⚛️ phenomenology

An effective topology-transfer nuisance for Genie-based neutrino-argon semi-inclusive analyses: a methodological proposal with MicroBooNE-constrained diagnostic

This paper proposes and validates a single empirical nuisance parameter, $f_{\rm mig}$ , to address the lack of topology-transfer capabilities in standard GENIE simulations for liquid-argon neutrino analyses, demonstrating through MicroBooNE data that this parameter effectively captures near-threshold proton mismodelling effects with a statistically significant constraint of $0.093 \pm 0.024$ .

Original authors: Nicolas Viaux

Published 2026-09-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Nicolas Viaux

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

Deep inside the universe, invisible particles called neutrinos stream through everything, passing through stars, planets, and even our own bodies without leaving a trace. To catch a glimpse of them, scientists build massive detectors filled with liquid argon, a super-cold metal that glows when a neutrino strikes an atom inside it. These detectors act like giant, three-dimensional cameras, capturing the tiny flashes of light left behind by the collision. By studying these flashes, physicists can learn how neutrinos behave and how they change as they travel, which is crucial for understanding the fundamental makeup of the cosmos. However, to make sense of the data, researchers rely on complex computer simulations that predict what should happen when a neutrino hits an atom. If the simulation gets the details wrong, the entire picture of the universe could be skewed.

A specific problem has emerged in these simulations regarding protons, the positively charged particles found in the center of atoms. When a neutrino strikes an argon atom, it can knock a proton loose. If that proton has enough energy, it travels far enough to be seen by the detector. But if the proton is just barely moving, it might stop before it can be detected, making it look like no proton was ever there at all. This distinction is critical because scientists sort their data into two groups: events where a proton is seen, and events where no proton is seen. The computer program used to simulate these events, known as Genie, has been missing a subtle effect that causes some of these barely-moving protons to lose just enough energy to disappear from the detector's view. This means the simulation is counting too many "proton-seen" events and too few "proton-missing" events, creating a gap between what the computer predicts and what the real detector observes.

Researchers at the MicroBooNE experiment, which uses a liquid argon detector in Illinois, noticed this discrepancy. They found that their real data contained significantly more events with no visible protons than the standard computer model predicted. To solve this, a team led by physicist Nicolas Viaux proposed a new way to fix the simulation without having to rebuild the entire computer model from scratch. Instead of trying to simulate every tiny force inside the atom, they introduced a single, simple adjustment factor. Think of this factor as a dial that the researchers can turn to move a small percentage of events from the "proton-seen" category into the "proton-missing" category, effectively correcting the balance.

The team tested this idea using two different sets of data from the MicroBooNE detector. In the first test, they looked at a specific sideband of data designed to study other types of particle interactions. This analysis required certain assumptions about the simulation's behavior, known as a "prior model." By adjusting their new dial, they found that the simulation matched the real data much better. The best setting for this dial suggested that about five percent of the events that the computer thought had a visible proton should actually be counted as having no visible proton. This adjustment not only fixed the count of missing protons but also relieved a strange pressure in the model where other parts of the simulation had to be artificially tweaked to compensate for the error.

To be absolutely sure this was the right answer, the team performed a second, even more direct test using a separate dataset that measured the rates of these events directly. This cleaner analysis avoided the extra assumptions used in the first test, relying only on the raw data without the prior model constraints. This direct scan gave an even stronger result. It showed that the simulation was missing a transfer of about nine percent of events from the visible to the invisible category. This result was statistically very strong, indicating that the effect is real and not just a random fluke in the data. However, it is important to note that this specific result is an effective measurement tailored to the MicroBooNE experiment's unique beam configuration; it is not yet a universal constant. The researchers confirmed that this missing piece of physics is likely due to the way protons lose energy as they travel through the dense nuclear material inside the argon atom, a process that the old simulation did not fully capture.

