pyhf#

The pure Python package pyhf is a statistical fitting package for high energy physics for purely binned, templated fits. It is a Python implementation of the HistFactory schema. The connection to zfit is done via the loss function, which can be created from a pyhf model.

Import the module with:

import zfit_physics.pyhf as zpyhf

The loss function can be created from a data and a pdf model with

nll = zpyhf.loss.nll_from_pyhf(data, pdf)

which optionally takes already created ZfitParameter as arguments.

Loss#

zfit_physics.pyhf.loss.nll_from_pyhf(data, pdf, *, params=None, init_pars=None, par_bounds=None, fixed_params=None, errordef=None)[source]#

Create a zfit loss function from a pyhf pdf and data.

Parameters:

  • data – Data to be used in the negative log-likelihood (NLL) calculation.

  • pdf – Probability density function model.

  • params – Optional; Model parameters for the fit. If None, initial parameters, bounds, and fixed status are recommended by the pdf.

  • init_pars – Optional; Initial parameter values. Ignored if params is provided.

  • par_bounds – Optional; Parameter bounds. Ignored if params is provided.

  • fixed_params – Optional; Boolean indicators of parameter fixed statuses. Ignored if params is provided.

  • errordef – Optional; Error definition, by default set to 0.5.