Lectures - Archive

Supercomputing in science


24 May
Natural and artificial intelligence
Institute of Neuroimonulogy SAS

What are the similarities and differences between the intelligence of the biological system itself and the artificial intelligence? The effort to model the processes taking place in the brain was one of the inspirations for the development of artificial intelligence. We will talk about how information is processed in the living brain, the extent to which such artificial intelligence inspiration is usable and justifiable, and where the two types of intelligence differ.
10 May
Current problems of reinforcement learning

Reinforcement learning - what it is and how it works, including demonstrations. The main points of the lecture was key achievements: Atari, GO, Hide and seek and demonstrations of deep reinforcement learning and AlphaGO. The problem of exploitation: rare rewards, such as in roles: Montezuma’s revenge, Hide and seek. We examine the state space as "effective."
26 Apr
Use of ML / AI for applications in chemistry - potential drugs on Covid-19
COO SAS - Computing Centre

How can we transform the structure of molecules into a form that neural networks understand? In this lecture, we want to summarize our efforts to replace the computationally expensive "docking" of molecules into the cavity of a target protein by machine learning and neural network methods. The target protein under study is 3CLpro SARS-CoV-2 (6WQF), which plays a key role in SARS-CoV-2 virus replication. How successful are these methods in preselecting large drug databases to select potential drugs for covid-19?
12 Apr
Introduction to applications of complex calculations in medicine
Department of Computer Graphics and Vision, Department of Applied Informatics FMFI UK, Rádiológia s.r.o.

In medicine, in recent years, observing the exponential growth of data and their processing requires increasingly sophisticated systems. The first part of the presentation outlined the needs and problems of modern medicine with emphasis on the field of radiology. In the second part we will get acquainted with the way of using neural networks to develop algorithms recognizing diagnoses in radiological images.
29 Mar
Deep learning and its basic applications for computer vision
Department of Computer Science FMFI UK

When we use some model of deep learning, it is almost impossible to understand how and why it works based on its architecture. To do this, we need to know the history of how and through what versions the researchers worked on the given architecture. In the lecture we will summarize the basic milestones of this history, starting with the perceptron and ending with the first successful deep neural networks.
15 Mar
Introduction to artificial neural networks and evolutionary algorithms
Department of Robotics and Cybernetics, FEI STU

The presentation defines the terms biological and mathematical neuron, neural networks and their learning, selected architectures of artificial neural networks (UNS) and UNS applications, neuromorphic computing technologies, evolutionary algorithms (EA), genetic algorithm, bio-inspired algorithms, parallelization of EA and HPC and briefly explain their principles.
1 Mar
Use of HPC in operational weather forecast in Slovakia
Slovak hydrometeorological institute

SHMU has been operating HPC since 2004. The primary application is the ALADIN model, developed within the international consortia ALADIN and currently ACCORD. This application contains hydride parallelism (combination of OpenMP, MPI) with the ability to set the length of the vector of internal cycles, so it is suitable for vector and scalar architectures.
15 Feb
Modelling of 2D systems properties using the Quantum Monte Carlo methods
Institute of Physics SAS

Prof. Štich is a partner of the TREX CoE - the European Center of Excellence in Exascale Computing (Horizon 2020 project). The lecture is focused on quantum Monte Carlo computational methods, for which the Nobel Prize in Physics was awarded in 2020 and which are well suited for the exascale computing. You will see a demonstration of calculations that required 30 million core hours to run.


7 Dec
Federated distributed computing
Ladislav Hluchý
Institute of Informatics SAS

Thanks to powerful servers and large data storages, a significant progress can be seen in distributed computing, where the performance achieved is very high (dozens of PFLOPS). In a not-too-distant past we saw the computing clusters being built from personal computers, progressing to powerful servers and data storage units later on. Since 2000 there was a boom of building grid-based distributed infrastructures with distributed operation systems. With the birth of cloud technologies, distributed computing has taken on a new dimension. IT companies such as Google, Amazon, Microsoft, IBM, ATOS brought their technologies to market. The federated cloud infrastructures were introduced, with the support of the European Commission, to avoid the so-called vendor lock-in situation. The individual national cloud infrastructures can join the federation, allowing researchers to run their calculations on an infrastructure that does not impose technological constraints.
23 Nov
Short guide to software parallelization
Jaroslav Suchánek
Biomedical Research Center, SAS

The lecture will highlight few of the most important aspects of example programs and analyses published on the webpage http://www.jsf.name/lectures.html and https://chipster.csc.fi, for example: MS Visual Studio Community and MS MPI. C/C++: preprocessor directives for multiplatform programming, function parameters called by value, reference and dereference, pointers, dynamic memory allocation, classes, stacks for better communication between processes, iterators, etc. MPI: point-to-point and collective communication. Genetic algorithms. Chipster: quick introduction to bioinformatics. Tips for useful reading, links and guidance for begginers.
9 Nov
Operation systems in multiprocessor clusters
Dušan Bernát
Faculty of mathematics, physics and informatics, Comenius University

The lecture focuses on explaining the basic terms, principles and issues in operation systems in multiprocessor architectures. The main areas of interest are process scheduling in a symmetric multiprocessing environment, system load measurement, memory allocation, and cluster file systems.
26 Okt
Development of technology and computer architecture – from 1-processor serial computers to supercomputers
Martin Šperka
Computer museum, Centre of Operations SAS

The presentation will summarize the main stages of accelerating computations due to the technology and architecture development from vacuum tube computers with serial ALUs (arithmetic-logical unit) to massively parallel architectures containing millions of processing units enabled by system-on-chip (SOC) technology with billions of transistors.
12 Okt
Computer simulations and calculations as an essential tool in science
Pavel Neogrády
Faculty of Natural Sciences, Comenius University

Do you want to learn why is the high-performance computing technology indispensable in science? This lecture will introduce - in a simple way - what types of issues the scientists face when modeling various systems and why is the HPC used to solve them. Simultaneously, we will explain methods and procedures essential in current scientific research.
29 Sep
Current trends and visions in HPC in Slovakia.
Lukáš Demovič, Filip Holka
Centre of Operations SAS

In the first part of the lecture you will learn what is a supercomputer and which technologies you can find inside. At the same time, we will show you current trends in building a new generation of supercomputers and how its’ performance and energy efficiency is evaluated. Supercomputer operation and monitoring issues will be covered, as well. The second part of the lecture will focus on the history of supercomputers in Slovakia and our plans for further development and participation in European structures and organizations.