The impact of prescriptions in phenomenological extractions of Transverse Momentum Dependent distributions

This paper demonstrates that the choice of $b_*$ prescription in the Collins-Soper-Sterman framework introduces a significant intrinsic theoretical uncertainty in Transverse Momentum Dependent distribution extractions, yielding consistent results for low-energy data but causing substantial discrepancies in intermediate-to-high energy predictions.

The Gist

Imagine you are trying to understand the shape of a cloud. You can't see the individual water droplets clearly, so you use a special camera (a mathematical formula) to take a picture. But this camera has a glitch: if you try to zoom in too close to the center of the cloud, the lens gets blurry and the math breaks down.

To fix this, scientists invented a "patch" called the $b^*$ prescription. It's like putting a piece of tape over the blurry part of the lens. The tape works well enough to take a decent photo, but here's the catch: different scientists use different kinds of tape. Some use clear tape, some use thick tape, and some tape is applied in slightly different ways.

This paper by Matteo Cerutti and Andrea Simonelli asks a simple but crucial question: Does the type of tape we use actually change what we think the cloud looks like?

The Cloud and the Camera (The Physics)

In the world of particle physics, scientists study Transverse Momentum Dependent (TMD) distributions. Think of these as 3D maps showing how tiny particles (quarks) move inside a proton (the "cloud").

To make these maps, they use a method called the CSS approach. This method is great at predicting how particles behave when they are moving slowly or when they are far apart. But when particles get very close together (high energy), the math gets messy and hits a "Landau pole"—a point where the numbers explode to infinity.

To stop the explosion, they use the $b^*$ prescription. This is a mathematical trick that says, "Okay, let's pretend the distance between particles stops getting smaller than a certain limit." It's a rule of thumb to keep the math from breaking.

The Experiment: Testing the Tape

The authors decided to stress-test this rule. They took two different "types of tape" (two different mathematical formulas for the $b^*$ prescription) and applied them to real-world data from particle collisions (Drell-Yan experiments).

Here is what they found, broken down into three simple scenarios:

1. The Low-Energy Test (The "Slow" Cloud)

When they looked at data where particles were moving relatively slowly (low energy), both types of tape worked perfectly.

• The Analogy: Imagine taking a photo of a slow-moving car. Whether you use clear tape or thick tape on your lens, the photo looks the same. The car is moving slowly enough that the "blur" doesn't matter.
• The Result: Both methods gave excellent agreement with the data and produced very similar maps of the particles' movement.

2. The Middle-Ground Test (The "Fast" Cloud)

Then, they looked at the "middle" speed—where particles are moving fast enough that the math is tricky, but not so fast that we have a perfect theory for them.

• The Analogy: Now imagine the car is speeding up. Suddenly, the type of tape matters! One type of tape makes the car look like it's drifting left, while the other makes it look like it's drifting right.
• The Result: The two different prescriptions produced significantly different maps of the particles in this middle region. One map looked like it followed the laws of physics perfectly; the other looked a bit "off."

3. The High-Energy Prediction (The "Super" Cloud)

Finally, they used their maps to predict what would happen in a much more powerful experiment (high-energy collisions at the CDF collider).

• The Analogy: They tried to predict the path of a race car going at 200 mph. The map made with the "good" tape predicted the car would stay on the track. The map made with the "bad" tape predicted the car would fly off the track.
• The Result: When they compared their predictions to the actual race results, only the "good" tape worked. The predictions based on the "bad" tape were completely wrong.

The Big Lesson

The paper concludes that the "tape" (the $b^*$ prescription) isn't just a harmless technical fix. It introduces a hidden bias.

• The Problem: If you only look at slow-moving data, you can't tell which tape is better. You might think your map is perfect, but it could be wrong in the middle region.
• The Solution: To get the right map, you must combine data from slow-moving particles AND fast-moving particles in the same analysis. This is called a "Global Fit."
  • The slow data tells you about the core of the proton.
  • The fast data tells you how the math should behave at high speeds.
  • By combining them, you force the "tape" to be the right kind, ensuring the whole map is accurate.

