Gaussian Process-based calculation of look-elsewhere trials factor

Abstract In high-energy physics it is a recurring challenge to efficiently and precisely (enough) calculate the global significance of, e.g., potential new resonance. We propose method that models in search region as Gaussian Process. The kernel of Process approximated with covariance matrix calculated carefully designed set background-only data sets, comparable number random sets used typical analysis relies on average upcrossings significance. trials factor for both low moderate significances can subsequently be desired precision computationally inexpensive sampling addition, once determined, computed analytically. our work we give some highlights analytic calculation also discuss peculiarities estimation finite grid. illustrate studies three complementary statistical models.