Logo

Research domains - SC4Simulations

print

The basic goals of the project in activities related to supercomputer simulations and the development of algorithms necessary for these simulations (SC4Simulations) can be divided into the following areas, solved by the respective research teams:

  • SC4Industry (Supercomputing for Industry): supercomputer simulations for solving industrial problems.
  • SC4Natural&BioSciences (Supercomputing for Natural and BioSciences): supercomputer modelling and simulation in natural sciences.
  • SC4Nanotechnologies (Supercomputing for Nanotechnologies): modelling with supercomputers in nanotechnologies.

The rapid development of information technologies in the last several years has opened up entirely new opportunities in the computer modelling of complex physical and general natural phenomena. Progress made in this field has found widespread applications in industry. We are currently witnesses to the penetration of quantitative approaches to problem-solving into everyday practice, replacing traditional processes such as produce-test-adjust or produce-sell-improve, which place considerable demands on time and finances. Computer modelling and simulations increasingly enable production problems to be reduced to equations or other relations in forms suitable for computer-based solutions. However, this process is becoming increasingly difficult, and it is not possible to draw a simple dividing line between basic and applied research – applications frequently throw up new problems whose solution often requires an approach integrating both basic and applied research.

Just such an integrated approach is represented in the research team of SC4Natural&BioSciences, which focuses on the development of effective mathematical methods for solving problems that are currently – due to their extent, non-smoothness or non-linearity – out of the scope of standard methods, and the subsequent incorporation of the new methods into the solution of selected complex scientific and engineering problems. Particular emphasis will be placed on the development of algorithms that are able to use effectively modern supercomputers and that are in a sense optimal or scalable. In these cases the main focus will be on algorithms with asymptotically linear complexity for some classes of relevant problems, in which the computation time decreases proportionally to the number of processors. The research will include the development of effective methods for solving basic direct tasks with special structure, including external problems, which will further be used to solve multiphysics problems, multiscale problems, inverse problems of identification of physical parameters, and optimal design problems. Research activities will include all phases of mathematical modelling, from the selection and correct mathematical formulation of problems to their discretization and the design of effective algorithms, the study of computation complexity, analysis of the reliability of results (a posteriori estimates), to the implementation of the new algorithms. With regard to practical applications, there will also be a focus on areas enabling the integration of existing software products. In the computer implementation of new algorithms, we shall use the fact that many algorithms used for solving entirely different tasks are the same. We plan to achieve the project goals by applying an interdisciplinary approach, integrating research teams in the solution of specific scientific and engineering problems. As part of the research, the current methods will be extended and new approaches will be developed with the aim of achieving demonstrably new results of international significance and practical interest; emphasis will also be placed on finding solutions to selected highly demanding problems across the entire range of issues covered by the research programme. The project proposal builds on existing results achieved by the most successful teams participating in the research plan CEZ 272400019 ‘Development of algorithms for the solution of complex industrial problems’, and responds to new trends in computer modelling, taking full advantage of progress in IT, thus enabling the monitoring and further development of the most recent trends in basic engineering research.

Many of the results achieved by the above-mentioned research will be applied by the team of SC4Industry part of the project. In addition to standard software packages (whether commercial or non-commercial) that will be available at the Supercomputing Centre, it will also be possible to use newly developed software packages produced as an output of research activities in SC4Natural&BioSciences, which will apply the latest findings in the field. The most exciting aspect of this part of the project is the mutual interconnection of the Centre’s own self-developed software tools with standard commercial and non-commercial packages. Similar interconnections have already been achieved by the applicant workplace in the past, e.g. when using self-designed parallel solvers for ANSYS, Comsol or PMD finite element systems, or in the AnyBody system solving biomechanical simulations, using the OOSol library developed at the applicant workplace. Thanks to this interconnection, end users of standard software packages will be able to take advantage of state-of-the-art methods in solving their practical problems, without having to learn about supercomputing or the theory behind these methods.

The research team SC4Nanotechnologies is conceived as a compactly unified whole, consisting of laboratories for basic research of nanotechnologies, an in-house technology base for the preparation of nanostructures, measurement and diagnostic equipment, and an application block. The main element connecting all of these individual activities is the creation of mathematical-physical models both in basic and applied research, plus the application of supercomputing methods for the effective solution of these models.

Benefits of SC4Simulations

In addition to the enormous potential for the development of new and effective methods of solving physical problems or problems in the natural sciences, the main benefit of SC4Simulations should be seen in connection with the acquisition of the first genuine supercomputer in the Czech Republic. The lack of a supercomputer currently represents the most serious obstacle to the development of scientific disciplines that involve the solution of computationally demanding tasks. The benefits of the planned new Supercomputing Centre can be clearly seen in the following activities:

  1. Extensive research computations using a computer cluster; these are an essential part of the research to be carried out in the areas IT4People, EC4Innovations, SC4Simulations and Theory4IT. It will be necessary to dedicate part of the cluster (estimated about 128 nodes) to IT4People, with accessibility 365/24/7 – because the computation tasks provide near-real time information and are connected with safety and crisis management. SC4Simulations and Theory4IT will use a symmetrical multiprocessor system to solve tasks involving large volumes of data, e.g. data mining, flow simulation, etc.
  2. The solution of real tasks (in response to the demands of private companies, local/national government or other research institutions) whose high computation demands require large-scale investments in computer systems. In solving these tasks, widespread use will be made of commercial software, which will be complemented by the self-designed software products resulting from the Centre’s research. Both the computer cluster and the symmetrical multiprocessor computer will be used, though the second system will evidently come under greater pressure, as commercial software is generally poor at solving data-independent distributed computations. These tasks will be solved by the Centre in conjunction with external institutions. The main partners are expected to be industrial companies and research workplaces – both in the Czech Republic and abroad – dealing with similar issues.
  3. Provision of computing capacities for private and public institutions, plus consultancy on supercomputing. The high computation demands of some types of real tasks often mean that these tasks are practically impossible for institutions to solve, as a solution necessitates high-value investment in adequate computer capacities. In time, even these capacities may prove inadequate or over-valued. Additionally, the technologies may only be used occasionally, as dictated by practical necessity. Institutions can avoid this set of problems by renting out computing capacities from the Centre, which will set aside a certain capacity (estimated at 128 nodes of the cluster plus 32 processors of the symmetrical multiprocessor computer) for this purpose; alternatively, the Centre’s full capacity may be provided if planned in advance. It is expected that demand for these services will come mainly from industrial companies and other scientific institutions.