import logging
import pyhf
from pyhf import events
from pyhf.exceptions import InvalidModifier
from pyhf.parameters import ParamViewer
from pyhf.tensor.manager import get_backend
log = logging.getLogger(__name__)
[docs]
def required_parset(sigmas, fixed: list[bool]):
n_parameters = len(sigmas)
return {
"paramset_type": "constrained_by_normal",
"n_parameters": n_parameters,
"is_scalar": False,
"inits": (1.0,) * n_parameters,
"bounds": ((1e-10, 10.0),) * n_parameters,
"fixed": tuple(fixed),
"auxdata": (1.0,) * n_parameters,
"sigmas": tuple(sigmas),
}
[docs]
class staterror_builder:
"""Builder class for collecting staterror modifier data"""
is_shared = True
[docs]
def __init__(self, config):
self.builder_data = {}
self.config = config
self.required_parsets = {}
[docs]
def collect(self, thismod, nom):
uncrt = thismod["data"] if thismod else [0.0] * len(nom)
mask = [bool(thismod)] * len(nom)
return {"mask": mask, "nom_data": nom, "uncrt": uncrt}
[docs]
def append(self, key, channel, sample, thismod, defined_samp):
self.builder_data.setdefault(key, {}).setdefault(sample, {}).setdefault(
"data", {"uncrt": [], "nom_data": [], "mask": []}
)
nom = (
defined_samp["data"]
if defined_samp
else [0.0] * self.config.channel_nbins[channel]
)
moddata = self.collect(thismod, nom)
self.builder_data[key][sample]["data"]["mask"].append(moddata["mask"])
self.builder_data[key][sample]["data"]["uncrt"].append(moddata["uncrt"])
self.builder_data[key][sample]["data"]["nom_data"].append(moddata["nom_data"])
[docs]
def finalize(self):
default_backend = pyhf.default_backend
for modifier_name, modifier in self.builder_data.items():
for sample_name, sample in modifier.items():
sample["data"]["mask"] = default_backend.concatenate(
sample["data"]["mask"]
)
sample["data"]["uncrt"] = default_backend.concatenate(
sample["data"]["uncrt"]
)
sample["data"]["nom_data"] = default_backend.concatenate(
sample["data"]["nom_data"]
)
if len(sample["data"]["nom_data"]) != len(sample["data"]["uncrt"]):
_modifier_type, _modifier_name = modifier_name.split("/")
_sample_data_len = len(sample["data"]["nom_data"])
_uncrt_len = len(sample["data"]["uncrt"])
msg = (
f"The '{sample_name}' sample {_modifier_type} modifier"
f" '{_modifier_name}' has data shape inconsistent with the sample.\n"
f"{sample_name} has 'data' of length {_sample_data_len} but {_modifier_name}"
f" has 'data' of length {_uncrt_len}."
)
raise InvalidModifier(msg)
for modname in self.builder_data:
parname = modname.split("/")[1]
nomsall = default_backend.sum(
[
modifier_data["data"]["nom_data"]
for modifier_data in self.builder_data[modname].values()
if default_backend.astensor(modifier_data["data"]["mask"]).any()
],
axis=0,
)
relerrs = default_backend.sum(
[
[
(
(modifier_data["data"]["uncrt"][binnr] / nomsall[binnr])
** 2
if nomsall[binnr] > 0
else 0.0
)
for binnr in range(len(modifier_data["data"]["nom_data"]))
]
for modifier_data in self.builder_data[modname].values()
],
axis=0,
)
# here relerrs still has all the bins, while the staterror are usually per-channel
# so we need to pick out the masks for this modifier to extract the
# modifier configuration (sigmas, etc..)
# so loop over samples and extract the first mask
# while making sure any subsequent mask is consistent
relerrs = default_backend.sqrt(relerrs)
masks = {}
for modifier_data in self.builder_data[modname].values():
mask_this_sample = default_backend.astensor(
modifier_data["data"]["mask"], dtype="bool"
)
if mask_this_sample.any():
if modname not in masks:
masks[modname] = mask_this_sample
else:
assert (mask_this_sample == masks[modname]).all()
# extract sigmas using this modifiers mask
sigmas = relerrs[masks[modname]]
# list of bools, consistent with other modifiers (no numpy.bool_)
fixed = default_backend.tolist(sigmas == 0)
# FIXME: sigmas that are zero will be fixed to 1.0 arbitrarily to ensure
# non-Nan constraint term, but in a future PR need to remove constraints
# for these
sigmas[fixed] = 1.0
self.required_parsets.setdefault(parname, [required_parset(sigmas, fixed)])
return self.builder_data
[docs]
class staterror_combined:
name = "staterror"
op_code = "multiplication"
[docs]
def __init__(self, modifiers, pdfconfig, builder_data, batch_size=None):
default_backend = pyhf.default_backend
self.batch_size = batch_size
keys = [f"{mtype}/{m}" for m, mtype in modifiers]
self._staterr_mods = [m for m, _ in modifiers]
parfield_shape = (self.batch_size or 1, pdfconfig.npars)
self.param_viewer = ParamViewer(
parfield_shape, pdfconfig.par_map, self._staterr_mods
)
self._staterror_mask = [
[[builder_data[m][s]["data"]["mask"]] for s in pdfconfig.samples]
for m in keys
]
global_concatenated_bin_indices = [
[[j for c in pdfconfig.channels for j in range(pdfconfig.channel_nbins[c])]]
]
self._access_field = default_backend.tile(
global_concatenated_bin_indices,
(len(self._staterr_mods), self.batch_size or 1, 1),
)
self._reindex_access_field(pdfconfig)
self._precompute()
events.subscribe("tensorlib_changed")(self._precompute)
[docs]
def _reindex_access_field(self, _pdfconfig):
default_backend = pyhf.default_backend
for syst_index, syst_access in enumerate(self._access_field):
singular_sample_index = [
idx
for idx, syst in enumerate(
default_backend.astensor(self._staterror_mask)[syst_index, :, 0]
)
if any(syst)
][-1]
for batch_index, batch_access in enumerate(syst_access):
selection = self.param_viewer.index_selection[syst_index][batch_index]
access_field_for_syst_and_batch = default_backend.zeros(
len(batch_access)
)
sample_mask = self._staterror_mask[syst_index][singular_sample_index][0]
access_field_for_syst_and_batch[sample_mask] = selection
self._access_field[syst_index, batch_index] = (
access_field_for_syst_and_batch
)
[docs]
def _precompute(self):
if not self.param_viewer.index_selection:
return
tensorlib, _ = get_backend()
self.staterror_mask = tensorlib.astensor(self._staterror_mask, dtype="bool")
self.staterror_mask = tensorlib.tile(
self.staterror_mask, (1, 1, self.batch_size or 1, 1)
)
self.access_field = tensorlib.astensor(self._access_field, dtype="int")
self.sample_ones = tensorlib.ones(tensorlib.shape(self.staterror_mask)[1])
self.staterror_default = tensorlib.ones(tensorlib.shape(self.staterror_mask))
[docs]
def apply(self, pars):
if not self.param_viewer.index_selection:
return None
tensorlib, _ = get_backend()
flat_pars = pars if self.batch_size is None else tensorlib.reshape(pars, (-1,))
statfactors = tensorlib.gather(flat_pars, self.access_field)
results_staterr = tensorlib.einsum("mab,s->msab", statfactors, self.sample_ones)
return tensorlib.where(
self.staterror_mask, results_staterr, self.staterror_default
)