Nucleon transversity from dihadron data and x-dependent lattice QCD constraint

arXiv:2611.12745 · hep-ph · Submitted 2026-11-09 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Phenomenology. The day in particle phenomenology: models, signatures and constraints.

Kai: We're looking at the paper: "Nucleon transversity from dihadron data and x-dependent lattice QCD constraint".

Mira: Here's the gist of the paper. Nucleon transversity from dihadron data and x-dependent lattice QCD constraint is presented in this work to extract collinear transversity distributions and dihadron fragmentation functions through a joint neural-network analysis of various production processes, incorporating lattice QCD constraints to improve flavor decomposition Experimental Access Routes Experimental access to transversity distributions requires convoluting to another chiral-odd quantity, such as the Collins effect in single-hadron semi-inclusive deepinelastic scattering (SIDIS) or azimuthal correlations in nearly back-to-back hadrons from electron–positron annihilation Dihadron production in SIDIS provides access to transversity distributions within the collinear factorization through the chiralodd interference dihadron fragmentation function (DiFF) H∢ q1 (z, Mh), which describes the fragmentation of a transversely polarized quark into a hadron pair Together with the unpolarized DiFF Dq1 (z, Mh), this function enters complementary observables in e+e− annihilation, SIDIS, and polarized pp collisions Transverse SSA in dihadron SIDIS provides flavor-sensitive combinations of transversity distributions and the interference DiFFs Transverse SSA in inclusive dihadron production in p↑p collisions provide complementary constraints through convolutions involving transversity and the interference DiFF Lattice QCD Input Role Lattice QCD provides complementary information through tensor charges, which are the first Mellin moments of the valence transversity distributions gqT(µ) = Z 1/0 dx

h q1 (x, µ) − h q¯1 (x, µ): These lattice determinations have been used as constraints in phenomenological transversity extractions, including simultaneous determinations of transversity and DiFFs Lattice matrix elements of spatially separated quark fields provide access to the x dependence of light-cone PDFs through large-momentum effective theory and related Euclidean approaches At intermediate positive x, the lattice data constrain h u1 − h d1 over a wide range, both overlapping the upper end of existing dihadron SIDIS measurements and extending to larger x Joint Analysis Methodology The study performs a simultaneous analysis of the unpolarized dihadron production cross section and Artru–Collins asymmetry in e+e− annihilation, along with transverse SSAs in dihadron SIDIS and pp collisions Neural networks are used to parametrize the transversity distributions and the DiFFs at an input scale of Q0 = 1 GeV, using hyperbolic-tangent hidden activations in the networks The parametrization includes three architectures for the DiFF network and three architectures for the transversity network The analysis assesses the impact of lattice input by comparing fits with and without it, noting that including the lattice distribution shifts down-valence transversity toward more negative values and enhances up-valence transversity at large x Results and Constraints The fit including the lattice input yields a central dv transversity distribution that exceeds the corresponding reference in magnitude at large x, though with sizable uncertainties The more negative down-valence contribution drives most of the increase in the isovector tensor charge These results demonstrate the complementary role of experimental measurements and lattice information on the x dependence in resolving transversity distributions The fit including lattice input favors a more negative down-valence transversity distribution and enhanced up-valence transversity distribution at large x, while preserving a comparable description of the experimental data The increase in g u−d T is driven mainly by the more negative down-valence contribution, bringing the central values closer to the lattice determinations shown in Fig.

Title and authors: Kai: So, we're looking at a paper called "Nucleon transversity from dihadron data and x-dependent lattice QCD constraint." It sounds like they are trying to figure out the structure of how quarks inside protons are polarized, which is called transversity.

Mira: Exactly. This work brings together two different ways to look at this problem: experimental data from things like dihadron production and lattice quantum chromodynamics calculations. It's trying to build a complete picture of nucleon transversity distributions.

Lev: From where I see it, the core idea is using these constraints to extract the collinear transversity functions and the dihadron fragmentation functions. It’s a joint neural network approach across different production processes like electron-positron annihilation, semi-inclusive deep-inelastic scattering, and polarized proton-proton collisions.

