Fantômas : An analysis of parton distributions in a pion with Bézier parametrizations
Maximiliano Ponce-Chavez · 2025
Read Fantômas : An analysis of parton distributions in a pion with Bézier parametrizations by Maximiliano Ponce-Chavez (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-15657.txt.
Introduction
Lucas Kotz and four co-authors systematically explore how parametrization choices affect pion parton distributions extracted from global data. Parton distributions describe momentum carried by quarks and gluons inside hadrons. Flexible forms are needed, but form choice can bias uncertainty estimates. Their Fantômas tool automates generation of polynomial parametrizations using Bézier curves inside a fitting program.
Applied to the charged pion, the analysis reveals strong correlation between sea quark and gluon distributions. Solutions with vanishing gluon and a large quark sea remain allowed by current data. That degeneracy shows true uncertainties are larger than single parametrization fits suggest. The automated tool lets analysts sample many functional forms. Better uncertainty accounting is vital for hadron structure physics.
Catalog record
| CREATOR | Maximiliano Ponce-Chavez |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 13 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-15657
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fantômas : An analysis of parton distributions in a pion with Bézier parametrizations Maximiliano Ponce-Chavez Abstract We systematically explore the parametrization dependence of the Parton Distribution Functions (PDFs) to better quantify the true uncertainty from global QCD analyses. To achieve this, we employ a novel technique that automates the generation of polynomial parametrizations for PDFs using Bézier curves. This technique is implemented in a C++ module, named Fantômas, which is now included within the xFitter program. As an example, we examine the charged pion PDF. Our analysis reveals that the sea and gluon distributions in the pion are strongly correlated, and PDF solutions featuring a vanishing gluon and a large quark sea are still experimentally allowed. [a]Fredrick Olness The Fantômas Framework for PDF Analyses : We present the Fantômas program [ 1 ] , a C++ package developed to parametrize a variety of Parton Distribution Functions (PDFs) using polynomial shapes realized by Bézier curves.This choice is theoretically grounded: Bézier parametrizations can approximate an arbitrary continuous function with desired accuracy, a consequence of the Stone-Weierstrass approximation theorem. The Fantômas program [ 1 , 2 ] is incorporated into the public code xFitter [ 3 ] . We employed this framework to gain valuable insights into the pion Parton Distribution Functions (PDFs). In contrast to the extensively studied proton PDFs, the pion PDFs are significantly less constrained, and we anticipate they will exhibit substantial parameterization dependence. Thus, the charged pion presents an ideal and timely case for examination using the Fantômas framework. We will use the xFitter study [ 4 ] as our starting point and build upon it to extend the analysis further using the HERA leading-neutron production in deeply inelastic scattering (DIS) as studied by the JAM collaboration [ 5 , 6 ] . The Fantômas Parametric Form : We extend the usual parameterization of the hadron PDFs at the initial scale Q_{0} using the following functional form: \displaystyle x\,f_{i}(x,Q_{0}^{2}) \displaystyle= \displaystyle F_{i}^{car}\times F_{i}^{mod}=\left\{A_{i}x^{B_{i}}(1-x)^{C_{i}}\right\}\times\left\{1+{\cal B}^{(N_{m})}[y(x)]\right\} (1) where, i runs over all partons. F_{i}^{car}(x) is the “carrier function” which contains the normalization coefficients A_{i} , chosen to satisfy the number and momentum sum rules. The powers x^{B_{i}} and (1-x)^{C_{i}} ensure appropriate limits as x\to 0 and 1 . The “modulator function” , F_{i}^{mod}(x) , provides the additional flexibility to explore a broader range of parameterization forms than the usual PDF analyses. In our methodology, the {\cal B} polynomial is chosen to be a Bézier curve. Specifically, {\cal B}^{(N_{m})}[y(x)] is Bézier curves of order N_{m} and the functional argument y(x) allows for a possible rescaling of the x -variable. They differ with usual parametrizations in that their generation can be associated with a set of hyperparameters (N_{m},y(x),\underline{x}) , where \underline{x} represents the abscissa of the control points