pdf

Column production methods related to the PDF weights.

Classes:

PDFWeightOutput(value)	Flag to denote which type of weights to output in the pdf_weights producer.
pdf_weights(*args, **kwargs)
pdf_weights_raw(*args, **kwargs)

class PDFWeightOutput(value)

Bases: StrEnum

Flag to denote which type of weights to output in the pdf_weights producer. Options:

• combined: Produces pdf_weight{,_up,_down}, applying some algorithm per event to combine all weights into

  an up/down variation pair.

• raw: Produces all pdf_weights_{hessian,alphas}.

Attributes:

combined
raw

combined = 'combined'

raw = 'raw'

class pdf_weights(*args, **kwargs)

Bases: Producer

Attributes:

cache_instances
mc_only
pdf_weight_output
uses

Methods:

call_func(events[, invalid_weights_action, ...])	Producer that determines and stores pdf weights with different methods.
init_func(**kwargs)	Default init function.
skip_func(**kwargs)	Default skip function.
update_cls_dict(cls_name, cls_dict, get_attr)

cache_instances = True

call_func(events, invalid_weights_action='raise', outlier_threshold=0.5, outlier_action='ignore', outlier_log_mode='warning', **kwargs)

Producer that determines and stores pdf weights with different methods.

The producer assumes that the nominal entry is always the first LHEPdfWeight value and that the nominal weight is already included in the LHEWeight. It can only be called with MC datasets.

The invalid_weights_action defines the procedure of how to handle events with a an unexpected number of pdf weights (not 101 or 103). Supported modes are: :rtype: Array

• "raise": an exception is raised

• "ignore": nominal weight is set to one, up/down weights too if not storing all weights

  and an empty vector for all weights otherwise

The outlier_action defines the procedure of how to handle events with a pdf uncertainty above the outlier_threshold. Supported modes are:

• "ignore": events are kept unmodified

• "remove": pdf weight nominal/up/down are all set to 0

• "raise": an exception is raised

Additionally, the verbosity of the procedure can be set with outlier_log_mode, which offers the following options:

• "none": no message is given

• "info": a logger.info message is given

• "debug": a logger.debug message is given

• "warning": a logger.warning message is given

Resources:

• https://arxiv.org/pdf/1510.03865.pdf

init_func(**kwargs)

Default init function.

Return type:

None

mc_only: bool = True

pdf_weight_output = 'combined'

skip_func(**kwargs) → bool

Default skip function.

Return type:

bool

static update_cls_dict(cls_name, cls_dict, get_attr)

uses = {'LHEPdfWeight'}

class pdf_weights_raw(*args, **kwargs)

Bases: pdf_weights

Attributes:

cache_instances
pdf_weight_output

Methods:

skip_func(**kwargs)

Default skip function.

cache_instances = True

pdf_weight_output = 'raw'

skip_func(**kwargs) → bool

Default skip function.

Return type:

bool