Superphysics and Supercomputers
This module takes place online.
Rapid development of computer technologies has led to many core computer architectures with thousands of computational elements in a single device. In addition to the quantitative growth, these devices operate simultaneously and in parallel. In addition, if operation of one computational element is similar to the rational analysis of a human brain, the parallel operation of thousand such computational elements is capable to reproduce activity of a human brain at its intuitive level.
Thus, artificial neural networks inspired by the work of the human brain have received a powerful boost for their application at the forefront of modern science, primarily in the analysis of data obtained at the Large Hadron Collider in CERN (Geneva), where experiments in high energy and nuclear physics are aimed at studying a wide range of phenomena from the Higgs boson to the Big Bang and black holes.
Programme – The course is taught in English (4 ECTS)
Requirements – Knowledge of advanced mathematics (calculus, statistics); at least one programming language (Python, Java, C++ or similar); at least 2 years of undergraduate studies or enrolment in master studies; topics of study: computer sciences, engineering, physics, mathematics, chemistry, biology or materials
Programme fee – EUR 500.00
Application deadline – n/a
If you are a student from the University of Massachusetts system, the University of Wisconsin system, and participating universities in Queensland you will participate as exchange students and will not pay fees directly to Frankfurt Digital Summer School. Please contact your study abroad advisor for more information on how and when to apply.
Module „Superphysics and Supercomputers“ is designed for two weeks and can be considered as introductory and complementary to the module „Artificial Intelligence (AI): Modern Methods“.
The module program includes 5 topics, as well as a final one-day workshop with student presentations on current topics in modern physics and supercomputers.
The aim of the module is to learn detailed information and acquire skills in the areas:
- Experimental High Energy Physics and Heavy Ion Physics,
- HPC and many-core CPU/GPU computer architectures,
- Parallel programming compatible with advanced computer architectures,
- Artificial human-like thinking,
- The future of parallel computing.
In accordance with these goals, lessons are organized into 5 topics.
Topic 1: Introduction to experimental physics.
This topic gives a basic knowledge of the Standard Model of elementary particles, discusses the types of modern experiments and their detector facilities, the role of high-performance computer architectures, parallel languages and fast algorithms for processing and analyzing experimental data. The specific features of acquiring experimental data are given and the stages of their processing are discussed in detail. The limits of applicability of traditional data processing methods and the necessity and capabilities of artificial intelligence methods are discussed. The importance of synergy between physics, computer science, programming and artificial intelligence methods for efficient operation of physics experiments is discussed.
Topic 2: Basics of C++
This topic provides the basics of C++ programming necessary to understand examples of modern algorithms for processing experimental data, the particular applications of which are discussed in this module. The basics of object-oriented programming, its advantages and disadvantages, as well as the limitations of using it when working on modern many-core computer architectures are discussed. Methods of optimal programming are discussed and an example of code analysis is given.
Topic 3: Many-core CPU/GPU architectures
This topic is entirely devoted to an overview of modern many-core computer architectures, a brief review of their history, and a detailed discussion of key stages and necessity of evolution in the development of CPU and GPU architectures, including the need for parallelism and its implementation levels. The distinctive features of Intel, AMD and Nvidia approaches are discussed.
Topic 4: Parallel programming
The principles of parallel programming are discussed in detail, as well as their relation to many-core computer architectures. The basics of parallel programming both within a core (SIMD, vectorization) and between cores within a single processor are given. The specifics of parallel programming on graphics cards, its difference from programming on the CPU, and the need to implement algorithms in a universal form with a single code that can work effectively on both the CPU and GPU are discussed.
Topic 5: Modern HEP and HI experiments
A brief review of the major world scientific centers (CERN, Switzerland; BNL, USA; FAIR, Germany) and their largest experiments in high-energy and heavy-ion physics: ATLAS, CMS, ALICE and LHCb (CERN), PHENIX and STAR (BNL), PANDA and CBM (FAIR). The future Compressed Baryonic Matter (CBM, FAIR) heavy ion experiment is explained in detail, the goals of its physical program, the design of its detector facility, the size and architecture of the computer farm, features of processing algorithms and stages of analysis of experimental data.