RESEARCH


ACL’s work is aimed at diversification of research process in Saint-Petersburg National Research University of Information Technologies, Mechanics and Optics (University ITMO) by solving certain scientific and technical problems and carrying out the research works according to international standards of modern science’s development and in order to promote and increase efficiency of academic and scientific activity of the students of Bachelor, Pre-Experience Master, Executive MBA and Doctoral programs.

The quintessence of ACL scientific development is a new generation of Supercomputer Technologies to support the decision making in critical situations. Technologically, this area is based on the paradigm of Urgent Computing. Currently, the concept of Urgent Computing (UC) has no unambiguous interpretation, and its interpretation depends on the research area of different groups of scientists. UC is traditionally used for solving problems which high computational complexity is combined with fixed terms (urgency) and criticality (importance, high priority) of the required outcomes. As a result, UC paradigm can be viewed as the process of high-performance computing of computer simulation for decision support in emergency situations given the limited time. The main objective of UC is to dynamically create and control the resource pool (data, software packages, computing power) in a distributed environment, which is sufficient to solve the problem of the user (in the form of the script specified by the user) within the specified time. Management of these resources is realized by low-level UC tools without user’s interference, and after the task is completed the resources are released automatically.

At present, despite the ideology of UC transparency and its practice-oriented focus, there is no common methodological approach to design and development of Early Warning Systems (EWS, Early Warning System), which implement the principles of UC. Apparently, it is due to the fact that in contrast to the traditional "big" problems supercomputer simulations, where the principal challenge is to map the computational algorithm for the supercomputer architecture, the UC objectives require consideration of a much wider range of factors. These tasks include the heterogeneity and variability of the computing environment along with the diversity of the algorithms and software packages for solving multidisciplinary problems (which is a characteristic feature of modern decision support systems). Therefore at the present moment there being developed highly specialized samples of EWS, focused on specific problems (often with a strong regional binding), for example, the prevention of forest fires, tornadoes, or the consequences of flooding.

In contrast, one of the main purposes of ACL scientific activity is to develop a UC methodology applicable in VLUC. VLUC is a hardware and software system which is in fact a problem-oriented source environment for design, development, testing and support of the EWS operation to prevent different classes of critical situations. To achieve this goal it is necessary to solve at least two problems – to develop VLUC hardware and software system and to demonstrate its efficiency and effectiveness in killer applications - real EWS in various subject areas. The criterion of success is a realistic demonstration of EWS, which represent in aggregate characteristics the new generation of killer models in relation to the models in use.

Thus, the complex nature of the ongoing studies of ACL is identified by ACL such scientific activities as emergency calculations, flood warning, epidemics and criminal threats analysis, etc.