⚛️ phenomenology

Color-Reconnection Effects on the nMPI-Nch Conditional Multiplicity Fluctuations in PYTHIA 8 pp Collisions

This paper investigates conditional charged-multiplicity fluctuations in PYTHIA 8 proton-proton collisions, finding that while color-reconnection models consistently reduce absolute fluctuation widths while increasing relative variations across various tunes and energies, historical uncertainties in baseline settings prevent definitive attribution of these effects solely to color reconnection.

Original authors: Subir Sarkar, Kalyan Dey

Published 2026-10-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Subir Sarkar, Kalyan Dey

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

Inside the heart of a particle accelerator, protons do not simply bounce off one another like billiard balls. When two protons collide at nearly the speed of light, they are more like two bags of tangled string being smashed together. Inside each proton, smaller particles called quarks and gluons swarm in a chaotic cloud. When the protons meet, these internal particles can collide with each other multiple times in a single crash, a phenomenon physicists call multiparton interactions. The result is a shower of new particles flying out in all directions. To understand how the universe works at its most fundamental level, scientists must understand exactly how these internal collisions translate into the final spray of debris we see in our detectors. A key piece of this puzzle is a process called color reconnection. Before the particles settle down and become the stable matter we know, the invisible forces holding them together rearrange themselves, linking different parts of the collision in new ways. This reshuffling changes how many particles are produced and how fast they move, but pinning down exactly how much this process changes the outcome has been a difficult challenge.

A team of researchers recently tackled this question by running massive simulations of proton collisions at three different energy levels: 2.76, 7, and 13 trillion electron volts. They used a sophisticated computer program known as PYTHIA 8, which acts as a virtual laboratory, allowing them to watch the entire life cycle of a collision from the initial smash to the final spray of particles. The scientists wanted to isolate the effect of color reconnection. To do this, they ran two sets of simulations for every energy level. In the first set, the color reconnection process was turned on, allowing the invisible strings to rearrange. In the second set, it was turned off, keeping the connections fixed as they were created. By comparing these two scenarios, they could see what changed when the strings were allowed to reconnect.

The researchers focused on a specific question: if you know exactly how many internal collisions happened in a crash, does turning on color reconnection change the number of particles that fly out? They grouped their simulated events by the exact count of these internal collisions. For every group where the number of internal collisions was the same, they counted the resulting charged particles. They found a consistent pattern across all three energy levels in the retained data. When color reconnection was active, the average number of particles produced was lower than when it was inactive. However, the spread of the data told a more subtle story. While the total number of particles dropped, the variation in that number did not drop by the same amount.

Imagine a classroom where the average test score drops, but the difference between the highest and lowest scores stays roughly the same. In that case, the class becomes more unpredictable relative to its new average. This is exactly what the researchers observed. When color reconnection was enabled, the average number of particles decreased significantly, but the width of the distribution—the range of possible outcomes—stayed relatively wide. This meant that the relative fluctuation, or how much the particle count varied compared to the average, actually increased. In simpler terms, the process made the final outcome more variable relative to the mean, even though the total number of particles was smaller. This pattern held true in the historical data across all three energy levels, though the researchers noted that for these older samples, some original computer settings were not fully preserved, meaning they could not definitively prove that color reconnection was the sole cause of the observed differences or their energy dependence.

To ensure this result was not just a quirk of one specific computer setup, the team tested it using different configurations of the simulation software, including settings used by major experiments at the Large Hadron Collider. They also compared their results with real data collected by the ALICE experiment, which measures the actual speed of particles in these collisions. The simulations with color reconnection turned on matched the real-world data trends more closely than the simulations without it. This suggests that the rearrangement of invisible forces is a real and necessary part of how nature handles these high-energy crashes.

However, the researchers were careful to note the limits of their findings. Much of their data came from historical simulations where some of the original computer settings were not fully preserved. While the pattern was consistent, they could not definitively prove that color reconnection was the only difference between the old simulations without re-running them from scratch with perfect records. The most reliable evidence came from a newer, fully documented set of simulations at the highest energy level, which confirmed the same behavior. Additionally, while they used other configurations like Monash and A14 to support their findings, the records for these samples lacked a fully audited historical production chain, preventing them from being treated as fully controlled evidence for the generality of the effect across all tunes. The study concludes that while color reconnection reduces the total number of particles produced in a collision with a fixed number of internal interactions, it simultaneously makes the final count more unpredictable relative to that average. This insight helps physicists refine their models of the subatomic world, ensuring that their virtual laboratories accurately reflect the complex, chaotic reality of the universe's smallest building blocks.

Technical Summary: Color-Reconnection Effects on the nMPI–Nch Conditional Multiplicity Fluctuations in PYTHIA 8 pp Collisions

Problem Statement
The paper investigates the relationship between multiparton interactions (MPI) and final-state charged-particle multiplicity ( $N_{ch}$ ) in proton–proton ($pp$) collisions, specifically focusing on how Color Reconnection (CR) modifies conditional fluctuations. While MPI models describe underlying-event activity, the final multiplicity is shaped by the full event evolution, including showers, hadronization, and CR. CR rearranges color connections between partons before string fragmentation, potentially reducing the string-length measure and altering particle production without changing the number of generated interactions.

