re_param = pipeline.params['regularisation']
sf_param = pipeline.params['stat_fluctuations']
lt_param = pipeline.params['livetime']
unfold_pipeline_cfg = pipeline.params['unfold_pipeline_cfg']
mean_perpe = []
mean_perbin = []
for idx, lt in enumerate(livetimes):
print '==========='
print 'livetime = {0}'.format(lt)
print '==========='
mean_perpe.append([])
lt_param.value = lt
pipeline.update_params(lt_param)
gen_pipe.update_params(lt_param)
u_pipe = Param(name='unfold_pipeline_cfg',
value=gen_pipe,
is_fixed=True,
prior=None,
range=None)
unfold_pipeline_cfg = u_pipe
pipeline.update_params(unfold_pipeline_cfg)
# Get nominal
re_param.value = 0 * ureg.dimensionless
pipeline.update_params(re_param)
nom_out = pipeline.get_outputs().pop()
def _calculate_fit_coeffs(data,
params,
fit_binning,
nu_params=None,
mu_params=None):
"""
Calculate the fit coefficients for each systematic, flavint,
bin for a polynomial.
"""
logging.debug('Calculating fit coefficients')
config = from_file(params['discr_sys_sample_config'].value)
degree = int(params['poly_degree'].value)
force_through_nominal = params['force_through_nominal'].value
if force_through_nominal:
def fit_func(vals, *poly_coeffs):
return np.polynomial.polynomial.polyval(
vals, [1.] + list(poly_coeffs))
else:
def fit_func(vals, *poly_coeffs):
return np.polynomial.polynomial.polyval(
vals, list(poly_coeffs))
# add free param for constant term
degree += 1
template_maker = Pipeline(params['pipeline_config'].value)
dataset_param = template_maker.params['dataset']
def parse(string):
return string.replace(' ', '').split(',')
sys_fit_coeffs = OrderedDict()
if nu_params is not None:
sys_list = parse(config.get('neutrinos', 'sys_list'))
nu_params = deepcopy(map(lambda x: x[3:], nu_params))
if set(nu_params) != set(sys_list):
raise AssertionError(
'Systematics list listed in the sample config file does '
'not match the params in the pipeline config file\n {0} '
'!= {1}'.format(set(nu_params), set(sys_list)))
for sys in sys_list:
ev_sys = 'neutrinos|' + sys
runs = parse(config.get(ev_sys, 'runs')[1:-1])
nominal = config.get(ev_sys, 'nominal')
mapset_dict = OrderedDict()
flavint_groups = None
for run in runs:
logging.info('Loading run {0} of systematic '
'{1}'.format(run, sys))
dataset_param.value = ev_sys + '|' + run
template_maker.update_params(dataset_param)
template = template_maker.get_outputs(
idx=int(params['stop_after_stage'].m))
if not isinstance(template, Data):
raise AssertionError(
'Template output is not a Data object, instead is '
'type {0}'.format(type(template)))
if flavint_groups is None:
flavint_groups = template.flavint_groups
else:
if set(flavint_groups) != set(template.flavint_groups):
raise AssertionError(
'Mismatch of flavint_groups - ({0}) does not '
'match flavint_groups '
'({1})'.format(flavint_groups,
template.flavint_groups))
outputs = []
for fig in template.keys():
outputs.append(
template.histogram(kinds=fig,
binning=fit_binning,
weights_col='pisa_weight',
errors=False,
name=str(NuFlavIntGroup(fig))))
mapset_dict[run] = MapSet(outputs, name=run)
nom_mapset = mapset_dict[nominal]
fracdiff_mapset_dict = OrderedDict()
for run in runs:
mapset = []
for flavintg_map in mapset_dict[run]:
# TODO(shivesh): error propagation?
flavintg = flavintg_map.name
mask = ~(nom_mapset[flavintg].hist == 0.)
