Scholarlink

<p>Resolving inconsistencies among historical photonuclear cross-section measurements is essential for reliable nuclear data evaluation. Significant discrepancies exist between the Livermore and Saclay datasets for the <sup>127</sup>I(<i>γ</i>, n) reaction, leading to long-standing uncertainties in evaluated databases. In this work, a Bayesian neural network (BNN) framework is applied to assess the systematic consistency of existing experimental data. The evaluation predicts that Bergère et al. (1969) measurements are mutually consistent within uncertainty, whereas both Livermore Bramblett et al. (1966) and Berman et al. (1987) measurements exhibit a systematic underestimation of the cross section, while available (<i>γ</i>, 2n) data remain consistent across laboratories. To independently test this prediction, new high-precision measurements of the <sup>127</sup>I(<i>γ</i>, n) cross section were performed at the SLEGS beamline using quasi-monochromatic <i>γ</i> rays produced via inverse Compton scattering. The new data, with total uncertainties below 4%, agree with the BNN evaluation and Bergè re et al. (1969) results over the full energy range, while confirming significant deviations from the Livermore measurements near the cross-section maximum. The agreement between data-driven evaluation and independent experiment demonstrates that machine-learning approaches can reliably identify systematic biases in legacy nuclear datasets. These results provide improved constraints on the <sup>127</sup>I photoneutron cross section and establish a validated framework for modern nuclear data evaluation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Photoneutron cross section data for 127I: Evaluation based on data systematics and new experiments

  • 期刊论文

  • 发表日期：

    2026年7月28日

查看全文
  • Qian-Kun Sun,
  • Zi-Rui Hao,
  • Yue Zhang,
  • Hang-Hua Xu,
  • Long-Xiang Liu,
  • Sheng Jin,
  • Kai-Jie Chen,
  • Yu-Xuan Yang,
  • Zhi-Cai Li,
  • Pu Jiao,
  • Zhen-Wei Wang,
  • Meng-Die Zhou,
  • Meng-Ke Xu,
  • Xiang-Fei Wang,
  • Yu-Long Shen,
  • Jia-Wen Ding,
  • Yong Zhu,
  • Shi-Hai Yue,
  • Li-Jun Gai,
  • Gong-Tao Fan,
  • Hong-Wei Wang

摘要

Resolving inconsistencies among historical photonuclear cross-section measurements is essential for reliable nuclear data evaluation. Significant discrepancies exist between the Livermore and Saclay datasets for the 127I(γ, n) reaction, leading to long-standing uncertainties in evaluated databases. In this work, a Bayesian neural network (BNN) framework is applied to assess the systematic consistency of existing experimental data. The evaluation predicts that Bergère et al. (1969) measurements are mutually consistent within uncertainty, whereas both Livermore Bramblett et al. (1966) and Berman et al. (1987) measurements exhibit a systematic underestimation of the cross section, while available (γ, 2n) data remain consistent across laboratories. To independently test this prediction, new high-precision measurements of the 127I(γ, n) cross section were performed at the SLEGS beamline using quasi-monochromatic γ rays produced via inverse Compton scattering. The new data, with total uncertainties below 4%, agree with the BNN evaluation and Bergè re et al. (1969) results over the full energy range, while confirming significant deviations from the Livermore measurements near the cross-section maximum. The agreement between data-driven evaluation and independent experiment demonstrates that machine-learning approaches can reliably identify systematic biases in legacy nuclear datasets. These results provide improved constraints on the 127I photoneutron cross section and establish a validated framework for modern nuclear data evaluation.