This discovery is important because it provides a practical tool for future experiments, including the massive DUNE detector currently under construction. By adding this simple adjustment to the computer models, scientists can ensure their simulations are more accurate, leading to more reliable measurements of how neutrinos change over long distances. However, before this tool can be used in DUNE, it must undergo further testing at the detector level to ensure it works correctly in that specific environment. The work does not claim to have solved every mystery of how protons behave inside an atom, but it has successfully identified a specific gap in the current models and offered a clear, effective way to fill it. The findings suggest that the universe is slightly more subtle in its handling of these tiny particles than our previous computer models allowed, and that a small correction can make a big difference in our understanding of the cosmos.

Technical Summary: An Effective Topology-Transfer Nuisance for Genie-Based Neutrino-Argon Semi-Inclusive Analyses

Problem Statement
Liquid-argon time projection chamber (LArTPC) oscillation analyses increasingly split neutrino interaction samples into proton-tagged ( $N_p$ ) and proton-less ( $0p$ ) topologies based on a $\sim 35$ MeV proton reconstruction threshold. A critical gap exists in the standard GENIE Monte Carlo Reweight library: it lacks a "dial" capable of transferring events across this $35$ MeV boundary. While GENIE's intranuclear cascade (INC) models nuclear binding via a constant removal energy ( $W_s \approx 25$ MeV), it lacks the position-dependent mean-field potential present in transport codes (e.g., GiBUU). This omission leads to a systematic underestimation of energy loss for near-threshold protons ( $T_p \approx 35\text{--}100$ MeV), causing GENIE to misclassify events that should be $0p$ as $N_p$ . This "proton-visibility migration" distorts the $0p/N_p$ ratio, directly impacting energy reconstruction strategies and oscillation systematic uncertainties. Existing GENIE reweight dials (e.g., MFP $_\pi$ , MFP $_N$ ) act within a topology but cannot redistribute events across the threshold.

Methodology
The author proposes a single empirical nuisance parameter, $f_{mig}$ , defined as a per-event reweight on nominal GENIE output. This parameter acts as an effective topology-transfer surrogate, absorbing an unspecified combination of near-threshold mismodelling effects (proton-side final-state interactions, nucleon-momentum distributions, RES/MEC kinematic shapes, and residual mean-field energy loss).

The methodology involves:

Physical Motivation: A semi-analytic estimate using the $^{40}$ Ar Woods-Saxon density profile and a Walecka-type relativistic mean-field potential (GiBUU NL3* depth) suggests a residual energy loss of $\approx 7.8$ MeV absent in GENIE. This provides a reference scale for the magnitude of $f_{mig}$ .
Parameterization: The migration is implemented as a linear transfer of events from the $N_p$ $N_{p}$ to the $0p$ $0 p$ topology:
- $N_{0p} \to N_{0p} + f_{mig} N_{Np}$
- $N_{Np} \to N_{Np} (1 - f_{mig})$
  This preserves the total event rate while altering the topology split.
Data Constraints: The parameter is constrained using two public MicroBooNE datasets:
- Primary Constraint: A direct scan on the simultaneous CC $0p/N_p$ differential cross-section release (PRD 110). This analysis uses unfolded cross-sections with no sector-weight regularisation, profiling only an overall normalisation.
- Secondary Constraint: A fit to the NC $\Delta$ radiative sideband covariance (64 bins, 4 channels). This fit includes five sector normalisation weights (CCQE+MEC, CCRES, CCDIS, NCRES, NCDIS) with ridge regularisation (Gaussian priors) and two migration fractions ( $f^{CC}_{mig}, f^{NC}_{mig}$ ).
Statistical Treatment: Significance is assessed using profile $\Delta\chi^2$ scans. Due to the low effective degrees of freedom ( $n_{eff} \approx 2.5$ ) in the sideband covariance and the physical boundary ( $f_{mig} \ge 0$ ), the author employs boundary-aware test statistics and validates results with 5,000 pseudo-experiments (toys) to calibrate empirical p-values.

