Optimization of factorization scale in QED Drell-Yan-like processes

This paper investigates the dependence of initial-state radiation corrections in $e^+e^-$-annihilation processes on the factorization scale by analyzing various prescriptions within leading and next-to-leading logarithmic approximations and optimizing the scale choice through comparison with complete two-loop results.

The Gist

Imagine you are trying to predict exactly how much energy is released when two particles, an electron and a positron, smash into each other and turn into new particles. In the world of high-energy physics, this is like trying to calculate the exact outcome of a complex billiard shot, but the balls are made of pure energy and the table is governed by quantum rules.

To get a precise answer, physicists use a mathematical tool called perturbation theory. Think of this as building a tower. You start with a solid base (the simplest calculation), then you add a second floor (a small correction), then a third floor (a smaller correction), and so on. The more floors you add, the more accurate your prediction becomes.

However, there's a catch. To build these floors, you have to choose a "reference height" or a factorization scale. This is like deciding where to set your ruler before you start measuring. If you set the ruler too low or too high, the measurements for the different floors of your tower get mixed up. Some parts of the calculation that should be small might look huge, and vice versa. This makes the tower wobbly and hard to predict.

The Problem: Where to Set the Ruler?

In this paper, the authors (Arbuzov, Voznaya, and Sadouski) investigate a specific type of particle collision (electron-positron annihilation) and ask: "What is the best place to set our ruler so that our calculations are as stable and accurate as possible?"

They look at three main ways people usually choose this scale:

1. The "Standard" Way: Set the ruler to the total energy of the collision.
2. The "Fastest Convergence" Way: Set the ruler where the math seems to settle down the quickest.
3. The "Minimal Sensitivity" Way: Set the ruler where a tiny change in the setting doesn't change the result much.

The Experiment: Testing the Scales

The authors have a unique advantage. For this specific particle collision, they already know the "perfect" answer for the first few floors of the tower (up to two loops of calculation). This is like having the blueprint of the finished building. They can now test their different ruler settings to see which one gets them closest to the blueprint without needing to build the entire, incredibly difficult third or fourth floor.

They tested three specific ruler settings:

• Setting A: The full collision energy ($\sqrt{s}$).
• Setting B: The full energy divided by a mathematical constant ($\sqrt{s/e}$).
• Setting C: The energy of the final particles produced ($\sqrt{sz}$).

The Findings: What Worked Best?

Here is what they discovered, using simple analogies:

• The "Standard" Way (Setting C): This is the most common method used by physicists. It works well when you are looking at the "middle" floors of the tower (Next-to-Leading Logarithmic order). However, for the very first, most basic floors (Leading Logarithmic order), it causes the math to wobble significantly. It's like using a ruler that is perfect for measuring a book but terrible for measuring a wall.
• The "Fastest Convergence" Way (Setting B): This turned out to be the winner for many situations. By setting the ruler to the collision energy divided by a specific number ($\sqrt{s/e}$), the "wobbly" parts of the calculation (the messy corrections) were absorbed neatly into the main structure. It made the tower stand straighter with fewer floors needed to get a good prediction.
• The "Minimal Sensitivity" Way: This also suggested using a high energy setting, similar to Setting A or B, which is a reasonable choice, though not always the absolute perfect one for every single scenario.

A Warning About "Safety Margins"

Physicists often estimate how wrong their calculations might be by moving the ruler slightly up and down (doubling or halving the scale) and seeing how much the result changes. If the result doesn't change much, they think, "Great, our answer is safe."

The authors found a trap here. When the particles "radiate" energy and drop down to a lower energy state (a phenomenon called "radiative return"), the standard method of moving the ruler up and down greatly underestimates the uncertainty. It's like checking if a bridge is safe by shaking it gently, but failing to notice that a specific type of wind (radiative return) could actually make it collapse. In these specific cases, the "safety margin" calculation gives a false sense of security.

The Conclusion

The paper concludes that for electron-positron collisions, the best way to set the mathematical ruler is often to use a value related to the total collision energy (specifically $\sqrt{s}$ or $\sqrt{s/e}$), rather than just the energy of the final particles.

This helps physicists build more stable "towers" of calculation, meaning they can predict experimental results with higher confidence. Since the math for electron collisions is a simpler version of the math used for proton collisions (like those at the Large Hadron Collider), these insights might also help improve predictions for those more complex machines.

In short: The authors found a better way to set the "ruler" for particle physics calculations, making the math more stable and revealing that the usual way of checking for errors can sometimes be dangerously optimistic.

Technical Summary

1. Problem Statement

In high-energy physics, precise theoretical predictions for observables require the calculation of higher-order radiative corrections within perturbation theory. For Quantum Electrodynamics (QED) processes like electron-positron annihilation ($e^+e^- \to \gamma^*/Z \to \mu^+\mu^-$), these corrections involve large logarithms of the form $L = \ln(\mu_F^2/\mu_R^2)$, where $\mu_F$ is the factorization scale and $\mu_R$ is the renormalization scale.

While the renormalization scale is typically fixed to the electron mass ($m_e$) to control mass singularities, the choice of the factorization scale $\mu_F$ is arbitrary at any finite order of perturbation theory. An arbitrary choice can lead to:

• Poor convergence of the perturbative series.
• Large contributions from higher-order terms (e.g., $O(\alpha^n L^0)$) that are difficult to calculate.
• Significant theoretical uncertainty if the scale dependence is not properly managed.

The paper addresses the challenge of optimizing the factorization scale to concentrate corrections into Leading Logarithmic (LL) and Next-to-Leading Logarithmic (NLL) terms, thereby improving convergence and reducing the impact of unknown higher-order terms.

