⚛️ phenomenology

Automated higher-order predictions for ultra-peripheral collisions with full impact-parameter dependence

This paper introduces an automated framework within the Sherpa event generator for simulating photon-induced processes in proton and nuclear collisions with full impact-parameter dependence and higher-order electroweak corrections, which is validated against ATLAS data and used to provide new predictions for exclusive $W$ - and $\tau$ -pair production.

Original authors: Frank Krauss, Peter Meinzinger

Published 2026-07-22

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Frank Krauss, Peter Meinzinger

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Large Hadron Collider (LHC) not just as a giant particle-smashing machine, but as a cosmic billiard table where the balls are protons and heavy atomic nuclei. Usually, we watch what happens when these balls crash head-on, shattering into a shower of new particles. But sometimes, the balls don't actually hit each other. They whiz past so closely that they feel a massive electromagnetic tug, like two magnets snapping past one another without touching. In physics, this is called an "ultra-peripheral collision." Because the particles are moving so fast, they are surrounded by a cloud of "virtual" photons—packets of light that act like a beam of light from a flashlight. When these two beams of light cross paths, they can collide and create new particles, turning the LHC into a giant, accidental photon collider. Scientists care about this because it lets them study rare quantum effects and test the fundamental laws of nature in a way that head-on crashes simply can't. To do this, they need to know exactly how strong that "light beam" is and how the size of the particle affects it, which requires incredibly precise math.

This paper introduces a new, automated tool built inside a famous computer program called SHERPA that acts like a super-smart simulator for these photon collisions. The authors, Frank Krauss and Peter Meinzinger, have created a framework that can automatically calculate what happens when protons or heavy nuclei (like lead) pass each other, taking into account the fact that these particles aren't just tiny dots, but have a real, finite size. Think of it like trying to predict the splash pattern when two waves crash; if you pretend the waves are just single points, your math is easy but wrong. If you account for the fact that the waves have width and shape, the prediction becomes much more accurate. The team's new tool does exactly this: it calculates the "photon flux" (the intensity of the light beam) based on the actual shape of the proton or nucleus and how close they pass each other, a distance physicists call the "impact parameter."

The researchers tested their new simulator against real data from the ATLAS experiment at the LHC, looking at collisions where protons or lead nuclei created pairs of muons (heavy cousins of electrons). They found that if you pretend the particles are point-like dots and ignore the space between them, your predictions are off. However, by including the full size of the particles and allowing the photons to be emitted from very close to the center of the nucleus (even inside the "radius" where you might expect them not to be), their simulation matched the real-world data much better. They also added "higher-order" corrections, which are like accounting for the tiny, messy details of physics—such as the particles occasionally spitting out extra soft photons—that happen in the real world but are often ignored in simple models. These corrections lowered their predicted numbers by about 10% in some cases, bringing them closer to what the detectors actually saw.

While the simulation worked very well for proton collisions, there was still a small gap of about 10% between their predictions and the data for lead-lead collisions. The authors suggest this isn't because their math is wrong, but likely because of a subtle "veto" in the real experiment: sometimes the heavy nuclei break apart in a way that creates extra debris in the detector, causing the computer to discard the event as a "failed" collision. This effect, known as electromagnetic dissociation, is hard to model perfectly but seems to be the culprit for the remaining difference.

Beyond just checking their math, the team used their new tool to make fresh predictions for processes that are currently being studied or are just on the horizon. They simulated the creation of pairs of W bosons (heavy force-carrying particles) and pairs of tau leptons (even heavier cousins of electrons). For the W bosons, their results matched existing measurements within the large margins of error, but they showed that the "messy" higher-order corrections could change the shape of the data by up to 15% in certain areas. For the tau pairs, they found that the way the particles decay and spin relative to each other has a huge impact, shifting the results by up to 7%. The paper concludes that this new automated framework is ready to be used by other scientists to explore these rare photon-induced events with greater precision, helping to uncover the secrets of the universe hidden in the gaps between colliding particles.

Technical Summary: Automated Higher-Order Predictions for Ultra-Peripheral Collisions with Full Impact-Parameter Dependence

Problem Statement
Ultra-peripheral collisions (UPCs) of protons and nuclei at the Large Hadron Collider (LHC) offer a unique environment to study photon-induced processes, effectively turning the collider into a photon-photon machine. While specialized simulation tools exist (e.g., STARlight, SuperCHIC), they often rely on simplified treatments of form factors, impact-parameter dependence, or higher-order corrections. Furthermore, existing general-purpose event generators have lacked a fully automated framework that integrates the Equivalent Photon Approximation (EPA) with full impact-parameter dependence, diverse form factor parametrizations, and automated higher-order electroweak corrections (NLO EW) and soft-photon resummation (YFS). This paper addresses the need for a unified, automated framework within the SHERPA event generator to provide precise, full particle-level simulations for these processes, validated against experimental data.

