[Paper Review] Measurement of CollinearDrop jet mass and its correlation with SoftDrop groomed jet substructure observables in $\sqrt{s}=200$ GeV $pp$ collisions by STAR

Y. Song|arXiv (Cornell University)|Jul 15, 2023

Particle physics theoretical and experimental studiesPhysics and Astronomy3 citations

TL;DR

This study presents the first measurement of CollinearDrop jet mass and its correlation with SoftDrop groomed jet substructure observables in $√{s}=200$ GeV $pp$ collisions at STAR. Using the MultiFold machine learning unfolding method, it demonstrates that higher CollinearDrop mass fractions enhance non-perturbative contributions and reveals an anti-correlation between grooming extent and the angular scale of the first hard splitting, providing insight into parton shower dynamics via early soft wide-angle radiation effects.

ABSTRACT

Jet substructure variables aim to reveal details of the parton fragmentation and hadronization processes that create a jet. By removing collinear radiation while maintaining the soft radiation components, one can construct CollinearDrop jet observables, which have enhanced sensitivity to the soft phase space within jets. We present a CollinearDrop jet measurement, corrected for detector effects with a machine learning method, MultiFold, and its correlation with groomed jet observables, in $pp$ collisions at $\sqrt{s}=200$ GeV at STAR. We demonstrate that the population of jets with a large non-perturbative contribution can be significantly enhanced by selecting on higher CollinearDrop jet mass fractions. In addition, we observe an anti-correlation between the amount of grooming and the angular scale of the first hard splitting of the jet.

Motivation & Objective

• To probe the soft and wide-angle radiation phase space in jets, which is suppressed in standard SoftDrop measurements.
• To measure the CollinearDrop jet mass ($\Delta M/M$) as a proxy for early-stage, non-perturbative radiation in parton showers.
• To investigate the correlation between CollinearDrop mass and SoftDrop observables ($R_{\mathrm{g}}$, $z_{\mathrm{g}}$) to understand how early radiation constrains later splitting dynamics.
• To validate event generator predictions (PYTHIA8, HERWIG7) against data using a novel multi-dimensional unfolding technique.
• To demonstrate the effectiveness of the MultiFold machine learning method for correcting detector effects while preserving multi-dimensional correlations in jet substructure observables.

Proposed method

• The CollinearDrop mass fraction $\Delta M/M = (M - M_{\mathrm{g}})/M$ is measured, where $M$ is the ungroomed jet mass and $M_{\mathrm{g}}$ is the SoftDrop groomed mass.
• SoftDrop grooming is applied with parameters $(z_{\mathrm{cut}}, \beta) = (0.01, 0)$, effectively measuring the mass lost due to soft wide-angle radiation.
• The MultiFold machine learning unfolding method is used to correct for detector effects, preserving correlations across multiple jet substructure observables.
• The method is validated against RooUnfold in one-dimensional distributions and applied to 2D and higher-dimensional correlation measurements.
• Jet reconstruction uses TPC tracks and BEMC towers with $20 < p_{\mathrm{T}} < 30$ GeV/$c$, excluding jets with $M = M_{\mathrm{g}}$ to enhance visibility of small $\Delta M$.
• Additional measurements of $\mu = \max(M_1, M_2)/M_{\mathrm{g}}$ are performed to probe mass sharing in hard splittings, corrected via Bayesian unfolding.

Experimental results

Research questions

• RQ1 How does the amount of soft wide-angle radiation, captured by $\Delta M/M$, correlate with the angular scale of the first hard splitting ($R_{\mathrm{g}}$) in jets?
• RQ2 How does the momentum imbalance of the first hard splitting ($z_{\mathrm{g}}$) relate to the non-perturbative mass contribution ($\Delta M/M$)?
• RQ3 To what extent do event generators like PYTHIA8 and HERWIG7 describe the observed correlations between CollinearDrop and SoftDrop observables?
• RQ4 Can the MultiFold machine learning unfolding method effectively preserve multi-dimensional correlations in jet substructure measurements while correcting for detector effects?
• RQ5 How does early-stage radiation constrain the phase space of later parton shower splittings, as revealed by the interplay of $\Delta M/M$, $R_{\mathrm{g}}$, and $z_{\mathrm{g}}$?

Key findings

• An anti-correlation is observed between the amount of SoftDrop grooming and the angular scale $R_{\mathrm{g}}$ of the first hard splitting, consistent with angular ordering in parton showers.
• Jets with larger $R_{\mathrm{g}}$ exhibit a sharper $\Delta M/M$ distribution peaked at small values, indicating that wide-angle splittings leave little room for additional grooming.
• Jets with smaller $R_{\mathrm{g}}$ show a broader $\Delta M/M$ distribution, suggesting that narrow splittings allow for more variable grooming, possibly due to earlier soft radiation.
• The $z_{\mathrm{g}}$ distribution becomes flatter with increasing $\Delta M/M$, indicating that larger non-perturbative mass contributions correlate with less momentum imbalance in the first splitting.
• PYTHIA8 Detroit and HERWIG7 H7.1-Default tunes describe the general trends in the data, though some tension is observed in the small $\Delta M$ region with HERWIG.
• The measurement of $\mu$ shows weaker dependence on $R_{\mathrm{g}}$ than $\Delta M/M$, and is well described by PYTHIA6 STAR tune and PYTHIA8 Monash tune, indicating that narrower splittings transfer less virtuality.