标题： 关于粒子物理中搜索和测量的聚焦统计功效 显示英文标题

标题： On Focusing Statistical Power for Searches and Measurements in Particle Physics

摘要： 粒子物理实验依赖于（广义）似然比检验（LRT）进行搜索和测量，这些检验包括复合假设检验。然而，这种检验并不保证是最优的，因为Neyman-Pearson引理仅适用于简单假设检验。因此，任何检验统计量的选择都会隐式地决定统计功效在参数空间中的变化方式。对具有复合检验的一般情况下的核心统计检验方法进行改进，将在各种实验中产生广泛的影响。我们讨论了一种替代的检验统计量，它使数据分析者能够将检验的功效集中在物理驱动的参数空间区域上。我们在ATLAS实验模拟的希格斯$\rightarrow\tau\tau$数据集和受LZ实验启发的暗物质数据集上展示了该技术相对于LRT的改进。我们还利用机器学习高效地执行Neyman构造，这是确保统计有效置信区间的必要步骤。 显示英文摘要

摘要： Particle physics experiments rely on the (generalised) likelihood ratio test (LRT) for searches and measurements, which consist of composite hypothesis tests. However, this test is not guaranteed to be optimal, as the Neyman-Pearson lemma pertains only to simple hypothesis tests. Any choice of test statistic thus implicitly determines how statistical power varies across the parameter space. An improvement in the core statistical testing methodology for general settings with composite tests would have widespread ramifications across experiments. We discuss an alternate test statistic that provides the data analyzer an ability to focus the power of the test on physics-motivated regions of the parameter space. We demonstrate the improvement from this technique compared to the LRT on a Higgs $\rightarrow\tau\tau$ dataset simulated by the ATLAS experiment and a dark matter dataset inspired by the LZ experiment. We also employ machine learning to efficiently perform the Neyman construction, which is essential to ensure statistically valid confidence intervals.