DATA ANALYSIS TECHNIQUES FOR NUCLEAR AND PARTICLE PHYSICS

Academic Year 2024/2025 - Teacher: Gioacchino ANASTASI

Expected Learning Outcomes

The course introduces the fundamental topics of data analysis for an experiment in nuclear and particle physics, dealing with both the programming tools and the statistical methods to employ. The students will acquire the ability to develop codes for the analysis of data, using object-oriented programming and the main packages of the ROOT software framework. Moreover he/she will learn to critically interpret the issues in programming and analysis during the practical lessons. He/she will learn to effectively communicate such knowledge thanks to the resolution of a practical test of programming and analysis and the following discussion of the topics assimilated during the lectures. The students will be motivated to proceed autonomously the study by encouraging the development of analyses to be employed during the initial steps of their master thesis research.

Referring in particular to the Dublin Descriptors, this course is intended to reinforce :
- Knowledge and understanding :
Understanding object-oriented programming. Knowledge of the fundamental tools and applications of the ROOT software. Knowledge of the fundamental techniques of statistical data analysis and the related test of hypotheses.
- Applying knowledge and understanding :
Capability of developing a data analysis framework, using the correct statistical and programming tools. Ability of interpret and solve the common issues and errors in coding in a UNIX environment.
- Making judgments :
Ability of identify the key elements for the data analysis in a nuclear or particle physics experiment. Discuss the choices for the development of a data analysis framework.
- Communication skills :
Ability of describing the main functionalities of a code for data analysis and the related implementations. Usage of the correct terminology in presenting the results of a statistical analysis.
- Learning skills :
Capability of study autonomously advanced statistical methods for data analysis. Ability to comprehend autonomously the key elements of data analysis in state-of-the-art researches in nuclear and particle physics.
----------

Course Structure

Lectures on theoretical parts (35 hours) followed by practical sessions (15 hours).
----------

Required Prerequisites

(Useful) Knowledge of the most common detection techniques and data-acquisition systems in the context of nuclear and particle physics. Knowledge of the theory of fundamental interactions in nuclear and particle physics, and of the interactions of radiation with matter.

----------

Attendance of Lessons

It is highly recommended to attend both lectures and practical sessions during the entire didactic period.

----------

Detailed Course Content

Introduction to data analysis techniques in High Energy Physics
Object-Oriented Programming
ROOT software
Probability density functions and Monte Carlo methods
Statistical methods for data analysis
----------

Textbook Information

G. Cowan, Statistical data analysis, O.U.P. (1998)
Lecture notes provided during classes and by using Teams.

Course Planning

	Subjects	Text References
1	Introduction to the data analysis in nuclear and particle physics
2	Introduction to UNIX-based operative systems (Basic commands. Conventions. Input/Output. Permissions. Processes & jobs.)
3	Introduction to C++ (part I : Variables & operators. Input/Output. Arrays, pointers & references. Functions.)
4	Introduction to C++ (part II : Object-oriented programming. Inheritance. Operator overloading. Template class.)
5	The ROOT analysis framework (part I : Introduction & GUI. Histogram. TGraph & TProfile.)
6	The ROOT analysis framework (part II : Scripts & ACLiC. TFile and TTree. Fit of graphs and distributions.)
7	Frequentist and Bayesian statistics approach. Random variables. Probability density and distribution functions. Examples of probability functions.
8	Introduction to the Monte Carlo method.
9	Test of hypothesis. Goodness-of-fit. Chi2 test. Estimators. Maximum likelihood method. Least-squares method.
10	Confidence intervals. Applications to Gaussian estimators and Maximum Likehood results. Limits for a signal, with and without background, in the frequentist and Bayesian approaches.
11	Development and data analysis for a real physics application.

Learning Assessment

Learning Assessment Procedures

The exam includes:
-) a practical part, during which the candidate is provided with experimental data to analyze by realizing a program with the knowledge obtained during the course;
-) an oral part, during which the methods employed in the practical test and the related results are discussed, together with a few questions about the topics encountered during the lectures (in particular regarding the statistical techniques of data analysis) to ascertain the knowledge level acquired by the candidate.

In the evaluation the pertinence of the answers, the level of detail in the exposed contents, the capability of making examples, the correct use of language and the clarity in the exposition will be taken into consideration.

The exam dates will be made available on the DFA website and on the "portale esami".

Examples of frequently asked questions and / or exercises

The concept of inheritance in C++
Built and use ROOT TTree & TFile
Fit of a distribution with the tools available in ROOT
Probability distribution functions (Poisson, binomial, multinomial, ...)
Monte Carlo method and random sampling
Estimators for mean and variance in the Maximum Likelihood technique

Degree Course in

Physics