div = np.zeros(flavintg_map.shape)
with np.errstate(divide='ignore', invalid='ignore'):
div[mask] = \
unp.nominal_values(flavintg_map.hist[mask]) /\
unp.nominal_values(nom_mapset[flavintg].hist[mask])
mapset.append(
Map(name=flavintg,
binning=flavintg_map.binning,
hist=div))
fracdiff_mapset_dict[run] = MapSet(mapset)
delta_runs = np.array([float(x)
for x in runs]) - float(nominal)
coeff_binning = OneDimBinning(name='coeff',
num_bins=degree,
is_lin=True,
domain=[-1, 1])
combined_binning = fit_binning + coeff_binning
params_mapset = []
for fig in template.keys():
# TODO(shivesh): Fix numpy warning on this line
pvals_hist = np.empty(map(int, combined_binning.shape),
dtype=object)
hists = [
fracdiff_mapset_dict[run][fig].hist for run in runs
]
zip_hists = np.dstack(hists)
for idx in np.ndindex(fit_binning.shape):
y_values = []
y_sigma = []
for run in fracdiff_mapset_dict:
y_values.append(
unp.nominal_values(
fracdiff_mapset_dict[run][fig].hist[idx]))
y_sigma.append(
unp.std_devs(
fracdiff_mapset_dict[run][fig].hist[idx]))
if np.any(y_sigma):
popt, pcov = curve_fit(fit_func,
delta_runs,
y_values,
sigma=y_sigma,
p0=np.ones(degree))
else:
popt, pcov = curve_fit(fit_func,
delta_runs,
y_values,
p0=np.ones(degree))
# perr = np.sqrt(np.diag(pcov))
# pvals = unp.uarray(popt, perr)
pvals_hist[idx] = popt
pvals_hist = np.array(pvals_hist.tolist())
params_mapset.append(
Map(name=fig,
binning=combined_binning,
hist=pvals_hist))
params_mapset = MapSet(params_mapset, name=sys)
if sys in sys_fit_coeffs:
sys_fit_coeffs[sys] = MapSet(
[sys_fit_coeffs[sys], params_mapset])
else:
sys_fit_coeffs[sys] = params_mapset
if mu_params is not None:
sys_list = parse(config.get('muons', 'sys_list'))
mu_params = deepcopy(map(lambda x: x[3:], mu_params))
if set(mu_params) != set(sys_list):
raise AssertionError(
'Systematics list listed in the sample config file does '
'not match the params in the pipeline config file\n {0} '
'!= {1}'.format(set(mu_params), set(sys_list)))
for sys in sys_list:
ev_sys = 'muons|' + sys
runs = parse(config.get(ev_sys, 'runs')[1:-1])
nominal = config.get(ev_sys, 'nominal')
map_dict = OrderedDict()
flavint_groups = None
for run in runs:
logging.info('Loading run {0} of systematic '
'{1}'.format(run, sys))
dataset_param.value = ev_sys + '|' + run
template_maker.update_params(dataset_param)
template = template_maker.get_outputs(
idx=int(params['stop_after_stage'].m))
if not isinstance(template, Data):
raise AssertionError(
'Template output is not a Data object, instead is '
'type {0}'.format(type(template)))
if not template.contains_muons:
raise AssertionError(
'Template output does not contain muons')
output = template.histogram(
kinds='muons',
binning=fit_binning,
# NOTE: weights cancel in fraction
weights_col=None,
errors=False,
name='muons')
map_dict[run] = output
nom_map = map_dict[nominal]
fracdiff_map_dict = OrderedDict()
for run in runs:
mask = ~(nom_map.hist == 0.)
