Speaker
Description
TFPWA is a general framework for partial wave analysis developed based on TensorFlow2. Partial wave analysis is a powerful method to determinate the 4-momentum distribution of multi-body decay final states and to extract the interested internal information. Base on a simple topological representation, TF-PWA can deal with most of the processes of partial wave analysis automatically and efficiently. The required inputs are only a simple YAML configuration file to describe the decay process and 4-momentum data. The helicity amplitude will be built though the configuration automatically. Helicity angles are also calculated automatically though the topology. The user-friendly interface for customizing decay models allow a large variant of popular or user-specific dynamic functions in constructing the amplitude conveniently. Taking the advantage of acceleration and optimization mechanism in TensorFlow, such as GPU calculation and automatic differentiation, the amplitude fitting speed is extremely fast. In a predefined main script, most of outputs will be generated though the configuration, including fit parameters, fit uncertainties, partial wave plots, fit fractions and toy generations. No additional coding efforts are required in most cases. There are already some published analyses based on TF-PWA framework at LHCb and BESIII. The source code is available at https://github.com/jiangyi15/tf-pwa.
| Consider for long presentation | No |
|---|
