dataset and associated files with 'data' set at the input fiducial
model. Uses the exact full-sky log-likelihood, scaled by fsky.
If you want to use numerical N_L CMB noise files, you can just replace the noise
.dat text file produced by this function.
:param fiducial_Cl: dictionary of Cls to use, combination of tt, te, ee, bb, pp;
note te must be included with tt and ee when using them
:param output_root: root name for output files, e.g. 'my_sim1'
:param output_dir: output directory
:param noise_muK_arcmin_T: temperature noise in muK-arcmin
:param noise_muK_arcmin_P: polarization noise in muK-arcmin
:param NoiseVar: alternatively if noise_muK_arcmin_T is None, effective
isotropic noise variance for the temperature (N_L=NoiseVar with no beam)
:param ENoiseFac: factor by which polarization noise variance is higher thab
NoiseVar (usually 2, for Planck about 4
as only half the detectors polarized)
:param fwhm_arcmin: beam fwhm in arcminutes
:param lmin: l_min
:param lmax: l_max
:param fsky: sky fraction
:param lens_recon_noise: optional array, starting at L=0, for the
pp lensing reconstruction noise, in [L(L+1)]^2C_L^phi/2pi units
:param cl_dict_lmin: l_min for the arrays in fiducial_Cl
:return: IniFile that was saved
"""
ini = IniFile()
dataset = ini.params
cl_keys = fiducial_Cl.keys()
use_CMB = set(cl_keys).intersection(set(CMB_keys))
use_lensing = lens_recon_noise
if use_CMB:
if NoiseVar is None:
if noise_muK_arcmin_T is None:
raise ValueError('Must specify noise')
NoiseVar = white_noise_from_muK_arcmin(noise_muK_arcmin_T)
if noise_muK_arcmin_P is not None:
ENoiseFac = (noise_muK_arcmin_P / noise_muK_arcmin_T)**2
elif noise_muK_arcmin_T is not None or noise_muK_arcmin_P is not None:
raise ValueError('Specific either noise_muK_arcmin or NoiseVar')
fields_use = ''
if 'tt' in cl_keys or 'te' in cl_keys:
fields_use = 'T'
if 'ee' in cl_keys or 'te' in cl_keys:
fields_use += ' E'
if 'bb' in cl_keys:
fields_use += ' B'
if 'pp' in cl_keys and use_lensing:
fields_use += ' P'
if 'tt' in cl_keys and 'ee' in cl_keys and 'te' not in cl_keys:
raise ValueError(
'Input power spectra should have te if using tt and ee -'
'using the exact likelihood requires the full covariance.')
else:
fields_use = 'P'
if not os.path.exists(output_dir):
os.makedirs(output_dir)
dataset['fields_use'] = fields_use
if use_CMB:
fwhm = fwhm_arcmin / 60
xlc = 180 * np.sqrt(8. * np.log(2.)) / np.pi
sigma2 = (fwhm / xlc)**2
noise_cols = 'TT EE BB'
if use_lensing:
noise_cols += ' PP'
elif use_lensing:
noise_cols = 'PP'
else:
raise ValueError('Must use CMB or lensing C_L')
noise_file = output_root + '_Noise.dat'
with open(os.path.join(output_dir, noise_file), 'w') as f:
f.write('#L %s\n' % noise_cols)
for ell in range(lmin, lmax + 1):
noises = []
if use_CMB:
# noinspection PyUnboundLocalVariable
noise_cl = ell * (ell + 1.) / 2 / np.pi * NoiseVar * np.exp(
ell * (ell + 1) * sigma2)
noises += [
noise_cl, ENoiseFac * noise_cl, ENoiseFac * noise_cl
]
if use_lensing:
noises += [lens_recon_noise[ell]]
f.write("%d " % ell + " ".join("%E" % elem
for elem in noises) + "\n")
dataset['fullsky_exact_fksy'] = fsky
dataset['dataset_format'] = 'CMBLike2'
dataset['like_approx'] = 'exact'
dataset['cl_lmin'] = lmin
dataset['cl_lmax'] = lmax
dataset['binned'] = False
dataset['cl_hat_includes_noise'] = False
save_cl_dict(os.path.join(output_dir, output_root + '.dat'),
fiducial_Cl,
cl_dict_lmin=cl_dict_lmin)
dataset['cl_hat_file'] = output_root + '.dat'
dataset['cl_noise_file '] = noise_file
ini.saveFile(os.path.join(output_dir, output_root + '.dataset'))
return ini
def make_forecast_cmb_dataset(input_cl_file,
output_root,
output_dir=None,
noise_muK_arcmin_T=None,
noise_muK_arcmin_P=None,
NoiseVar=None,
ENoiseFac=2,
fwhm_arcmin=None,
lmin=2,
lmax=None,
fsky=1,
fields_use=None,
lens_recon_noise=None,
cl_data_cols=''):
"""
Make a simulated .dataset and associated files with 'data' set at the input fiducial model.