Kai: Right. The experimental access routes they mention involve things like using the Collins effect in single hadron semi inclusive deep inelastic scattering or azimuthal correlations in nearly back to back hadrons from electron positron annihilation dihadron production. It’s about getting transversity distributions through the chiral odd interference dihadron fragmentation function, DiFF, which describes how a transversely polarized quark fragments into a hadron pair.

Mira: And this DiFF goes hand in hand with the unpolarized DiFF, which is what you see in e plus minus annihilation experiments like e plus minus to pi plus pi X. These functions enter complementary observables across different processes like e plus minus annihilation, semi inclusive deep inelastic scattering, and polarized proton proton collisions.

Lev: Then they also look at transverse single spin asymmetries in dihadron SIDIS and transverse single spin asymmetries in inclusive dihadron production in polarized proton-proton collisions to get more constraints through convolutions involving transversity and the interference DiFFs. It's a lot of experimental input they are combining.

Kai: Now, what about lattice QCD? The paper says it provides complementary information through tensor charges, which are the first Mellin moments of the valence transversity distributions g q T mu equals Z over zero dx h q one x mu minus h q bar one x mu.

Mira: Those lattice determinations have been used as constraints in phenomenological transversity extractions, including simultaneous determinations of transversity and DiFFs. They are basically using those lattice numbers to help pin down the experimental results.

Title and authors: Lev: These lattice matrix elements of spatially separated quark fields give access to the x dependence of light cone parton distribution functions through large momentum effective theory and related euclidean approaches. Specifically, at intermediate positive x, they constrain h u one minus h d one over a wide range, which overlaps with the upper end of existing dihadron SIDIS measurements and extends to larger x.

Kai: The methodology they use is a simultaneous analysis of the unpolarized dihadron production cross section and the Artru Collins asymmetry in e plus minus annihilation, along with transverse SSAs in dihadron SIDIS and pp collisions.

Mira: They use neural networks to parametrize the transversity distributions and the DiFFs at an input scale of Q zero equals one giga electron volt using hyperbolic tangent hidden activations. They even have three architectures for the DiFF network and three architectures for the transversity network.

Lev: Then they assess how including lattice input changes things by comparing fits with and without it. The paper notes that including the lattice distribution shifts down valence transversity toward more negative values and enhances up valence transversity at large x.

Kai: So, in results, the fit including the lattice input yields a central down valence transversity distribution that exceeds its reference in magnitude at large x, although with sizable uncertainties. The more negative down valence contribution drives most of the increase in the isovector tensor charge.

Mira: These results show that experimental measurements and lattice information play complementary roles in resolving the x dependence of transversity distributions. They favor a more negative down valence transversity distribution and enhanced up valence transversity distribution at large x, while still describing the experimental data well.

Lev: The increase in g u minus d T is driven mainly by that more negative down valence contribution, bringing the central values closer to the lattice determinations shown in Fig.

Kai: So, what's interesting for us is how this joint analysis helps us see how these different pieces fit together when we look at the data from e plus minus annihilation and those polarized proton-proton collisions. It shows that combining them gives a much more constrained picture than either one alone.

Mira: And it highlights the complementary role of lattice calculations in providing those fundamental constraints on the valence quark structure that are hard to pin down purely from experiments across all momentum fractions.

Title and authors: Lev: The main limitation they point out is that the dihadron SIDIS data available for these analyses only reach x less than or equal to zero point three, whereas the proton proton data sample a broader range of partonic momentum fractions through convolutions involving various flavor channels. Consequently, the large x flavor decomposition and the full moments remain sensitive to the assumed functional form and extrapolation.

Kai: So they are admitting that there's still room for improvement in reaching those very high x regions where we really want to know a lot about transversity.

Mira: And it's a fair point. The paper says that lattice QCD provides constraints, but the way they use those constraints is what drives the shift in the valence distribution toward more negative values and enhanced up valence at large x.

Lev: I think that's where the real physics lies, because it suggests that we need these complementary inputs to move beyond just fitting data points to a simple functional form.

Kai: So looking ahead, what does this mean for us? It means we need more of these joint analyses using AI to handle the complexity of parameterizing those functions across all those processes simultaneously.

