Statistical Methods in Experimental Particle Physics

The course consists of a series of guided exercises in computational statistics of gradually increasing complexity, progressing from basic simulations, estimators, confidence intervals, fits, up to realistic case studies in experimental particle physics. Lectures will be held in the classroom once or twice a week. Students are expected to turn in the results at the next class, where they will be discussed, and they will eventually produce written summaries. The concluding case studies will exemplify in a reduced form some specific real research problems in experimental particle physics. More time will be allowed for their solution, and the corresponding written reports will be part of the final exam.  

Below is a non-exhaustive list of examples of guided exercises:

• branching fraction measurement of a given decay process (multidimensional unbinned likelihood fit);
• particle Identification using Cherenkov light emission(test of hypotheses, likelihood ratio,…), 
• lifetime measurement of a given decay process (resolution convolution and background subtraction,…)
• search for a rare decay process over background (significance, interval estimation, look-elsewhere effect);
• vertex fit in a silicon tracking detector (least squares fit and constraints)
• measurement of the detection asymmetry of neutral kaons passing through a block of material (profile likelihood, profile likelihood with nuisance parameters)
• …

The first part of the course will introduce concepts of classical statistical data analysis such as probability, information, point estimation, interval estimation and hypothesis testing (including goodness-of-fit tests), as well as practical examples of how to use the ROOT data analysis software in the C++ programming language. Each student can perform their work using their own preferred software platform, although ROOT is strongly recommended (and supported) by the teacher. However, it is a prerequisite for this course that students have the necessary knowledge to use a programmable software system capable of performing simple calculations, generating random numbers, creating histograms, and plotting graphs.