defining the shape through shifts of the coordinates \delta\underline{P}(\underline{x}) as optimized parameters. In the limit {\cal B}^{(N_{m})}=0 , we recover a traditional baseline parametrization F_{i}^{car}(x) . Fantômas Pion Fit : The xFitter pion PDF analysis included Drell-Yan (DY) pair production by \pi^{-} scattering on a tungsten target by E615 (140 data points) and NA10 (140 data points), as well as prompt-photon ( \gamma ) production in \pi^{-} and \pi^{+} scattering on a tungsten target by WA70 (99 data points). The kinematical coverage of those data is most sensitive to the pion PDFs at large x values. DY processes are most sensitive to valence distributions, with the gluon contributing mainly through the DGLAP evolution. The prompt-photon data provide additional constraints on the gluon distribution above x\gtrsim 0.1 . To better facilitate the separation of the quark sea ( S ) and the gluon ( g ) PDFs at small- x ( x<0.1 ), we followed the proposal of the JAM collaboration [ 7 , 5 ] and included pion Deep Inelastic Scattering (DIS) 29 data points from H1 at HERA. Our choice of data differs largely from JAM’s– JAM’s fit use as much LN data as DY, while Fantômas’s has 10 times more DY data. The PDF Fits : We proceed with an in-depth investigation of the functional form dependencies enabled by the Fantômas methodology. We determine the PDFs at the next-to-leading order (NLO) in the QCD coupling strength \alpha_{s} , and assume the same flavor composition of pion PDFs at the initial scale Q_{0}^{2}=1.9\mbox{ GeV}^{2} as in the xFitter study [ 4 ] . In addition to the gluon PDF g(x) , we introduce the total valence and sea quark distributions, V(x) and S(x) , and determine the PDFs for individual flavors on the assumptions that the PDFs for the two constituent (anti)quarks are the same in \pi^{+} and \pi^{-} , and the light quark sea is flavor-blind. In the fits performed on the complete data set, we independently vary the Bézier curve’s hyperparameters, including the order, which lies between 0 and 3; altogether, the fits presented have a minimum of 5 and a maximum of 13 free parameters. Within the Fantômas framework we generate {\sim}100 baseline PDF sets as outlined in Ref. [ 1 ] . The expected variance of the \chi^{2} for N_{\rm pts}=408 and varying N_{\rm par} between 7 and 13 amounts to \delta\chi^{2}\simeq 30 at the 1\sigma level [ 8 ] ; most of the PDF sets are well within this range. The spread of these baseline PDFs quantifies the parametrization dependence that contributes to the epistemic uncertainty [ 9 ] . We want to combine the epistemic uncertainty (characterized by the parameterization dependence) with the aleatoric uncertainty (characterized by stochastic fluctuations of data) to obtain a total PDF uncertainty. The PDF Uncertainties : In many practical applications with limited parameter dependence, a useful estimate of epistemic uncertainty can often be obtained by sufficiently sampling acceptable PDF models [ 9 ] . This procedure is more direct than introducing either the global or dynamic tolerance. We follow this paradigm in three steps to estimate the total PDF uncertainty. 1) Epistemic Uncertainty: We select five out of {\sim}100 explored baseline PDF configurations that yield a good \chi^{2} and whose central fits collectively cover a broad span of possible PDF magnitudes to estimate the epistemic uncertainty. 2) Aleatory Uncertainty: For each of these five selected models, we generate an uncertainty band using the Hessian method with \Delta\chi^{2}=1 to estimate their aleatory uncertainties. These five PDFs with error bands 1 1 1 The combined ensemble is generated using the mp4lhc package [ 10 ] . For each of the five input Hessian ensembles shown in Fig. 1 , which contain 2N_{\rm eig} eigenvector PDFs, we produce 100 Monte Carlo replicas. These replicas are generated using a linear sampling procedure based on symmetric eigenvector variations. are displayed in Fig. 1 . 3) Combined Uncertainty: Finally, we generate a combined PDF error ensemble from these five bands using the METAPDF method to achieve a more robust total uncertainty estimate [ 10 ] . The METAPDF package was updated for arbitrary hadrons in Ref. [ 11 ] . The resulting ensemble of pion error PDFs is named “ FantoPDF ” and the result for the gluon is displayed in Fig. 2(a) . To summarize, the five fits that form the final FantoPDF ensemble were selected for their distinct shapes, ensuring a broad coverage of the parameter space