Tree boosting for learning EFT parameters

We present a new tree boosting algorithm designed for the measurement of parameters in context effective field theory (EFT). To construct algorithm, we interpret optimized loss function traditional decision as maximal Fisher information Poisson counting experiments. promote interpretation to general EFT predictions and develop suitable method. The resulting “Boosted Information Tree” approximates score, derivative log-likelihood with respect parameter. It thus provides sufficient statistic vicinity reference point parameter space where estimator is trained. training exploits per-event likelihood ratios different values available simulated data sets. Program Title: BIT (Boosted Trees) CPC Library link program files: https://doi.org/10.17632/9fjyb5hyxt.1 Developer's repository link: https://github.com/HephyAnalysisSW/BIT Licensing provisions: GPLv3 Programming language: Python2 Python3 Nature problem: Providing discriminator estimation standard model theory. Solution method: A tree-based “augmented” set regress score thereby test an