The path to proton structure at 1% accuracy

Abstract

We present a new set of parton distribution functions (PDFs) based on a fully global dataset and machine learning techniques: NNPDF4.0. We expand the NNPDF3.1 determination with 44 new datasets, mostly from the LHC. We derive a novel methodology through hyperparameter optimization, leading to an efficient fitting algorithm built upon stochastic gradient descent. We use NNLO QCD calculations and account for NLO electroweak corrections and nuclear uncertainties. Theoretical improvements in the PDF description include a systematic implementation of positivity constraints and integrability of sum rules. We validate our methodology by means of closure tests and “future tests” (i.e. tests of backward and forward data compatibility), and assess its stability, specifically upon changes of PDF parametrization basis. We study the internal compatibility of our dataset, and investigate the dependence of results both upon the choice of input dataset and of fitting methodology. We perform a first study of the phenomenological implications of NNPDF4.0 on representative LHC processes. The software framework used to produce NNPDF4.0 is made available as an open-source package together with documentation and examples.

NNLO QCD corrections in full colour for jet production observables at the LHC

“…Instead of the PDF set PDF4LHC15 nnlo 100 [73] which was used for the LC-only calculation, we now use NNPDF40 nnlo as 01180 [74] as the default PDF set for all LO, NLO and NNLO coefficients. The single jet inclusive cross section is calculated as a function of p T for seven slices in absolute rapidity:…”

“…They constrain in particular the shape of the gluon distribution [28]. Jet data from LHC single jet inclusive measurements at 7 TeV and 8 TeV were newly included in NNPDF4.0 [74] and the difference in predictions between PDF4LHC15 and NNPDF4.0 highlights their impact.…”

“…We redo these predictions at LC and add the newly computed SLC contributions to obtain the FC predictions. The set-up is the same, except that we use NNPDF40 nnlo as 01180 [74] with α s (M Z ) = 0.118 as PDF instead of the MMHT2014 NNLO PDF sets [81] for all calculations. Jets are reconstructed using the anti-k T algorithm with cone size R = 0.4 in the rapidity range |y| < 3, requiring that the leading jet has a minimum p T of 100 GeV and the subleading jet a minimum p T of 50 GeV.…”

“…Especially the bins with larger values of y b correspond to very asymmetric parton-parton collisions, thereby directly assessing larger values of proton momentum fraction x than in most single and double differential measurements. The impact of various data sets on single inclusive jet production and dijet production at 7 TeV and 8 TeV on the determination of the PDFs has been investigated in detail in [28] and subsequently in the context of the global NNPDF4.0 PDF fit in [74]. In these studies, the CMS 8 TeV triple differential jet production data was found to provide the strongest PDF constraints among all jet data set but to exercise a substantial pull on the gluon distribution around x ∼ 0.3.…”

“…Ironically, the newly determined NNPDF4.0 default PDF set [74] provides a worse description of the CMS 8 TeV triple differential dijet data [80] than the NNPDF3.1 [82] and Comparison of the triple differential dijet distributions at leading colour NNLO⊗NP⊗EW using NNPDF40 nnlo as 01180 (red), NNPDF31 nnlo as 0118 (grey) and MMHT2014 nnlo (green) as the PDF.…”