# Reweighting

Up to now, when new data is released, full refit needed:

→ tune parametrisation and statistic treatment

→ can be done only by PDF fitting collaborations themselves

We define the reweighting method:

→ it is based on Bayes’ theorem

→ no need to refit

→ anybody can produce a new set of PDF replicas

Bayes’ theorem can be stated in terms of probability densities:

where f are PDFs and y data.

Applying the theorem we can define weights like this:

with

These weights allow for an inclusion of new data in the fit by updating the probability distribution.

The NNPDF2.2 parton fit includes through reweighting D0, ATLAS and CMS charge asymmetry datasets.