div = np.zeros(nom_map.shape)
with np.errstate(divide='ignore', invalid='ignore'):
div[mask] = \
unp.nominal_values(map_dict[run].hist[mask]) /\
unp.nominal_values(nom_map.hist[mask])
fracdiff_map_dict[run] = Map(name='muons',
binning=nom_map.binning,
hist=div)
delta_runs = np.array([float(x)
for x in runs]) - float(nominal)
coeff_binning = OneDimBinning(name='coeff',
num_bins=degree,
is_lin=True,
domain=[-1, 1])
combined_binning = fit_binning + coeff_binning
pvals_hist = np.empty(map(int, combined_binning.shape),
dtype=object)
hists = [fracdiff_map_dict[run].hist for run in runs]
zip_hists = np.dstack(hists)
for idx in np.ndindex(fit_binning.shape):
y_values = []
y_sigma = []
for run in fracdiff_mapset_dict:
y_values.append(
unp.nominal_values(
fracdiff_mapset_dict[run][fig].hist[idx]))
y_sigma.append(
unp.std_devs(
fracdiff_mapset_dict[run][fig].hist[idx]))
if np.any(y_sigma):
popt, pcov = curve_fit(fit_func,
delta_runs,
y_values,
sigma=y_sigma,
p0=np.ones(degree))
else:
popt, pcov = curve_fit(fit_func,
delta_runs,
y_values,
p0=np.ones(degree))
# perr = np.sqrt(np.diag(pcov))
# pvals = unp.uarray(popt, perr)
pvals_hist[idx] = popt
pvals_hist = np.array(pvals_hist.tolist())
params_map = Map(name='muons',
binning=combined_binning,
hist=pvals_hist)
if sys in sys_fit_coeffs:
sys_fit_coeffs[sys] = MapSet(
[sys_fit_coeffs[sys], params_map])
else:
sys_fit_coeffs[sys] = params_map
return sys_fit_coeffs
def __init__(self,
pipelines,
label=None,
set_livetime_from_data=True,
profile=False):
self.label = label
self._source_code_hash = None
self.metadata = OrderedDict()
self._profile = profile
self._pipelines = []
if isinstance(pipelines,
(str, PISAConfigParser, OrderedDict, Pipeline)):
pipelines = [pipelines]
for pipeline in pipelines:
if not isinstance(pipeline, Pipeline):
pipeline = Pipeline(pipeline, profile=profile)
self._pipelines.append(pipeline)
data_run_livetime = None
if set_livetime_from_data:
#
# Get livetime metadata if defined in any stage in any pipeline
#
for pipeline_idx, pipeline in enumerate(self):
for stage_idx, stage in enumerate(pipeline):
if not (hasattr(stage, "metadata")
and isinstance(stage.metadata, Mapping)
and "livetime" in stage.metadata):
continue
if data_run_livetime is None:
data_run_livetime = stage.metadata["livetime"]
if stage.metadata["livetime"] != data_run_livetime:
raise ValueError(
"Pipeline index {}, stage index {} has data"
" livetime = {}, in disagreement with"
" previously-found livetime = {}".format(
pipeline_idx,
stage_idx,
stage.metadata["livetime"],
data_run_livetime,
))
# Save the last livetime found inside the pipeline's metadata
# TODO: implement metadata in the pipeline class instead
self.metadata['livetime'] = data_run_livetime
#
# Set param `params.livetime` for any pipelines that have it
#
if data_run_livetime is not None:
data_run_livetime *= ureg.sec
for pipeline_idx, pipeline in enumerate(self):
if "livetime" not in pipeline.params.names:
continue
pipeline.params.livetime.is_fixed = True
if pipeline.params.livetime != data_run_livetime:
logging.warning(
"Pipeline index %d has params.livetime = %s, in"
" disagreement with data livetime = %s defined by"
" data. The pipeline's livetime param will be"
" reset to the latter value and set to be fixed"
" (if it is not alredy).",
pipeline_idx,
pipeline.params.livetime.value,
data_run_livetime,
)
pipeline.params.livetime = data_run_livetime
#for pipeline in self:
# pipeline.select_params(self.param_selections,
# error_on_missing=False)
# Make sure that all the pipelines have the same detector name (or None)
self.detector_name = 'no_name'
for p in self._pipelines:
name = p.detector_name
if name != self.detector_name and self.detector_name != 'no_name':
raise NameError(
'Different detector names in distribution_maker pipelines')