Why This Matters

This study is a warning to the scientific community. It says: "Don't just trust the math because it fits the data you have right now."

The choices scientists make to fix broken math (like the $b^*$ prescription) can secretly change the results. To understand the true structure of matter, we need to be careful about these choices and use as much data as possible—from the slowest to the fastest collisions—to ensure our "maps" of the universe are real, not just artifacts of our mathematical patches.

In short: The way we patch our math matters. If we want to see the universe clearly, we need to check our work against the whole picture, not just the easy parts.

Technical Summary

1. Problem Statement

The paper addresses a critical source of theoretical uncertainty in the extraction of Transverse Momentum Dependent (TMD) parton distribution functions (PDFs) within the Collins–Soper–Sterman (CSS) formalism.

• The Core Issue: The CSS formalism requires resummation of large logarithms in impact-parameter space ($b_T$) to handle the non-perturbative regime. However, perturbative QCD breaks down at large $b_T$ due to the Landau pole. To circumvent this, a $b_*$ prescription is introduced, which maps the physical $b_T$ to a modified variable $b_*$ that saturates at a maximum value $b_{\text{max}}$, keeping the calculation within the perturbative domain.
• The Ambiguity: The choice of the functional form of $b_*(b_T)$ and the value of $b_{\text{max}}$ are arbitrary phenomenological inputs. While the standard CSS approach assumes that non-perturbative functions ($g_K$ and $g_{q/h}$) can compensate for these choices to render physical predictions independent of the prescription, the authors argue that in practice, this compensation is incomplete.
• The Gap: It is unclear whether different $b_*$ prescriptions lead to physically equivalent TMDs, particularly in the intermediate transverse momentum region ($1 \text{--} 2 \text{ GeV} \ll |k_\perp| \ll Q$), where neither fixed-order perturbation theory nor pure non-perturbative modeling is sufficient.

2. Methodology

The authors perform a "phenomenological stress-test" to quantify the impact of these prescriptions.

• Framework: They utilize the NangaParbat fitting framework at Next-to-Next-to-Leading Logarithmic (NNLL) accuracy.
• Dataset:
  • Low-Energy: Drell-Yan (DY) data from Fermilab experiments E288 and E605 (covering $Q \approx 4\text{--}11$ GeV). These data primarily constrain the non-perturbative (low $|k_\perp|$) region.
  • High-Energy (Prediction): Z-boson production data from CDF Run I ($\sqrt{s} = 1.96$ TeV, $Q \approx 66\text{--}116$ GeV). This dataset probes the intermediate and large $|k_\perp|$ regions.
• Non-Perturbative Model: A simple Gaussian model is used for the intrinsic TMD part ($g_1$) and the Collins-Soper kernel ($g_2$), keeping the functional form fixed to isolate the effect of the $b_*$ variations.
• Variables Tested: Four distinct configurations were tested by varying:
  1. The functional form of $b_*$:
     • $b^{\text{CSS}}_*$ (Original square-root form from Ref. [15]).
     • $b^{\text{exp}}_*$ (Exponential form from Ref. [53], providing a sharper transition).
  2. The saturation scale $b_{\text{max}}$:
     • $b_{\text{max}} = c_1$ (Standard, $\approx 1.123$ GeV$^{-1}$).
     • $b_{\text{max}} = c_1/2$ (Modified, raising the non-perturbative threshold to $\approx 2$ GeV).

3. Key Contributions and Results

A. Low-Energy Data Agreement (The Illusion of Stability)

• Finding: All four $b_*$ configurations yield equally good fits to the low-energy E288 and E605 data ($\chi^2/N_{\text{dat}} \approx 1$).
• Implication: The extracted TMDs in the low transverse momentum region ($|k_\perp| \lesssim 1$ GeV) are compatible across all prescriptions. The non-perturbative parameters absorb the differences in the $b_*$ prescription, masking the underlying theoretical ambiguity when only low-energy data is considered.