Mira: And it suggests that future work should focus on extending the kinematic reach, especially pushing that dihadron SIDIS data to higher x values so we can test these constraints further.

Lev: From an engineering standpoint, incorporating lattice input means we have a better idea of what the required precision is for our experimental measurements because the lattice calculations give us a target for how well we need to constrain those moments.

Kai: So, in summary, this paper on "Nucleon transversity from dihadron data and x-dependent lattice QCD constraint" shows that combining these different experimental data sets with lattice calculations via a neural network gives us a better understanding of the nucleon's spin structure.

Mira: It confirms that the constraints from lattice QCD are really important for resolving the flavor decomposition, especially in how those valence distributions behave at larger momentum fractions.

Lev: And it shows that even with limitations in current data coverage, we can use these constraints to make informed predictions about transversity observables across a wider kinematic range.

The paper's summary: Kai: So, we're looking at how this paper pulls together everything—the experimental data from those dihadron collisions and those lattice QCD calculations—to get a handle on nucleon transversity distributions.

Mira: Right. Basically, they use a joint neural network analysis across different experiments like electron-positron annihilation and polarized proton-proton collisions to extract the collinear transversity functions and the dihadron fragmentation functions at once.

Lev: And they didn't just fit the data; they used lattice QCD calculations of tensor charges as a hard constraint to guide that fitting process, which is really what makes this study unique.

Kai: The main result here is that when you include that lattice input, especially for the down valence transversity distribution, it shifts toward more negative values at large momentum fractions, and it enhances the up valence transversity distribution there.

Mira: That’s significant because it shows that these two inputs—the experimental constraints and the theoretical constraints from the lattice—actually work together to resolve how much flavor content is in those distributions across different momentum scales.

Lev: From a hardware standpoint, this means we need to be very precise when we measure those asymmetries at high x values, because the lattice input is telling us what that structure *should* look like before we even look at the final data.

Kai: So, it confirms that just looking at one type of experiment isn't enough; you need that lattice information to get a stable picture of the x dependence.

Mira: Exactly. The paper shows that this combined approach gives us a better description of the isovector tensor charge than either method could do alone.

Lev: It’s about moving beyond just fitting data points into some kind of real understanding of how these quarks are organized inside the proton at different momenta.

Kai: And while they get a good picture, they also point out that the dihadron SIDIS data has a kinematic limit, meaning we're still extrapolating to reach those very large x values where the structure gets really interesting.

Mira: That limitation is something we have to keep in mind when thinking about what these constraints can predict for new physics beyond just the Standard Model.

Lev: The paper sets up a clear path forward then—we need more high x data, and we need to see how this AI fitting framework handles even higher complexity as we try to push those kinematic limits.

The paper's improvements: Kai: So, to wrap up the paper's suggestions, they aren't just stopping there; they’re proposing ways to push this analysis further using AI and more experimental data.

Mira: Right. They suggest that we need an improved AI system that can handle simultaneous extraction of both the collinear transversity distributions and the dihadron fragmentation functions across all those different production processes.

Lev: The core improvement is building a neural network that can flexibly parametrize those quark and antiquark transversity distributions using inputs for both x and the natural logarithm of x, which lets them map out h u one, h d one, h u bar one, and h d bar one.

Kai: That’s powerful because it means the AI can learn a much richer structure for these functions rather than just fitting a simple shape to the data points.

Mira: And on top of that, they propose a neural network that takes both the momentum fraction z and the hadron mass M h as input to extract those dihadron fragmentation functions with their specific z and M h dependence.

Lev: That’s smart because it links the quark structure directly to how the particles actually fly out in those collisions, giving us a more complete picture of the process.

Kai: They also stress that they should keep incorporating lattice QCD calculations as constraints throughout this process, not just at one point, but dynamically.

Mira: And they want to see how incorporating that lattice input affects the fitting—specifically how it drives the down valence transversity toward those more negative values we talked about earlier.

Lev: It means the next step for building real hardware is to design error correction schemes that can handle this level of complexity and parameterization across so many variables simultaneously.

Kai: So, what this changes for us is that we have a roadmap for how to use advanced AI tools to tackle these complex QCD problems in a more integrated way than before.

