pith. sign in

arxiv: 2606.22750 · v1 · pith:SKT7K7OWnew · submitted 2026-06-22 · ⚛️ nucl-ex

Development of a Neural Network-Based Background Suppression Technique for Sigma N Cusp Spectroscopy at J-PARC

K. Amemiya , Y. Ichikawa , S.H. Hayakawa , K. Tanida , S.H. Kim , F. Oura , H.I. Lee , R.J. Saito
show 3 more authors
K. Shimazaki R. Sasaki Y. Nakayama
This is my paper

Pith reviewed 2026-06-26 06:10 UTC · model grok-4.3

classification ⚛️ nucl-ex
keywords neural networkbackground suppressionΣN cuspJ-PARCHyperon TPCtrack multiplicitymissing mass spectrumsignal to noise ratio
0
0 comments X

The pith

Neural network background suppression doubles usable statistics for ΣN cusp spectroscopy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a neural network to identify clean events with two charged tracks in the Hyperon Time Projection Chamber for the d(K-, π-) reaction at J-PARC. This allows use of Mt=2 events that have more statistics but heavy background, by training on track topology and energy loss to separate signal from quasi-free backgrounds. The method achieves a signal-to-noise ratio comparable to the conventional three-track cut while avoiding mass-dependent biases that distort the spectrum. Combining the machine-learning selected Mt=2 sample with the standard Mt=3 sample effectively doubles the total usable events. This increase in statistics improves the precision for extracting parameters of the ΣN cusp near threshold.

Core claim

A binary classification neural network using HypTPC track topology and dE/dx as input features discriminates signal from quasi-free backgrounds in Mt=2 events, achieving a signal-to-noise ratio comparable to the strict Mt=3 condition while preserving the integrity of the spectral shape. By combining this independent ML-selected Mt=2 sample with the conventional Mt=3 sample, the total usable statistics are effectively doubled compared to traditional methods, significantly enhancing the sensitivity for determining the ΣN cusp parameters.

What carries the argument

The binary classification neural network that takes HypTPC track topology and energy loss (dE/dx) as inputs to select clean Mt=2 events without mass-dependent acceptance bias.

If this is right

  • The ML-selected Mt=2 sample can be combined independently with the Mt=3 sample to double statistics.
  • The combined dataset maintains an undistorted spectral shape for accurate cusp parameter determination.
  • The approach achieves signal-to-noise performance matching the strict multiplicity cut but with higher event yield.
  • Overall sensitivity to ΣN cusp parameters is significantly enhanced due to increased statistics.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • This technique may reduce the need for strict track multiplicity cuts in similar hyperon production experiments.
  • Future experiments could test the neural network on simulated data with known cusp parameters to verify shape preservation.
  • The doubled statistics could allow resolution of finer structures in the missing-mass spectrum beyond the target 0.4 MeV.

Load-bearing premise

The neural network selects Mt=2 events without introducing a mass-dependent acceptance bias or distorting the spectral shape.

What would settle it

A direct comparison of the spectral shape or extracted cusp parameters from the ML-selected Mt=2 sample versus the Mt=3 sample, where any significant mass-dependent difference would indicate bias.

read the original abstract

A clear spectral enhancement, known as the ``$\Sigma N$ cusp'', has been observed near the $\Sigma N$ threshold in the $d(K^-, \pi^-)$ reaction. To understand the dynamical origin of this enhancement, the J-PARC E90 experiment aims to investigate the missing-mass spectrum with an unprecedented resolution of 0.4 MeV ($\sigma$). In this experiment, a Hyperon Time Projection Chamber (HypTPC) is utilized to detect charged decay products and suppress severe contamination from quasi-free (QF) background processes. While a conventional track multiplicity condition of three (Mt=3) effectively suppresses these QF events, it restricts the signal statistics to approximately 17\% and introduces a mass-dependent acceptance bias that distorts the spectrum. In contrast, events with a track multiplicity of two (Mt=2) offer roughly double the statistical power ($\sim$39\%) with minimal mass dependence, but they suffer from heavy background contamination. To fully exploit the Mt=2 events, we developed an innovative background suppression technique based on a neural network. By constructing a binary classification model using the HypTPC track topology and energy loss ($dE/dx$) as input features, we successfully discriminated the signal from QF backgrounds. This machine learning approach achieves a signal-to-noise ratio comparable to the strict Mt=3 condition while preserving the integrity of the spectral shape. By combining this independent ML-selected Mt=2 sample with the conventional Mt=3 sample, the total usable statistics are effectively doubled compared to traditional methods, significantly enhancing the sensitivity for determining the $\Sigma N$ cusp parameters.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript describes the development of a neural network classifier for suppressing quasi-free background in Mt=2 events from the d(K^-, π^-) reaction at J-PARC E90. Using HypTPC track topology and dE/dx as inputs, the NN is claimed to achieve signal-to-noise performance comparable to the conventional Mt=3 multiplicity cut while avoiding mass-dependent acceptance bias, thereby allowing the Mt=2 sample (~39% of events) to be combined with the Mt=3 sample (~17%) to roughly double usable statistics for ΣN cusp spectroscopy.