:param input_cl_file: input fiducial CL
:param output_root: root name for output files, e.g. 'my_sim1'
:param output_dir: output directory
:param noise_muK_arcmin_T: temperature noise in muK-arcmin
:param noise_muK_arcmin_P: polarization noise in muK-arcmin
:param NoiseVar: effective isotropic noise variance for the temperature (N_L=NoiseVar with no beam)
:param ENoiseFac: factor by which polarization noise variance is higher (usually 2, for Planck about 4
as only half the detectors polarized)
:param fwhm_arcmin: beam fwhm in arcminutes
:param lmin: l_min
:param lmax: l_max
:param fsky: sky fraction
:param fields_use: optional list of fields to restict to (e.g. 'T E')
:param lens_recon_noise: optional array, starting at L=0, for the PP lensing reconstruction noise, in [L(L+1)]^2C_L^phi/2pi units
:param cl_data_cols: if not specified in file header, order of columns in input CL file (e.g. 'TT TE EE BB PP')
:return:
"""
use_lensing = lens_recon_noise
use_CMB = noise_muK_arcmin_T or NoiseVar is not None
ini = IniFile()
dataset = ini.params
if not cl_data_cols:
cl_data_cols = lastTopComment(input_cl_file)
if not cl_data_cols:
raise Exception(
'input CL file must specific names of columns (TT TE EE..)')
else:
dataset['cl_hat_order'] = cl_data_cols
if use_CMB:
if NoiseVar is None:
if noise_muK_arcmin_T is None:
raise ValueError('Must specify noise')
NoiseVar = (noise_muK_arcmin_T * np.pi / 180 / 60.)**2
if noise_muK_arcmin_P is not None:
ENoiseFac = (noise_muK_arcmin_P / noise_muK_arcmin_T)**2
elif noise_muK_arcmin_T is not None or noise_muK_arcmin_P is not None:
raise ValueError('Specific either noise_muK_arcmin or NoiseVar')
if not fields_use:
fields_use = ''
if 'TT' or 'TE' in cl_data_cols: fields_use = 'T'
if 'EE' or 'TE' in cl_data_cols: fields_use += ' E'
if 'BB' in cl_data_cols: fields_use += ' B'
if 'PP' in cl_data_cols and use_lensing: fields_use += ' P'
else:
fields_use = fields_use or 'P'
if output_dir is None:
output_dir = os.path.join(os.path.dirname(__file__), '..', 'data',
output_root)
if not os.path.exists(output_dir): os.makedirs(output_dir)
dataset['fields_use'] = fields_use
if use_CMB:
fwhm = fwhm_arcmin / 60
xlc = 180 * np.sqrt(8. * np.log(2.)) / np.pi
sigma2 = (fwhm / xlc)**2
noise_cols = 'TT EE BB'
if use_lensing: noise_cols += ' PP'
elif use_lensing:
noise_cols = 'PP'
noise_file = output_root + '_Noise.dat'
with open(os.path.join(output_dir, noise_file), 'w') as f:
f.write('#L %s\n' % noise_cols)
for l in range(lmin, lmax + 1):
NoiseCl = l * (l + 1.) / 2 / np.pi * NoiseVar * np.exp(
l * (l + 1) * sigma2)
noises = []
if use_CMB:
noises += [NoiseCl, ENoiseFac * NoiseCl, ENoiseFac * NoiseCl]
if use_lensing: noises += [lens_recon_noise[l]]
f.write("%d " % l + " ".join("%E" % elem
for elem in noises) + "\n")
dataset['fullsky_exact_fksy'] = fsky
dataset['dataset_format'] = 'CMBLike2'
dataset['like_approx'] = 'exact'
dataset['cl_lmin'] = lmin
dataset['cl_lmax'] = lmax
dataset['binned'] = False
dataset['cl_hat_includes_noise'] = False
shutil.copy(input_cl_file, os.path.join(output_dir, output_root + '.dat'))
dataset['cl_hat_file'] = output_root + '.dat'
dataset['cl_noise_file '] = noise_file
ini.saveFile(os.path.join(output_dir, output_root + '.dataset'))
def makeGrid(batchPath, settingName=None, settings=None, readOnly=False, interactive=False):
batchPath = os.path.abspath(batchPath) + os.sep
# 0: chains, 1: importance sampling, 2: best-fit, 3: best-fit and Hessian
cosmomcAction = 0