Mira: It shifts the focus from just extracting one number to building an entire framework where lattice theory, experimental data, and machine learning all feed into each other.

Lev: This suggests that the overhead for running this kind of complex analysis on real hardware will be significant, but the potential gain in constraint quality is worth looking at.

Conclusion: Kai: So, to wrap up this paper on "Nucleon transversity from dihadron data and x-dependent lattice QCD constraint," they’ve shown that combining those two major inputs gives us a much more reliable picture of how quarks are polarized inside the nucleon.

Mira: Right. The implication is that we can finally start to map out the x dependence of these distributions with a level of certainty that was previously impossible because we were only looking at one side or the other.

Lev: From my side, it means we have a much clearer target for what those high momentum fraction constraints need to look like when we design future experiments.

Kai: It’s about using AI to bridge the gap between what we measure in accelerators and what theory predicts on the lattice.

Mira: Exactly. And they suggest that by pushing our AI models to handle both the collinear distributions and those fragmentation functions at once, we can build a much more complete picture of the nucleon structure.

Lev: It’s a lot of parameterization work, but if that neural network architecture works as described, it should give us better insight into how these different components interact.

Kai: And they point out that while this analysis is strong, the dihadron SIDIS data still has some kinematic limits at very high x values.

Mira: True. So the next big step for this research direction is pushing those experimental boundaries to cover those higher momentum fractions where the effects might be even more pronounced.

Lev: We need better experimental reach, because if we want to use these lattice constraints to make solid predictions about nucleon structure at high x, we have to be able to measure it there.

Kai: So they’ve given us a solid method for combining experiments and theory, but the next hurdle is getting more data where the physics gets really interesting.

Mira: That's right. We need to keep looking at how these constraints hold up when we apply them to other processes like inclusive dihadron production in polarized proton-proton collisions.

Lev: And that’s what I’m looking at next—we want to see if this constraint framework can be applied universally, not just within the specific range they studied here.

Hongxin Dong, Yuan-Yuan Huang, Tianbo Liu, *Xiaomin Shen, Peng Sun, †Xiaonu Xiong, ℶJian-Hui Zhang § Yuxiang Zhao ÉŸYiyu Zhou and Ruilin Zhu

Nanjing Normal University · Institute of Modern Physics, Chinese Academy of Sciences · Key Laboratory of Particle Physics and Particle Irradiation (MOE), Institute of Frontier and Interdisciplinary Science, Shandong University · Southern Center for Nuclear-Science Theory (SCNT), Institute of Modern Physics, Chinese Academy of Sciences · University of Chinese Academy of Sciences · School of Physics and Electronics, Central South University · School of Science and Engineering, The Chinese University of Hong Kong (Shenzhen) · Institute Of Particle Physics, Central China Normal University

hep-ph

Submitted: 2026-11-09

Updated: 2026-11-09

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 89/100

The gist: Nucleon transversity from dihadron data and x-dependent lattice QCD constraint is presented in this work to extract collinear transversity distributions and dihadron fragmentation functions through a…

Key concepts

Nucleon Transversity Distributions
These are fundamental distributions describing how quarks inside a nucleon are polarized when transversely polarized. They are chiral-odd quantities that require specific experimental setups, like dihadron production or the Collins effect in deep-inelastic scattering, to be measured.
Dihadron Fragmentation Functions (DiFF)
DiFFs describe how a transversely polarized quark fragments into a pair of hadrons. They are crucial for accessing transversity distributions through observables in processes like SIDIS and $e^+e^-$ annihilation. The analysis aims to extract these functions simultaneously with the transversity.
Lattice QCD Constraints
Lattice QCD provides theoretical calculations, such as tensor charges, that constrain the Mellin moments of valence transversity distributions. These lattice results are used in the analysis to guide phenomenological extractions, helping to resolve ambiguities in how quark and antiquark transversity behave across different momentum fractions (x).
Joint Neural Network Analysis
A joint neural network is employed to simultaneously parametrize both the transversity distributions and the DiFFs. This AI approach allows for a flexible representation of these complex functions, incorporating input variables like x and $\ln x$, to extract both collinear PDFs and fragmentation functions from combined experimental observables.