Significance. If the performance and shape-preservation claims are substantiated, the technique would meaningfully increase statistical sensitivity for extracting ΣN cusp parameters at the target 0.4 MeV resolution, addressing a key limitation of traditional multiplicity cuts in hyperon missing-mass experiments.

major comments (1)
  1. [Abstract] Abstract: the central claim that the NN-selected Mt=2 sample achieves SNR comparable to Mt=3 while preserving spectral shape (no mass-dependent bias) is asserted without any quantitative metrics, efficiency curves versus missing mass, closure-test results on Monte Carlo, or sideband comparisons. This empirical performance statement is load-bearing for the doubling-of-statistics conclusion and requires explicit validation in the results section.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback. The major comment is addressed below with a commitment to strengthen the manuscript's empirical validation.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the NN-selected Mt=2 sample achieves SNR comparable to Mt=3 while preserving spectral shape (no mass-dependent bias) is asserted without any quantitative metrics, efficiency curves versus missing mass, closure-test results on Monte Carlo, or sideband comparisons. This empirical performance statement is load-bearing for the doubling-of-statistics conclusion and requires explicit validation in the results section.

    Authors: We agree that the abstract's performance claims require explicit quantitative support in the results section to substantiate the SNR comparability and lack of mass-dependent bias. The revised manuscript adds efficiency curves versus missing mass, Monte Carlo closure-test results, and sideband comparisons in the results section. These additions directly validate the claims and support the doubling-of-statistics conclusion. The abstract is also updated to reference these validations. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents an empirical ML classifier for background suppression in experimental data, with performance claims (SNR comparable to Mt=3, preserved spectral shape, doubled statistics) framed as outcomes of applying a trained neural network to held-out or independent samples. No equations, derivations, fitted parameters renamed as predictions, or self-citation chains appear in the provided text. The central claim rests on observable classifier behavior rather than any self-definitional or constructionally forced step, making the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are stated. The central claim rests on the empirical performance of an unspecified neural-network model whose training procedure, loss function, and validation strategy are not detailed here.

pith-pipeline@v0.9.1-grok · 5875 in / 1230 out tokens · 34455 ms · 2026-06-26T06:10:27.182441+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

16 extracted references

  1. [1]

    R. H. Dalitz, Nucl. Phys. A354, 101 (1981)

    1981

  2. [2]

    R. H. Dalitz and A. Deloff, Czech. J. Phys. B32, 1021 (1982)

    1982

  3. [3]

    T. H. Tan, Phys. Rev. Lett.23, 395 (1969)

    1969

  4. [4]

    Braunet al., Nucl

    O. Braunet al., Nucl. Phys. B124, 45 (1977)

    1977

  5. [5]

    Eastwoodet al., Phys

    D. Eastwoodet al., Phys. Rev. D3, 2603 (1971)

    1971

  6. [6]

    Pigotet al., Nucl

    C. Pigotet al., Nucl. Phys. B249, 172 (1985)

    1985

  7. [7]

    Budzanowskiet al., Phys

    A. Budzanowskiet al., Phys. Lett. B692, 10 (2010)

    2010

  8. [8]

    Acharyaet al.(ALICE Collaboration), Phys

    S. Acharyaet al.(ALICE Collaboration), Phys. Lett. B833, 137272 (2022)

    2022

  9. [9]

    Ichikawaet al., Prog

    Y. Ichikawaet al., Prog. Theor. Exp. Phys.2014, 101D03 (2014)

    2014

  10. [10]

    Ichikawa, K

    Y. Ichikawa, K. Tanidaet al.(J-PARC E90 Collaboration), J-PARC Proposal E90 (2021)

    2021

  11. [11]

    Amemiya, Master’s thesis, Tohoku University (2026)

    K. Amemiya, Master’s thesis, Tohoku University (2026)

    2026

  12. [12]

    Ichikawa, JPS Conf

    Y. Ichikawa, JPS Conf. Proc.33, 011103 (2021)

    2021

  13. [13]

    S. H. Kimet al., Nucl. Instrum. Methods Phys. Res. A940, 359-370 (2019)

    2019

  14. [14]

    Miyagawa and H

    K. Miyagawa and H. Yamamura, Phys. Rev. C60, 024003 (1999)

    1999

  15. [15]

    Akibaet al., Proc

    T. Akibaet al., Proc. 25th ACM SIGKDD Int. Conf. on Knowledge Discovery & Data Mining, 2623 (2019)

    2019

  16. [16]

    S. M. Lundberg and S.-I. Lee, Adv. Neural Inf. Process. Syst.30, 4765 (2017). 25

    2017