if not settings:
if not settingName:
if not pathIsGrid(batchPath):
raise Exception('Need to give name of setting file if batchPath/config does not exist')
readOnly = True
sys.path.insert(0, batchPath + 'config')
sys.modules['batchJob'] = batchjob # old name
settings = __import__(IniFile(batchPath + 'config/config.ini').params['setting_file'].replace('.py', ''))
else:
settings = __import__(settingName, fromlist=['dummy'])
batch = batchjob.batchJob(batchPath, settings.ini_dir)
if hasattr(settings, 'skip'): batch.skip = settings.skip
batch.makeItems(settings, messages=not readOnly)
if readOnly:
for jobItem in [b for b in batch.jobItems]:
if not jobItem.chainExists():
batch.jobItems.remove(jobItem)
batch.save()
print('OK, configured grid with %u existing chains' % (len(batch.jobItems)))
return batch
else:
batch.makeDirectories(settings.__file__)
batch.save()
# priors and widths for parameters which are varied
start_at_bestfit = getattr(settings, 'start_at_bestfit', False)
params = getattr(settings, 'params', default_params)
param_extra = getattr(settings, 'param_extra_opts', default_param_extra_opts)
for jobItem in batch.items(wantSubItems=False):
jobItem.makeChainPath()
ini = IniFile()
for param in jobItem.param_set:
ini.params['param[' + param + ']'] = params[param]
if param_extra is not None and param in param_extra:
ini.params.update(param_extra[param])
if hasattr(settings, 'extra_opts'):
ini.params.update(settings.extra_opts)
ini.params['file_root'] = jobItem.chainRoot
cov_dir_name = getattr(settings, 'cov_dir', 'planck_covmats')
covdir = os.path.join(batch.basePath, cov_dir_name)
covmat = os.path.join(covdir, jobItem.name + '.covmat')
if not os.path.exists(covmat):
covNameMappings = getattr(settings, 'covNameMappings', None)
mapped_name_norm = jobItem.makeNormedName(covNameMappings)[0]
covmat_normed = os.path.join(covdir, mapped_name_norm + '.covmat')
covmat = covmat_normed
if not os.path.exists(covmat) and hasattr(jobItem.data_set,
'covmat'): covmat = batch.basePath + jobItem.data_set.covmat
if not os.path.exists(covmat) and hasattr(settings, 'covmat'): covmat = batch.basePath + settings.covmat
else:
covNameMappings = None
if os.path.exists(covmat):
ini.params['propose_matrix'] = covmat
if getattr(settings, 'newCovmats', True): ini.params['MPI_Max_R_ProposeUpdate'] = 20
else:
hasCov = False
ini.params['MPI_Max_R_ProposeUpdate'] = 20
covmat_try = []
if 'covRenamer' in dir(settings):
covmat_try += settings.covRenamer(jobItem.name)
covmat_try += settings.covRenamer(mapped_name_norm)
if hasattr(settings, 'covrenames'):
for aname in [jobItem.name, mapped_name_norm]:
covmat_try += [aname.replace(old, new, 1) for old, new in settings.covrenames if old in aname]
for new1, old1 in settings.covrenames:
if old1 in aname:
name = aname.replace(old1, new1, 1)
covmat_try += [name.replace(old, new, 1) for old, new in settings.covrenames if old in name]
if 'covWithoutNameOrder' in dir(settings):
if covNameMappings:
removes = copy.deepcopy(covNameMappings)
else:
removes = dict()
for name in settings.covWithoutNameOrder:
if name in jobItem.data_set.names:
removes[name] = ''
covmat_try += [jobItem.makeNormedName(removes)[0]]
covdir2 = os.path.join(batch.basePath, getattr(settings, 'cov_dir_fallback', cov_dir_name))
for name in covmat_try:
covmat = os.path.join(batch.basePath, covdir2, name + '.covmat')
if os.path.exists(covmat):
ini.params['propose_matrix'] = covmat