A key challenge addressed is distinguishing whether observed differences in multiplicity distributions between CR-ON and CR-OFF states are solely due to CR or confounded by unverified historical production settings (e.g., effective lifetimes, decay overrides, and seed values). Previous studies have examined multiplicity at fixed MPI counts, but this work aims to rigorously compare conditional means, absolute widths, and relative widths (coefficients of variation) across multiple energies and tune configurations, while explicitly acknowledging limitations in the provenance of historical baseline samples.

Methodology
The study utilizes PYTHIA 8.316 to simulate $pp$ collisions at center-of-mass energies of 2.76, 7, and 13 TeV.

Samples: The baseline consists of 5 million events per energy and CR state using the Monash 2013 tune. An independent 13 TeV comparison uses the CMS CP5, Monash, and ATLAS A14 tunes.
Definitions:
- MPI Count ( $n_{MPI}$ ): The internal generator interaction count.
- Multiplicity ( $N_{ch}$ ): Charged particles with $|\eta| < 0.8$ and $0.15 < p_T < 20$ GeV/c.
- Selection: Events are conditioned on integer $n_{MPI} = m$ (with $m > 2$ ). Bins with fewer than 500 events in either CR state are discarded.
Metrics: For each matched bin $m$ $m$ , the authors calculate:
- Conditional mean ( $\mu_m$ ).
- Absolute standard deviation ( $\sigma_m$ ).
- Coefficient of variation ( $C_m = \sigma_m / \mu_m$ ).
- Scaled variance ( $F_m = \sigma_m^2 / \mu_m$ ).
Validation: The analysis employs AI-assisted code development (OpenAI Codex) for parsing and plotting. Validation includes reanalysis of 60 million source event records, empirical bootstrap checks (2,000 replicates), and verification of inclusive distributions.
Provenance Caveats: The authors explicitly note that historical baseline samples lack certified records for effective lifetimes and decay settings, preventing a fully controlled causal attribution for those specific datasets. The CMS CP5 comparison, however, uses documented, identical non-CR settings and particle data dumps, providing a more robust test case.

Key Results

Conditional Width Ordering: Across all retained matched bins at 2.76, 7, and 13 TeV, enabling CR (CR-ON) results in:
- A smaller conditional mean ( $\mu_m$ ) compared to CR-OFF.
- A smaller absolute width ( $\sigma_m$ ) compared to CR-OFF.
- Crucially, the fractional decrease in the mean is larger than the fractional decrease in the absolute width. Consequently, the coefficient of variation ( $C_m$ ) is larger for CR-ON than for CR-OFF in every matched bin.
- This ordering ( $R_{\mu,m} < R_{\sigma,m} < 1$ ) implies that while the absolute spread of multiplicity narrows with CR, the relative fluctuations increase because the mean drops more significantly.
Energy and Tune Dependence:
- The pattern of smaller absolute width but larger relative width persists across all three energies and in independent tune configurations (Monash, A14, and CP5) at 13 TeV.
- The scaled variance ( $F_m$ ) shows different behavior; it is larger for CR-ON in Monash and A14 but nearly equal (or slightly smaller) for CP5, indicating that the ordering of $F_m$ is not universal across tunes.
Final-State Estimators:
- The study correlates $n_{MPI}$ with five final-state activity estimators (midrapidity multiplicity, forward multiplicity, transverse multiplicity, transverse scalar- $p_T$ sum, and forward-backward asymmetry).
- CR-OFF samples generally exhibit steeper conditional responses and stronger correlations with $n_{MPI}$ than CR-ON samples.
- CR-ON samples follow the trend of ALICE experimental data for mean transverse momentum ( $\langle p_T \rangle$ ) versus $N_{ch}$ more closely than CR-OFF, though this comparison is limited by differences in primary-particle definitions.
Normalization Sensitivity: The paper demonstrates that the ordering of relative fluctuations depends on the normalization scale. While CR-ON shows a larger coefficient of variation (normalized to the local conditional mean), it also shows a larger width when normalized to the inclusive mean of the respective CR state.

Significance and Claims
The paper claims to establish a consistent descriptive pattern in retained PYTHIA samples: enabling Color Reconnection reduces the conditional mean multiplicity by a larger fraction than it reduces the absolute standard deviation, leading to an increase in relative fluctuations (coefficient of variation).

However, the authors are modest regarding the causal implications:

Historical Baselines: Due to missing production records (effective lifetimes/decays) in the historical 2.76, 7, and 13 TeV Monash samples, the observed energy dependence cannot be definitively attributed to CR alone. These results are presented as descriptive comparisons of retained samples.
Controlled Evidence: The CMS CP5 13 TeV comparison, with verified identical non-CR settings, provides the strongest evidence for the CR-dependent ordering.
Generalizability: The authors state that the generality of these findings across different tunes and energies requires new, fully controlled baseline productions with complete records. They do not claim the results are universal across all event generators (e.g., Herwig) or that they resolve all model uncertainties.

In conclusion, the work provides a detailed characterization of how CR affects conditional multiplicity fluctuations in PYTHIA 8, highlighting a counter-intuitive result where absolute narrowing coexists with relative broadening, while carefully delineating the limits of causal attribution due to historical data constraints.

Technical Summary: Color-Reconnection Effects on the nMPI–Nch Conditional Multiplicity Fluctuations in PYTHIA 8 pp Collisions

More like this