2. Methodology

The authors analyze the $e^+e^-$ annihilation process, treating it analogously to the QCD Drell-Yan process using the QED structure function approach. This allows for a systematic inclusion of higher-order contributions amplified by large logarithms.

Key Methodological Steps:

• Analytical Comparison: The study leverages the unique availability of complete analytical results up to $O(\alpha^2)$ (two-loop) for this process. This allows for a direct comparison between approximate calculations (LL, NLL, NNLL) and exact results.
• Scale Variation: The authors test several prescriptions for choosing $\mu_F$:
  1. Conventional Scale Setting (CSS): $\mu_F = \sqrt{s z}$ (invariant mass of the final state pair), where $s$ is the center-of-mass energy and $z$ is the energy fraction.
  2. Fixed Energy Scales: $\mu_F = \sqrt{s}$ and $\mu_F = \sqrt{s/e}$.
  3. Optimization Principles: Theoretical frameworks like Fastest Apparent Convergence (FAC), Principle of Minimal Sensitivity (PMS), Brodsky-Lepage-Mackenzie (BLM), and Principle of Maximal Conformality (PMC) are discussed, though the authors note BLM/PMC are less applicable here due to the dominance of anomalous dimensions over running coupling effects.
• Quantitative Analysis:
  • Differential Distributions: The cross-section $d\sigma/ds'$ is analyzed across different bins of $z$ (energy fraction) to see how scale choices affect the redistribution of LL, NLL, and NNLL terms.
  • Total Cross-Section: Integrals are performed with varying lower cutoffs ($z_{min}$) to include or exclude the "radiative return" to the $Z$ resonance.
  • Uncertainty Estimation: The standard procedure of varying $\mu_F$ by a factor of 2 ($\mu_F \to \mu_F/2, 2\mu_F$) is tested to see if it accurately estimates the magnitude of unknown higher-order corrections.

3. Key Contributions

• Systematic Scale Optimization: The paper provides a quantitative analysis of how different $\mu_F$ choices redistribute contributions between logarithmic orders ($L^k$) in the perturbative expansion.
• Identification of Optimal Scales: The authors demonstrate that for $e^+e^-$ annihilation, scales proportional to the initial center-of-mass energy ($\mu_F \approx \sqrt{s}$ or $\mu_F \approx \sqrt{s/e}$) are superior to the conventional $\mu_F = \sqrt{sz}$ in the Leading Logarithmic approximation.
• Validation of Uncertainty Estimates: The study critically evaluates the standard "factor of 2" variation method, revealing its limitations in specific kinematic regimes (specifically near resonances).

4. Key Results

A. Differential Distributions

• LL and NLL Absorption: When $\mu_F = \sqrt{s/e}$, the NLL contributions in the $O(\alpha)$ order are almost completely absorbed into the LL terms. Similarly, in the $O(\alpha^2)$ order, the bulk of the $O(\alpha^2 L^1)$ effect is absorbed into the $O(\alpha^2 L^2)$ term.
• Conventional Scale Failure: The conventional choice $\mu_F = \sqrt{sz}$ (characteristic of the hard subprocess) performs poorly in the LL approximation, failing to absorb lower-order logarithmic terms effectively. However, it performs reasonably well in the NLL approximation for specific beam energies.
• Resummation: The choice $\mu_F = \sqrt{s/e}$ aligns with the Fastest Apparent Convergence (FAC) method, suggesting it minimizes the size of higher-order corrections.

B. Total Cross-Section and Radiative Return

• Radiative Return Impact: When integrating down to low $z$ (e.g., $z_{min}=0.1$), the "radiative return" to the $Z$ resonance ($sz \approx M_Z^2$) generates large corrections.
• Scale Sensitivity: The optimal intersection point of LL and NLL curves shifts depending on the order of calculation and experimental cuts. While $\mu_F = \sqrt{s/e}$ is generally preferred, the optimal scale is not universal and depends on the specific energy and kinematic cuts.

C. Uncertainty Estimation (Scale Variation)

• Overestimation at Resonance: At the $Z$-peak ($\sqrt{s} = M_Z$), varying the scale by a factor of 2 provides a reasonable estimate of uncertainties.
• Underestimation off-Peak: Crucially, for energies above the $Z$-peak (e.g., $\sqrt{s} = 240$ GeV or 3 TeV) with low $z_{min}$ (including radiative return), the standard scale variation method seriously underestimates the true uncertainty. The estimated shift ($\delta$) is significantly smaller than the actual missing higher-order contribution ($\Delta$). This indicates that standard uncertainty estimates may be insufficient for precision physics in these regimes.

5. Significance and Implications

• Precision Physics: The findings are critical for current and future high-luminosity $e^+e^-$ colliders (like FCC-ee or CLIC) where precision measurements of the $Z$ boson and Higgs production require sub-percent theoretical accuracy.
• QCD Relevance: Since the QED Drell-Yan process is an Abelian reduction of the QCD Drell-Yan process, the insights regarding scale dependence, the failure of standard uncertainty estimates near resonances, and the benefits of specific scale choices (like $\sqrt{s/e}$) offer valuable guidance for optimizing calculations in QCD at the LHC.
• Methodological Shift: The paper suggests moving away from the rigid application of $\mu_F = \sqrt{sz}$ for initial-state radiation in annihilation processes and adopting scales that better align with the initial state energy to maximize perturbative convergence.

In conclusion, the authors demonstrate that optimizing the factorization scale is not merely a technicality but a necessity for controlling theoretical uncertainties, particularly in regimes involving radiative returns to resonances where standard error estimation techniques fail.