print('covmat ' + jobItem.name + ' -> ' + name)
hasCov = True
break
if not hasCov: print('WARNING: no matching specific covmat for ' + jobItem.name)
ini.params['start_at_bestfit'] = start_at_bestfit
updateIniParams(ini, jobItem.data_set.params, batch.commonPath)
for deffile in settings.defaults:
ini.defaults.append(batch.commonPath + deffile)
if hasattr(settings, 'override_defaults'):
ini.defaults = [batch.commonPath + deffile for deffile in settings.override_defaults] + ini.defaults
ini.params['action'] = cosmomcAction
ini.saveFile(jobItem.iniFile())
if not start_at_bestfit:
setMinimize(jobItem, ini)
variant = '_minimize'
ini.saveFile(jobItem.iniFile(variant))
# add ini files for importance sampling runs
for imp in jobItem.importanceJobs():
if batch.hasName(imp.name.replace('_post', '')): raise Exception(
'importance sampling something you already have?')
for minimize in (False, True):
if minimize and not getattr(imp, 'want_minimize', True): continue
ini = IniFile()
updateIniParams(ini, imp.importanceSettings, batch.commonPath)
if cosmomcAction == 0 and not minimize:
for deffile in settings.importanceDefaults:
ini.defaults.append(batch.commonPath + deffile)
ini.params['redo_outroot'] = imp.chainRoot
ini.params['action'] = 1
else:
ini.params['file_root'] = imp.chainRoot
if minimize:
setMinimize(jobItem, ini)
variant = '_minimize'
else:
variant = ''
ini.defaults.append(jobItem.iniFile())
ini.saveFile(imp.iniFile(variant))
if cosmomcAction != 0: break
if not interactive: return batch
print('Done... to run do: python python/runbatch.py ' + batchPath)
if not start_at_bestfit:
print('....... for best fits: python python/runbatch.py ' + batchPath + ' --minimize')
print('')
print('for importance sampled: python python/runbatch.py ' + batchPath + ' --importance')
print('for best-fit for importance sampled: python python/runbatch.py ' + batchPath + ' --importance_minimize')
def make_forecast_cmb_dataset(input_cl_file, output_root, output_dir=None, noise_muK_arcmin_T=None,
noise_muK_arcmin_P=None, NoiseVar=None, ENoiseFac=2, fwhm_arcmin=None,
lmin=2, lmax=None, fsky=1, fields_use=None,
lens_recon_noise=None, cl_data_cols=''):
"""
Make a simulated .dataset and associated files with 'data' set at the input fiducial model.
:param input_cl_file: input fiducial CL
:param output_root: root name for output files, e.g. 'my_sim1'
:param output_dir: output directory
:param noise_muK_arcmin_T: temperature noise in muK-arcmin
:param noise_muK_arcmin_P: polarization noise in muK-arcmin
:param NoiseVar: effective isotropic noise variance for the temperature (N_L=NoiseVar with no beam)
:param ENoiseFac: factor by which polarization noise variance is higher (usually 2, for Planck about 4
as only half the detectors polarized)
:param fwhm_arcmin: beam fwhm in arcminutes
:param lmin: l_min
:param lmax: l_max
:param fsky: sky fraction
:param fields_use: optional list of fields to restict to (e.g. 'T E')
:param lens_recon_noise: optional array, starting at L=0, for the PP lensing reconstruction noise, in [L(L+1)]^2C_L^phi/2pi units
:param cl_data_cols: if not specified in file header, order of columns in input CL file (e.g. 'TT TE EE BB PP')
:return:
"""
use_lensing = lens_recon_noise
use_CMB = noise_muK_arcmin_T or NoiseVar is not None
ini = IniFile()
dataset = ini.params
if not cl_data_cols:
cl_data_cols = lastTopComment(input_cl_file)
if not cl_data_cols:
raise Exception('input CL file must specific names of columns (TT TE EE..)')
