Abstract
APPRENTICE is a tool developed for event generator tuning. It contains a range of conceptual improvements and extensions over the tuning tool Professor. Its core functionality remains the construction of a multivariate analytic surrogate model to computationally expensive Monte-Carlo event generator predictions. The surrogate model is used for numerical optimization in chi-square minimization and likelihood evaluation. Apprentice also introduces algorithms to automate the selection of observable weights to minimize the effect of mis-modeling in the event generators. We illustrate our improvements for the task of MC-generator tuning and limit setting.
Highlights
Monte Carlo-based (MC) event generators are necessary tools for interpreting data at the high energy frontier
For the large datasets of collider data that are available for MC tuning, a common heuristic to evaluate the goodness of a prediction is: χ2(p, w)
When the theory prediction tb(p) is based on simulated data from the MC event generator, the computational cost is high, severely limiting the number of parameter choices p that can be explored in the tuning
Summary
Monte Carlo-based (MC) event generators are necessary tools for interpreting data at the high energy frontier. Where SO is the set of observables O used in the tune, each observable has a weight wO represented by a vector w, tb(p)/Rb is the theory prediction/reference data in a given bin b of an observable and the ∆tb/∆Rb’s are error estimates on these quantities. Our problem is to minimize χ2(p, w) as a function of the adjustable parameters p and possibly the observable weights w. One needs a range of theory predictions tb for different possible parameter choices p and a method or principle for choosing w. Based on a modest number of MC simulations, it constructs a polynomial approximation surrogate to these predictions in each bin of a histogram representing the observables that drive the tuning. An "optimal" set of tuned parameters p is determined by minimizing the heuristic (1). We motivate the use of apprentice by listing the problems that apprentice can currently solve and by highlighting the advantages of using apprentice to solve these problems, especially for HEP applications
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
