Speaker
Prof.
liuti simonetta
Description
I will discuss the application of an alternative type of neural network, the Self-Organizing
Maps (SOMs), to extract parton distribution functions from various hard scattering processes.
SOMs provide a complementary algorithm to NNPDFs yielding a parametrization that is free
from the bias implicit in choosing specific analytic forms. At the same time it enables us to
extrapolate to kinematical regions where data are not available. I will show in particular the
extraction using SOMs of the ratio d/u in the x=1 limit, including the treatment of nuclear effects.
Author
Simonetta Liuti
(University of Virginia)