else:
dataset['cl_hat_order'] = cl_data_cols
if use_CMB:
if NoiseVar is None:
if noise_muK_arcmin_T is None:
raise ValueError('Must specify noise')
NoiseVar = (noise_muK_arcmin_T * np.pi / 180 / 60.) ** 2
if noise_muK_arcmin_P is not None:
ENoiseFac = (noise_muK_arcmin_P / noise_muK_arcmin_T) ** 2
elif noise_muK_arcmin_T is not None or noise_muK_arcmin_P is not None:
raise ValueError('Specific either noise_muK_arcmin or NoiseVar')
if not fields_use:
fields_use = ''
if 'TT' or 'TE' in cl_data_cols: fields_use = 'T'
if 'EE' or 'TE' in cl_data_cols: fields_use += ' E'
if 'BB' in cl_data_cols: fields_use += ' B'
if 'PP' in cl_data_cols and use_lensing: fields_use += ' P'
else:
fields_use = fields_use or 'P'
if output_dir is None:
output_dir = os.path.join(os.path.dirname(__file__), '..', 'data', output_root)
if not os.path.exists(output_dir): os.makedirs(output_dir)
dataset['fields_use'] = fields_use
if use_CMB:
fwhm = fwhm_arcmin / 60
xlc = 180 * np.sqrt(8. * np.log(2.)) / np.pi
sigma2 = (fwhm / xlc) ** 2
noise_cols = 'TT EE BB'
if use_lensing: noise_cols += ' PP'
elif use_lensing:
noise_cols = 'PP'
noise_file = output_root + '_Noise.dat'
with open(os.path.join(output_dir, noise_file), 'w') as f:
f.write('#L %s\n' % noise_cols)
for l in range(lmin, lmax + 1):
NoiseCl = l * (l + 1.) / 2 / np.pi * NoiseVar * np.exp(l * (l + 1) * sigma2)
noises = []
if use_CMB: noises += [NoiseCl, ENoiseFac * NoiseCl, ENoiseFac * NoiseCl]
if use_lensing: noises += [lens_recon_noise[l]]
f.write("%d " % l + " ".join("%E" % elem for elem in noises) + "\n")
dataset['fullsky_exact_fksy'] = fsky
dataset['dataset_format'] = 'CMBLike2'
dataset['like_approx'] = 'exact'
dataset['cl_lmin'] = lmin
dataset['cl_lmax'] = lmax
dataset['binned'] = False
dataset['cl_hat_includes_noise'] = False
shutil.copy(input_cl_file, os.path.join(output_dir, output_root + '.dat'))
dataset['cl_hat_file'] = output_root + '.dat'
dataset['cl_noise_file '] = noise_file
ini.saveFile(os.path.join(output_dir, output_root + '.dataset'))
ini.params['temp_cmb'] = 2.7255
ini.params['CMB_outputscale'] = 2.7255e6 ** 2.
ini.defaults.append('params.ini')
bf = types.BestFit(root + '.minimum', setParamNameFile=root + '.paramnames', want_fixed=True)
for camb, cosmomc in list(pars.items()):
par = bf.parWithName(cosmomc)
if par is not None: ini.params[camb] = par.best_fit
ini.params['scalar_amp(1)'] = float(ini.params['scalar_amp(1)']) / 1e9
nmassive = 1
neffstandard = 3.046 / 3
ini.params['massless_neutrinos'] = float(ini.params['massless_neutrinos']) - neffstandard * nmassive
ini.params['massive_neutrinos'] = int(round(neffstandard * nmassive))
ini.params['nu_mass_degeneracies'] = neffstandard * nmassive
ini.params['share_delta_neff'] = False
ini.params['tensor_spectral_index(1)'] = -float(ini.params['initial_ratio(1)']) / 8
inPars = IniFile(root + '.inputparams')
if inPars.bool('use_nonlinear_lensing', True):
ini.params['do_nonlinear'] = 3
# Note, if you want accurate spectrun on small scales, may need to increase accuracy
ini.saveFile(sys.argv[2])
print('OK, though note this does not support all parameter extensions from LCDM')