self.detector_name = name
# set parameters with an identical name to the same object
# otherwise we get inconsistent behaviour when setting repeated params
# See Isues #566 and #648
# Also, do this for all selections!
original_selection = self.param_selections
all_selections = set()
for pipeline in self:
for stage in pipeline.stages:
all_selections.update(
stage._param_selector._selector_params.keys())
for selection in all_selections:
self.select_params(selection)
all_params = self.params
for pipeline in self:
pipeline.update_params(all_params,
existing_must_match=True,
extend=False)
self.select_params(original_selection)
def main():
args = parse_args()
set_verbosity(args.v)
if args.plot:
import matplotlib as mpl
mpl.use('pdf')
import matplotlib.pyplot as plt
from pisa.utils.plotter import Plotter
cfg = from_file(args.fit_settings)
sys_list = cfg.get('general', 'sys_list').replace(' ', '').split(',')
stop_idx = cfg.getint('general', 'stop_after_stage')
for sys in sys_list:
# Parse info for given systematic
nominal = cfg.getfloat(sys, 'nominal')
degree = cfg.getint(sys, 'degree')
force_through_nominal = cfg.getboolean(sys, 'force_through_nominal')
runs = eval(cfg.get(sys, 'runs'))
#print "runs ", runs
smooth = cfg.get(sys, 'smooth')
x_values = np.array(sorted(runs))
# Build fit function
if force_through_nominal:
function = "lambda x, *p: np.polynomial.polynomial.polyval(x, [1.] + list(p))"
else:
function = "lambda x, *p: np.polynomial.polynomial.polyval(x, list(p))"
# Add free parameter for constant term
degree += 1
fit_fun = eval(function)
# Instantiate template maker
template_maker = Pipeline(args.template_settings)
if not args.set_param == '':
for one_set_param in args.set_param:
p_name, value = one_set_param.split("=")
#print "p_name,value= ", p_name, " ", value
value = parse_quantity(value)
value = value.n * value.units
param = template_maker.params[p_name]
#print "old ", p_name, "value = ", param.value
param.value = value
#print "new ", p_name, "value = ", param.value
template_maker.update_params(param)
inputs = {}
map_names = None
# Get sys templates
for run in runs:
for key, val in cfg.items('%s:%s'%(sys, run)):
if key.startswith('param.'):
_, pname = key.split('.')
param = template_maker.params[pname]
try:
value = parse_quantity(val)
param.value = value.n * value.units
except ValueError:
value = parse_string_literal(val)
param.value = value
param.set_nominal_to_current_value()
template_maker.update_params(param)
# Retreive maps
template = template_maker.get_outputs(idx=stop_idx)
if map_names is None: map_names = [m.name for m in template]
inputs[run] = {}
for m in template:
inputs[run][m.name] = m.hist
# Numpy acrobatics:
arrays = {}
for name in map_names:
arrays[name] = []
for x in x_values:
arrays[name].append(
inputs[x][name] / unp.nominal_values(inputs[nominal][name])
)
a = np.array(arrays[name])
arrays[name] = np.rollaxis(a, 0, len(a.shape))
# Shift to get deltas
x_values -= nominal
# Binning object (assuming they're all the same)
binning = template.maps[0].binning
shape = [d.num_bins for d in binning] + [degree]