B. Intermediate Region Divergence

• Finding: Significant discrepancies emerge in the intermediate transverse momentum region ($1 \text{--} 2$ GeV $< |k_\perp| < Q$).
  • Configurations with $b_{\text{max}} = c_1/2$ (modified) shift the "node" (where the TMD changes sign) to lower momenta and fail to reproduce the behavior predicted by analytic resummation in transverse-momentum space.
  • Configurations with $b_{\text{max}} = c_1$ (standard) align much better with the perturbative resummation predictions.
• Significance: The choice of prescription acts as an intrinsic theoretical uncertainty that is not visible in low-energy fits but fundamentally alters the shape of the TMD in the transition region.

C. High-Energy Predictive Power

• Finding: When the TMDs extracted from low-energy data are used to predict CDF Run I Z-boson data:
  • Standard Configurations ($b_{\text{max}}=c_1$): Provide excellent agreement with high-energy data ($\chi^2/N_{\text{dat}} \approx 0.6\text{--}0.8$).
  • Modified Configurations ($b_{\text{max}}=c_1/2$): Fail significantly ($\chi^2/N_{\text{dat}} \approx 2.9\text{--}6.2$), underestimating the cross-section at intermediate $q_T$.
• Conclusion: The "modified" prescriptions introduce artifacts in the intermediate region that are incompatible with the evolution of TMDs to higher energy scales.

D. The Role of Global Fits

• Finding: Including high-energy data (CDF) in the global fit improves the description of high-energy data for the "modified" configurations but deteriorates the fit quality for low-energy data (E605).
• Insight: This demonstrates a tension between energy regimes. A single non-perturbative model cannot simultaneously satisfy the constraints of low-energy data (sensitive to the core) and high-energy data (sensitive to the evolution and intermediate region) if the $b_*$ prescription is poorly chosen.
• Recommendation: Global fits combining low- and high-energy data are essential to constrain the $b_*$ prescription and reduce phenomenological bias.

E. The $b_{\text{min}}$ Prescription (UV Regularization)

• The paper also investigates the $b_{\text{min}}$ prescription, used to regulate the UV divergence of the integral relating TMDs to collinear PDFs.
• Result: While $b_{\text{min}}$ distorts the perturbative matching at large $|k_\perp|$, its numerical impact on low-$q_T$ phenomenology is negligible. The authors show that TMDs extracted without $b_{\text{min}}$ still satisfy the integral relation to collinear PDFs (within power corrections) if the integral is properly UV-regularized, suggesting $b_{\text{min}}$ is a phenomenological choice rather than a fundamental requirement.

4. Significance and Conclusions

1. Intrinsic Uncertainty: The $b_*$ prescription is not merely a technical tool but a source of systematic theoretical uncertainty. Different choices lead to different physical interpretations of the TMDs in the intermediate region, even if they fit low-energy data equally well.
2. Necessity of Global Fits: Relying solely on low-energy Drell-Yan data is insufficient for reliable TMD extraction. High-energy data (e.g., LHC, CDF) are crucial because they probe the intermediate momentum region where the choice of prescription manifests.
3. Validation of Standard Approach: The study validates the standard CSS choice ($b_{\text{max}} = c_1$) as being more consistent with analytic resummation and high-energy data than alternative "modified" choices.
4. Future Directions: The authors suggest that while global fits mitigate prescription dependence, the ultimate solution may lie in more flexible parametrizations (e.g., neural networks) or alternative frameworks that treat the prescription as an intrinsic part of the non-perturbative modeling rather than an external regulator.

In summary, the paper demonstrates that phenomenological prescriptions in TMD extractions are not benign; they introduce significant biases that can only be constrained by a global analysis spanning multiple energy scales.