Methodology
The authors developed a framework within SHERPA v3.1 that calculates photon-induced processes based on the EPA formalism. Key methodological components include:

Kinematics and Flux Calculation: The photon virtuality $Q^2$ and fractional energy $x$ are defined, and the differential photon flux is derived from the spin density matrix. The flux is expressed as a function of the impact parameter $b_\gamma$ via a Fourier transform involving Bessel functions.
Form Factor Parametrizations: The framework implements a comprehensive set of form factors to model the finite size of emitters:
- Protons: Budnev parametrization, dipole, approximate dipole, Gaussian, and point-like approximations.
- Heavy Ions (e.g., Lead): Woods-Saxon nuclear density distribution (computed numerically via Fourier transform), its analytic approximation, and point-like limits.
- Transition: For extended form factors, the calculation transitions to the point-like approximation at large impact parameters to optimize numerical integration while maintaining accuracy in the relevant kinematic region.
Impact-Parameter Sampling and Survival Factors: The impact parameter $b_\gamma$ $b_{γ}$ is sampled from a distribution optimized for efficiency. To ensure physical exclusivity (preventing strong interactions), the framework employs two survival probability models ( $P^2_{surv}$ $P_{s u r v}^{2}$ ):
- Hard-sphere rejection: For nuclei, rejecting events where the impact parameter $b < R_1 + R_2$ .
- Probabilistic survival: For proton-proton collisions, using a parametrization fitted to data to model the probability of avoiding inelastic collisions.
Higher-Order Corrections: The framework incorporates higher-order effects automatically via two distinct methods:
- NLO Electroweak (EW): Calculated using local subtraction with Catani-Seymour dipoles.
- Soft-Photon Resummation: Implemented via the Yennie-Frautschi-Suura (YFS) formalism to all orders.
Validation and Prediction: The framework was validated against ATLAS measurements of exclusive dimuon ( $\mu^+\mu^-$ ) production in $pp $($ \sqrt{s}=13$ TeV) and Pb+Pb ( $\sqrt{s}_{NN}=5.02$ TeV) collisions. It was subsequently used to predict exclusive $W^+W^-$ production in $pp$ and $\tau^+\tau^-$ production in both $pp$ and Pb+Pb collisions.

Key Results

Form Factor Sensitivity: Comparisons of different form factors for dimuon production show that while they agree within a few percent for ion beams, the choice of form factor and the treatment of the impact parameter cutoff are critical. Specifically, restricting the minimum emission distance $b_{\gamma, min}$ to the nuclear radius ( $R$ ) significantly underestimates the cross-section. The authors find that contributions from $b_{\gamma} < R$ are physically relevant, and a default cutoff of $b_{\gamma, min}/R = 0.3$ is required for convergence.
Survival Probability Modeling: For ion beams, the inclusion of overlap rejection (hard-sphere) reduces the cross-section by approximately 10–30% depending on the phase space cuts. For protons, the difference between hard-sphere and probabilistic rejection models is small (~3%).
Higher-Order Corrections: Including NLO EW corrections and YFS resummation reduces the leading-order (LO) cross-section predictions by up to 10% in the lowest mass bins. This reduction is attributed to the use of "bare" leptons in experimental analyses, which are sensitive to soft and collinear final-state radiation. The YFS resummation generally yields a slightly stronger reduction than NLO EW at low masses.
Data Validation:
- Dimuon Production: The framework describes ATLAS $pp$ data well. For Pb+Pb data, the most realistic calculation (including full impact-parameter dependence, survival factors, and higher-order corrections) still overpredicts the data by approximately 10%. The authors attribute this residual discrepancy to electromagnetic dissociation of nuclei producing central hadronic activity, which violates the experimental exclusivity veto, a finding consistent with recent literature.
- W-Pair Production: Predictions for exclusive $W^+W^-$ production in $pp$ collisions are compatible with ATLAS measurements within large experimental uncertainties. Differential corrections (NLO and YFS) reach up to -15% in kinematic tails (large transverse momentum).
- Tau-Pair Production: Predictions for $\tau^+\tau^-$ production in $pp$ and Pb+Pb collisions highlight that spin correlations and QED radiation at decay vertices dominate the corrections (shifting cross-sections by +5% for leptonic and -7% for hadronic channels), while genuine NLO EW corrections to the production process remain at the percent level.

Significance and Claims
The paper claims to present the first automated framework within a general-purpose event generator (SHERPA) that combines full impact-parameter dependence, a wide range of form factor parametrizations, and automated higher-order electroweak corrections for UPCs. The authors emphasize that their approach allows for arbitrary final states, including Beyond Standard Model (BSM) scenarios via the UFO interface.

The study demonstrates that precise modeling of the impact parameter and the inclusion of higher-order radiative corrections are essential for matching experimental data. While the framework successfully describes $pp$ data, the persistent ~10% discrepancy in Pb+Pb collisions underscores the complexity of modeling nuclear survival and dissociation effects. The authors conclude that their tool provides a robust foundation for future studies of photon-induced processes, including jet production and applications to future electron-ion colliders (EIC), where similar formalisms apply. The framework is intended for release as part of SHERPA v3.1.

More like this