shape_small = [d.num_bins for d in binning]
outputs = {}
errors = {}
for name in map_names:
# Now actualy perform some fits
outputs[name] = np.ones(shape)
errors[name] = np.ones(shape)
for idx in np.ndindex(*shape_small):
y_values = unp.nominal_values(arrays[name][idx])
y_sigma = unp.std_devs(arrays[name][idx])
if np.any(y_sigma):
popt, pcov = curve_fit(fit_fun, x_values, y_values,
sigma=y_sigma, p0=np.ones(degree))
else:
popt, pcov = curve_fit(fit_fun, x_values, y_values,
p0=np.ones(degree))
perr = np.sqrt(np.diag(pcov))
for k, p in enumerate(popt):
outputs[name][idx][k] = p
errors[name][idx][k] = perr[k]
# TODO(philippeller): the below block of code will fail
# Maybe plot
#if args.plot:
# fig_num = i + nx * j
# if fig_num == 0:
# fig = plt.figure(num=1, figsize=( 4*nx, 4*ny))
# subplot_idx = nx*(ny-1-j)+ i + 1
# plt.subplot(ny, nx, subplot_idx)
# #plt.snameter(x_values, y_values, color=plt_colors[name])
# plt.gca().errorbar(x_values, y_values, yerr=y_sigma,
# fmt='o', color=plt_colors[name],
# ecolor=plt_colors[name],
# mec=plt_colors[name])
# # Plot nominal point again in black
# plt.snameter([0.0], [1.0], color='k')
# f_values = fit_fun(x_values, *popt)
# fun_plot, = plt.plot(x_values, f_values,
# color=plt_colors[name])
# plt.ylim(np.min(unp.nominal_values(arrays[name]))*0.9,
# np.max(unp.nominal_values(arrays[name]))*1.1)
# if i > 0:
# plt.setp(plt.gca().get_yticklabels(), visible=False)
# if j > 0:
# plt.setp(plt.gca().get_xticklabels(), visible=False)
if smooth == 'gauss':
for name in map_names:
for d in range(degree):
outputs[name][...,d] = gaussian_filter(outputs[name][...,d],sigma=1)
if smooth == 'gauss_pid':
for name in map_names:
split_idx = binning.names.index('pid')
tot = len(binning)-1
for d in range(degree):
for p in range(len(binning['pid'])):
outputs[name][...,p,d] = gaussian_filter(
np.swapaxes(outputs[name], split_idx, tot)[...,p,d],
sigma=1
)
outputs[name] = np.swapaxes(outputs[name], split_idx, tot)
# Save the raw ones anyway
outputs['pname'] = sys
outputs['nominal'] = nominal
outputs['function'] = function
outputs['map_names'] = map_names
outputs['binning_hash'] = binning.hash
to_file(outputs, '%s/%s_sysfits_%s_%s.json'%(args.out_dir, sys,
args.tag, smooth))
if args.plot:
for d in range(degree):
maps = []
for name in map_names:
maps.append(Map(name='%s_raw'%name, hist=outputs[name][...,d],
binning=binning))
maps = MapSet(maps)
my_plotter = Plotter(
stamp='',
outdir=args.out_dir,
fmt='pdf',
log=False,
label=''
)
my_plotter.plot_2d_array(
maps,
fname='%s_%s_%s_%s'%(sys, args.tag, d, smooth),
)
def main():
global SIGMA
args = vars(parse_args())
set_verbosity(args.pop('v'))
center_zero = args.pop('center_zero')
make_pdf = False
if args['pdf']:
make_pdf = True
args['pdf'] = False
outdir = args.pop('outdir')
fileio.mkdir(outdir, mode=0755)
SIGMA *= args.pop('sigma')
cfx_pipe = Pipeline(args.pop('cfx_pipeline'))
signal = args.pop('signal').replace(' ', '').split(',')
output_str = []
for name in signal:
if 'muons' in name or 'noise' in name:
raise AssertionError('Are you trying to unfold muons/noise?')