Terminology

Summary

Nucleon transversity from dihadron data and x-dependent lattice QCD constraint is presented in this work to extract collinear transversity distributions and dihadron fragmentation functions through a joint neural-network analysis of various production processes, incorporating lattice QCD constraints to improve flavor decomposition

Experimental Access Routes

Experimental access to transversity distributions requires convoluting to another chiral-odd quantity, such as the Collins effect in single-hadron semi-inclusive deepinelastic scattering (SIDIS) or azimuthal correlations in nearly back-to-back hadrons from electron–positron annihilation Dihadron production in SIDIS provides access to transversity distributions within the collinear factorization through the chiralodd interference dihadron fragmentation function (DiFF) H∢ q1 (z, Mh), which describes the fragmentation of a transversely polarized quark into a hadron pair Together with the unpolarized DiFF Dq1 (z, Mh), this function enters complementary observables in e+e− annihilation, SIDIS, and polarized pp collisions Transverse SSA in dihadron SIDIS provides flavor-sensitive combinations of transversity distributions and the interference DiFFs Transverse SSA in inclusive dihadron production in p↑p collisions provide complementary constraints through convolutions involving transversity and the interference DiFF

Lattice QCD Input Role

Lattice QCD provides complementary information through tensor charges, which are the first Mellin moments of the valence transversity distributions gqT(µ) = Z 1/0 dx [h q1 (x, µ) − h q¯1 (x, µ)] These lattice determinations have been used as constraints in phenomenological transversity extractions, including simultaneous determinations of transversity and DiFFs Lattice matrix elements of spatially separated quark fields provide access to the x dependence of light-cone PDFs through large-momentum effective theory and related Euclidean approaches At intermediate positive x, the lattice data constrain h u1 − h d1 over a wide range, both overlapping the upper end of existing dihadron SIDIS measurements and extending to larger x

Joint Analysis Methodology

The study performs a simultaneous analysis of the unpolarized dihadron production cross section and Artru–Collins asymmetry in e+e− annihilation, along with transverse SSAs in dihadron SIDIS and pp collisions Neural networks are used to parametrize the transversity distributions and the DiFFs at an input scale of Q0 = 1 GeV, using hyperbolic-tangent hidden activations in the networks The parametrization includes three architectures for the DiFF network and three architectures for the transversity network The analysis assesses the impact of lattice input by comparing fits with and without it, noting that including the lattice distribution shifts down-valence transversity toward more negative values and enhances up-valence transversity at large x

Results and Constraints

The fit including the lattice input yields a central dv transversity distribution that exceeds the corresponding reference in magnitude at large x, though with sizable uncertainties The more negative down-valence contribution drives most of the increase in the isovector tensor charge These results demonstrate the complementary role of experimental measurements and lattice information on the x dependence in resolving transversity distributions The fit including lattice input favors a more negative down-valence transversity distribution and enhanced up-valence transversity distribution at large x, while preserving a comparable description of the experimental data The increase in g u−d T is driven mainly by the more negative down-valence contribution, bringing the central values closer to the lattice determinations shown in Fig.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI system can perform simultaneous extraction of collinear transversity distributions and dihadron fragmentation functions using a joint neural network analysis across electron-positron annihilation, semi-inclusive deep-inelastic scattering, and polarized proton-proton collisions.

  2. The improved system can incorporate lattice QCD calculations of isovector transversity to constrain up–down flavor differences for quarks and antiquarks over a broad momentum fraction range complementary to data, shifting the down-valence distribution toward more negative values and enhancing up-valence transversity at large x.

  3. The AI system can utilize neural networks parametrized with hyperbolic-tangent hidden activations to flexibly represent quark and antiquark transversity distributions with input variables of x and ln x, allowing for the parametrization of h uv 1, h dv 1, h u¯ 1, and h d¯ 1.

  4. The improved system can simultaneously extract DiFFs with z and Mh dependence by parametrizing the two independent light-quark functions Du 1 and H∢ u 1 using a neural network that takes (z, Mh) as input.

Sources

Related papers