Effectiveness of nonflow suppression using multi-particle correlators

This paper challenges the conventional view that multi-particle correlators effectively suppress non-flow effects by demonstrating, through toy models of particle decay and momentum conservation, that these estimators can actually produce flow harmonic estimates that deviate significantly from true values, particularly in small systems.

The Gist

The Great Flow Detective: When "Smart" Tools Get Tricked

Imagine you are at a massive, chaotic party (a particle collision in a physics lab). Your goal is to figure out the "dance rhythm" of the crowd. In physics, this rhythm is called collective flow. It's the way thousands of particles move together in a coordinated, swirling pattern, much like a fluid.

However, the party isn't perfect. There are "non-flow" distractions:

1. The Clingy Couples (Particle Decay): Sometimes, two particles are born from the same "parent" and stick together, moving in a specific direction that has nothing to do with the main dance.
2. The Group Hug (Momentum Conservation): Physics has a rule that the total movement of the group must balance out. If one person jumps left, someone else must jump right. This creates a hidden connection between strangers that isn't part of the dance.

For years, physicists have used a tool called multi-particle correlators to find the real dance rhythm. The logic was simple: "If we look at groups of 4 or 6 people at a time instead of just pairs, the random 'clingy couples' will get lost in the noise, and we'll see the true dance." It was believed that higher-order tools (looking at more people) were the ultimate "non-flow suppressors."

This paper says: "Not so fast."

The authors ran a simulation using two "toy models" (simplified versions of reality) to test if these smart tools actually work as well as we thought. They found that in small, crowded systems, these tools can actually get more confused than simpler methods.

Here is the breakdown of their findings using everyday analogies:

1. The "Clingy Couples" Test (Toy Model I)

Imagine you are trying to measure the average dance speed of a crowd.

• The Setup: You have a crowd dancing to a beat (the input flow). Then, you secretly add 50 pairs of "clingers" who are holding hands and spinning in a specific, weird direction.
• The Expectation: The "multi-particle" tool (looking at groups of 4 or 6) should ignore the clingers and tell you the true dance speed.
• The Reality: The tool didn't ignore them. Instead, it started measuring the distorted dance speed caused by the clingers.
  • The Twist: The tool's result depended heavily on how the clingers were holding hands. If they held hands at a 90-degree angle, the tool's reading would drop. If they held hands at 180 degrees, it would rise.
  • The Analogy: It's like trying to measure the speed of a river by looking at a group of ducks. If you add a few ducks tied together with a rope, and you look at groups of 4 ducks, your calculation of the river's speed gets messed up by the rope. The "smart" tool didn't filter out the rope; it just calculated the speed of the river including the rope's weird pull.
  • The Surprise: In some cases, a "dumber" tool (like the Event-Plane method, which just looks at the general direction) actually gave a result closer to the original true dance speed, while the "smart" multi-particle tool gave a result that was further away.

2. The "Group Hug" Test (Toy Model II)

Now, imagine a room full of people who are not dancing at all (no background flow). They are just standing there.

• The Rule: Physics says, "If you move, someone else must move the opposite way to balance it out."
• The Expectation: If there is no dance, the flow measurement should be zero.
• The Reality: The multi-particle correlator tool said, "Hey, there is a tiny bit of flow!"
  • The Analogy: It's like a room of people standing still. Because they are all holding hands in a giant circle (momentum conservation), if you ask a group of 4 people, "Are you moving together?" the math says "Yes, slightly," even though they are just standing still.
  • The Problem: The tool invented a "ghost dance" that didn't exist. Other methods (like the Event-Plane method) correctly said, "No, there is no dance here." The multi-particle tool was tricked by the rules of the game itself.

The Big Takeaway

The paper concludes that higher-order multi-particle correlators are not a magic bullet.

• They don't always find the truth: In small systems (like small particle collisions), these tools often get closer to the "apparent" flow (the distorted reality with the noise included) rather than the "input" flow (the true signal we want to find).
• They are sensitive to the noise: Instead of ignoring the "clingy couples" or the "group hugs," these tools sometimes amplify the weird patterns those distractions create.
• Context matters: If you want to know the true underlying rhythm of the universe, simply counting groups of 4 or 6 particles might actually lead you further astray than simpler methods, especially when the crowd is small.

In short: Just because a tool is more complex and looks at more data points doesn't mean it's better at filtering out the noise. Sometimes, the noise is so clever that it tricks the smartest tools into seeing a dance that isn't there, or measuring a dance that has been warped by the noise.

Technical Summary

1. Problem Statement

In relativistic heavy-ion collisions, the extraction of collective flow harmonics ($v_n$) is complicated by non-flow contributions. Non-flow refers to correlations arising from sources other than the collective expansion of the Quark-Gluon Plasma (QGP), such as resonance decays, jet fragmentation, and global momentum conservation.

• Current Consensus: Multi-particle correlators (specifically higher-order cumulants like $v_n\{4\}$, $v_n\{6\}$) are widely regarded as the gold standard for suppressing non-flow. The heuristic argument is that non-flow is typically short-range (few-particle), so its contribution diminishes rapidly as the order of the correlator increases, while the collective flow (a many-body phenomenon) remains coherent.
• The Gap: It remains unclear whether these higher-order estimators effectively reconstruct the underlying input flow harmonics (the true background flow) or if they merely converge to a distorted "apparent" flow caused by the non-flow mechanisms themselves. This is particularly critical in small collision systems (e.g., p+p, p+A) where non-flow can dominate.