elif 'all_nu' in name:
output_str = [str(NuFlavIntGroup(f)) for f in ALL_NUFLAVINTS]
else:
output_str.append(NuFlavIntGroup(name))
output_str = [str(f) for f in output_str]
cfx_pipe._output_names = output_str
# Turn off stat fluctuations
stat_param = cfx_pipe.params['stat_fluctuations']
stat_param.value = 0 * ureg.dimensionless
cfx_pipe.update_params(stat_param)
# Get nominal Map
re_param = cfx_pipe.params['regularisation']
re_param.value = 0 * ureg.dimensionless
cfx_pipe.update_params(re_param)
nom_out = cfx_pipe.get_outputs()
re_param.reset()
cfx_pipe.update_params(re_param)
params = ParamSet()
for param in cfx_pipe.params:
if param.name != 'dataset':
params.extend(param)
free = params.free
logging.info('Free params = {0}'.format(free))
contin = True
for f in free:
if 'hole_ice' not in f.name and 'dom_eff' not in f.name:
continue
# if 'atm_muon_scale' in f.name:
# contin = False
# if contin:
# continue
logging.info('Working on parameter {0}'.format(f.name))
if f.prior.kind != 'uniform':
# Use deltaLLH = SIGMA to define +/- sigma for non-uniform
scan_over = np.linspace(*f.range, num=1000) * f.range[0].u
llh = f.prior.llh(scan_over)
dllh = llh - np.min(-llh)
mllh_idx = np.argmin(-llh)
if mllh_idx == 0:
l_sig_idx = 0
else:
l_sig_idx = np.argmin(np.abs(dllh[:mllh_idx] - SIGMA))
u_sig_idx = np.argmin(np.abs(dllh[mllh_idx:] - SIGMA)) + mllh_idx
l_sigma = scan_over[l_sig_idx]
u_sigma = scan_over[u_sig_idx]
else:
l_sigma = f.range[0]
u_sigma = f.range[1]
logging.info('Setting {0} lower sigma bound to '
'{1}'.format(f.name, l_sigma))
f.value = l_sigma
cfx_pipe.update_params(f)
l_out = cfx_pipe.get_outputs()
logging.info('Setting {0} upper sigma bound to '
'{1}'.format(f.name, u_sigma))
f.value = u_sigma
cfx_pipe.update_params(f)
u_out = cfx_pipe.get_outputs()
f.reset()
cfx_pipe.update_params(f)
f_outdir = outdir + '/' + f.name
l_outdir = f_outdir + '/' + 'lower'
u_outdir = f_outdir + '/' + 'upper'
fileio.mkdir(f_outdir)
fileio.mkdir(l_outdir)
fileio.mkdir(u_outdir)
compare(outdir=l_outdir,
ref=MapSet([nom_out]),
ref_label='baseline',
test=MapSet([l_out]),
test_label=r'-sigma',
**args)
compare(outdir=u_outdir,
ref=MapSet([nom_out]),
ref_label='baseline',
test=MapSet([u_out]),
test_label=r'+sigma',
**args)
l_in_mapset = l_outdir + '/' + 'fract_diff__-sigma___baseline.json.bz2'
u_in_mapset = u_outdir + '/' + 'fract_diff__+sigma___baseline.json.bz2'
l_in_map = MapSet.from_json(l_in_mapset).pop() * 100.
u_in_map = MapSet.from_json(u_in_mapset).pop() * 100.
if make_pdf:
outfile = f_outdir + '/systematic_effect.pdf'
else:
outfile = f_outdir + '/systematic_effect.png'
title = r'% effect on ' + r'${0}$'.format(l_in_map.tex) + \
' event counts for {0} parameter'.format(f.name)
sub_titles = (r'(-\sigma - {\rm baseline}) \:/\: {\rm baseline}',
r'(+\sigma - {\rm baseline}) \:/\: {\rm baseline}')
make_plot(
maps=(l_in_map, u_in_map),
outfile=outfile,
logv=False,
center_zero=center_zero,
vlabel=r'({\rm change} - {\rm baseline}) \:/\: {\rm baseline} (%)',
title=title,
sub_titles=sub_titles)