Key Contributions

Identification of a Missing Nuisance: The paper identifies a specific gap in the GENIE reweight basis: the inability to model topology-crossing final-state interactions near the proton detection threshold.
Proposal of $f_{mig}$ : It introduces a methodological diagnostic parameter that can be implemented as a flat per-event weight without event regeneration, making it suitable for future oscillation analyses (e.g., DUNE) pending validation.
Diagnostic By-product: The analysis demonstrates that introducing $f_{mig}$ relieves a compensatory suppression of the CCRES sector weight observed in migration-blind fits. In the absence of a migration dial, the fit artificially suppresses CCRES ( $w_{CCRES} \approx 0.69$ ) to accommodate the $0p$ excess; with $f_{mig}$ , the CCRES weight relaxes toward unity ( $w_{CCRES} \approx 1.07$ ).
Robustness Checks: The paper includes extensive checks on prior sensitivity, covariance structure (Peelle's Pertinent Puzzle), and orthogonality with existing hadronic dials (e.g., MFP $_N$ ).

Results

Primary Constraint (PRD 110): The direct CC scan yields a best-fit migration fraction of $\hat{f}_{mig} = 0.093 \pm 0.024$ . This corresponds to a preference for $f_{mig} > 0$ at 3.83 $\sigma$ (Wilks) and 3.24 $\sigma$ (empirical), with no sector-weight regularisation applied.
Secondary Constraint (Sideband): The NC $\Delta$ sideband fit yields $\hat{f}^{CC}_{mig} = 0.050^{+0.029}_{-0.022}$ . The significance ranges from 2.00 $\sigma$ (uninformative prior) to 2.92 $\sigma$ (physics-informed priors based on external cross-section data) under Wilks, and 2.36 $\sigma$ –3.16 $\sigma$ under empirical calibration.
Consistency: The two results are statistically compatible ( $\approx 1.2\sigma$ separation). The higher value in the PRD 110 scan is attributed to the lack of sector-weight degrees of freedom to absorb normalisation mismatches, forcing the topology tension entirely into $f_{mig}$ .
Physical Interpretation: The fitted magnitude ( $\sim 5\%$ for the sideband, $\sim 9\%$ for the direct scan) is consistent with the $\sim 5\text{--}10\%$ proton-transparency differences reported between cascade (GENIE) and transport (GiBUU) generators for near-threshold protons.

Significance and Claims
The paper claims that $f_{mig}$ is a methodological diagnostic and an effective topology-transfer surrogate, rather than a unique measurement of a specific microscopic mechanism (e.g., mean-field energy loss alone). The parameter absorbs a combination of near-threshold effects including FSI, kinematic mismodelling, and reconstruction thresholds.

The author emphasizes that:

The result is conditional on the prior model used in the sideband fit and the specific covariance structure of the MicroBooNE datasets.
The primary empirical evidence comes from the prior-free direct CC scan, which establishes a strong preference ( $>3\sigma$ ) for a topology-transfer effect.
The parameter is not yet validated for direct use in DUNE oscillation analyses; it requires independent detector-level validation (e.g., with SBND or DUNE-ND) to confirm the migration fraction at the reconstructed level and to distinguish between competing physical origins.
The parameter serves as a candidate Genie dial to address the systematic deficit in the $0p$ topology, which currently forces compensatory adjustments in sector normalisations that may bias oscillation results.

The paper concludes that while the specific physical origin (residual mean field vs. other kinematic effects) cannot be isolated with current data, the existence of a $\sim 5\text{--}10\%$ topology-transfer effect is strongly supported by MicroBooNE data, necessitating a new nuisance parameter in future LArTPC analyses.

Technical Summary: An Effective Topology-Transfer Nuisance for Genie-Based Neutrino-Argon Semi-Inclusive Analyses

More like this