2. Methodology

The authors employ a controlled numerical and analytical approach using two specific toy models designed to mimic distinct classes of non-flow effects. They compare the performance of multi-particle correlators against two alternative methods: the Event-Plane (EP) method and the Maximum Likelihood Estimator (MLE).

• Toy Model I (Particle Decay):

  • Simulates non-flow by adding $M$ pairs of particles with a fixed opening angle ($\phi_{open}$) on top of a background of $N$ particles generated from a standard flow distribution ($v_2=0.1, v_3=0.06$).
  • Two scenarios are tested: pairs correlated with the symmetry plane and pairs uncorrelated with it.
  • The study varies the number of pairs ($M$) and the opening angle ($\phi_{open}$).

• Toy Model II (Global Momentum Conservation):

  • Simulates a system with zero initial collective flow but strict global momentum conservation constraints.
  • Events are generated using the T-generation algorithm (via ROOT's TGenPhaseSpace), which enforces $\sum \vec{p}_T = 0$ on an otherwise isotropic distribution.
  • This isolates the artificial flow signals generated purely by kinematic constraints.

• Analytical Framework:

  • The authors derive analytic expressions for the apparent flow harmonics ($\tilde{v}_n$) and the cumulants ($c_n\{2k\}$) for both models.
  • They analyze the combinatorial contributions of background particles versus non-flow pairs to the two- and four-particle correlation functions.

3. Key Contributions

1. Challenge to the "Suppression" Paradigm: The paper demonstrates that higher-order multi-particle correlators do not always recover the input flow. Instead, they often converge to the apparent flow of the distorted final state, which can deviate significantly from the true underlying physics.
2. Analytic Explanation of Oscillations: The authors provide a rigorous analytic derivation explaining the characteristic oscillations observed in $v_2\{2\}$ and $v_2\{4\}$ as a function of the particle pair opening angle. They show these oscillations arise from specific combinatorial terms involving $\cos(n\phi_{open})$.
3. Discovery of Artificial Flow in Momentum Conservation: In Toy Model II (zero input flow), the authors show that multi-particle cumulants yield non-zero flow harmonics (e.g., $v_1\{2\} \neq 0$) purely due to momentum conservation constraints. This artificial signal is absent in single-particle estimators (like EP and MLE), revealing a unique sensitivity of cumulants to this specific non-flow mechanism.
4. Comparison with MLE: The study validates the Maximum Likelihood Estimator (MLE) as a robust alternative that often tracks the input flow more accurately than correlators in the presence of non-flow, particularly in small systems.

4. Key Results

A. Toy Model I (Particle Decay)

• Convergence to Apparent Flow: Higher-order correlators ($v_n\{4\}, v_n\{6\}$) track the apparent flow (the flow of the final distorted distribution) more closely than lower-order estimators. However, the apparent flow itself is often far from the input flow.
• Underestimation vs. Overestimation:
  • In the uncorrelated pair scenario, all estimators underestimate the input flow, but correlators underestimate it more severely than the EP or MLE methods.
  • In the correlated scenario, $v_n\{2\}$ and $v_n\{4\}$ exhibit oscillations dependent on $\phi_{open}$. While $v_n\{4\}$ suppresses the non-flow contribution relative to $v_n\{2\}$, it still fails to recover the true input $v_n$ when the non-flow significantly alters the global distribution.
• Oscillation Period: The deviation from the input value oscillates with a period $T = 2\pi/n$, a feature unique to particle correlators and derived analytically.

B. Toy Model II (Momentum Conservation)

• Artificial Non-Zero Flow: Even with an isotropic input distribution ($v_n = 0$), multi-particle cumulants yield non-zero values (e.g., $c_1\{2\} < 0$, $c_2\{2\} > 0$).
• Scaling Laws: The magnitude of these artificial cumulants scales as $c_n\{2k\} \propto 1/(M-2k)^{nk}$, where $M$ is the multiplicity.
• Method Discrepancy:
  • Correlators: Detect a "fake" flow signal.
  • Event-Plane/MLE: These methods, which rely on single-particle distributions or event-by-event plane reconstruction, correctly identify the flow as zero (or negligible) because momentum conservation does not bias the single-particle azimuthal distribution, only the multi-particle correlations.
• Sign Reversal: The signs of the cumulants derived from the EP method (assuming pure flow) differ from those calculated directly from the momentum-conserving distribution, highlighting a fundamental mismatch in how the two approaches interpret the data.

5. Significance and Conclusion

• Re-evaluation of Small System Physics: The findings suggest that in small collision systems (where non-flow is large), the "robustness" of higher-order cumulants is conditional. They effectively suppress short-range non-flow but may introduce or amplify biases related to global constraints (like momentum conservation) or specific decay topologies.
• Interpretation Warning: A small value of $v_n\{4\}$ (or a sign change between $v_n\{2\}$ and $v_n\{4\}$) should not be automatically interpreted as a pure signature of collective flow or its suppression. It may instead reflect the specific structure of non-flow correlations.
• Alternative Approaches: The Maximum Likelihood Estimator (MLE) and Event-Plane methods are shown to be viable, and in some cases superior, alternatives for extracting the true underlying flow harmonics when non-flow mechanisms distort the particle distribution.
• Future Direction: The authors conclude that suppressing explicit few-particle non-flow in a correlator does not guarantee a faithful reconstruction of the input background flow. Future analyses must account for the specific nature of non-flow sources (decay vs. momentum conservation) rather than relying solely on the order of the cumulant.

In summary, the paper provides a critical theoretical and numerical correction to the standard interpretation of multi-particle flow measurements, demonstrating that higher-order correlators are not a universal "cure" for non-flow and can sometimes yield estimates further from the truth than conventional methods.