Space

Keldysh Institute of Applied Mathematics

Keldysh Research Centre

8/10 Onezhskaya Street Moscow 125438 Russia
Tel

+7 095 456-46-08
+7 095 456-87-56

Fax

+7 095 456-82-28

Telex

412761 NIITP SU

E-mail

niitp@atis.kiae.su

Mathematical modelling became an effective means of theoretical research of processes and objects in the middle of the last century. It has been used mainly to determine the basic quantitative characteristics of phenomena that cannot be experimentally replicated in a natural or laboratory environment. An example is the task of defining the orbits of rocket and space technology products and the power of a thermonuclear device. Over time, mathematical modelling was used to interpret experiments and optimize the process of developing complex technical objects. In this case, the limited resources of available computers required the use of approximations About the size of the tasks and the set of processes to be considered.

This limited not only the effectiveness of mathematical modelling, but also the possibility of experimental testing of its methods. The consequence was a well-founded distrust of technical specialists and experimenters to the results of mathematical modelling. Supercomputer modeling is fundamentally different from the approximate analogues. The quantitative increase in computer resources has resulted in a new quality. The creation of parallel supercomputers allowed to base models on classical, centuries-tested three-dimensional equations of mathematical physics, expressing fundamental Conservation laws, avoiding approximations of the dimension of the task and the set of effects to be taken into account. Technical objects, not the simplest geometric figures.

The main advantage of supercomputer modeling, distinguishing it from approximate theoretical estimations, is the following. Supercomputer mathematical model is able to reflect not only the natural conditions in which the investigated processes are developed or functioning objects, but also the conditions of reproducing their laboratory experiments. Thus the supercomputer models are reliably podtverzhdaemy and can be used for full predictive calculations. In particular, in the absence of natural tests in the reduction of nuclear weapons only by means of mathematical modelling it is possible to reproduce the nuclear explosion and its impact.

Comparison of levels of development of supercomputer modelling in Russia and abroad gives a contradictory picture. Formally, Russian codes can solve the same tasks as foreign ones. In this case, the quality of the code software in large calculations is not confirmed. Intensive computer codes implementing fundamental mathematical models in Russia are developed in the framework of research, not experimental-design works. The issue of industrial development is not even put in most cases. As a result, advanced scientific ideas in the field of mathematical modelling, numerical methods and technologies of high-performance computation, developed in the framework of research works, do not get sufficiently perfect software Implementation and, as a result, are not confirmed. In this case, 3d modeling requires a disproportionately high cost of resources to compile and debug programs, both compute and interface. More important is the design and documentation of the software. This is not the subject of scientific research.

Result in the level of mathematical models determined by the completeness of the described physical picture and the correctness of the equations used for this mathematical physics, as well as the efficiency of the algorithms used to solve the equations of the model, Russian codes are not inferior to foreign ones. According to the quality of software implementation, determined by the degree of debuggability of the software, and the completeness of the experimental confirmation Russian codes are much inferior to the western.

The supercompute resources in Russia and the low level of the software of Russian supercomputer codes led to the fact that the Russian industry uses for the development of technical facilities foreign Software, mainly packages ABAQUS, ANSYS and NASTRAN. Jetipakety are used by defense companies. This situation is unacceptable. The problem of providing Russian industry, first of all by defense, by means of predictive supercomputer modelling should be solved in the nearest years. When planning a solution to this problem, it should be reported that parity in the development of supercomputers in the near future is not achievable. It is also impossible to allocate sufficient funds for bringing Russian scientific codes to the industrial level in the coming years.

The only possible breakthrough is the quality of mathematical models and the efficiency of numerical algorithms. It is necessary to develop a set of scientific supercomputer codes on the basis of achievements of institutions subordinated to Russian Federal State Budget Institution "Rossijskajaakademijanauk" (hereinafter – RAS), State Corporation for Atomic energy "Rosatom" and the Russian MOD, to create a supercomputer center on the basis of a new supercomputer with a performance of at least 10 pflop/s on the basis of the federal State Institution "Federal Research Center Institute Applied mathematics named after M.V. Keldysh of the Russian Academy of Sciences "(IPM M. Keldysh RAS) and organize the development of supercomputer technologies together with the enterprises of industry.

At the same time, the lack of performance of supercomputer will be compensated by the quality of models, efficiency of algorithms and expert experience of representatives of the industry planning calculation, and insufficient quality of debugging of programs – participation in calculation of their developers. Implementation of the breakthrough in the creation of qualitative mathematic models and effective numerical algorithms, debugging of scientific supercomputer codes and creation on this basis of the supercomputer center of collective use is a mission of IPM imeni M.V. Keldysh for the period up to 2020. The mission of the project is to create within the IPM. M.V. Keldysh ran a unified research and administrative infrastructure to provide breakthrough scientific developments on promising areas of applied mathematics, mathematical modelling, theoretical programming, Applied Mechanics, Bioinformatics, and biomedicine on the basis of modern computational resources.

Institute of Applied Mathematics was founded in 1953 to solve complex mathematical problems connected with state programs of space exploration, development of nuclear and thermonuclear energy, on the basis of creation and Extensive use of computer hardware and software. Its organizer and director (1953 – 1978) was President of the USSR Academy of Sciences Mstislav Vsevolodovich Keldysh.

From the first years the activity of the Institute focused on the decision of large applied tasks is based on fundamental scientific researches in the field of mathematics, mechanics, cybernetics, informatics, which are conducted by its employees. 3 Academicians, 4 correspondent members of the Russian Academy of Sciences, 91 doctor and 155 candidates of sciences work in IPM. Among them are 14 laureates of the Lenin Prize, 30 laureates of the State Prize, 5 laureates of the USSR Council of Ministers prize.

The ballistic Center organized by the Institute, starting with the launch of the first artificial Earth satellite, successfully solves problems of ballistic-navigational support of flights of manned ships, long-term orbital stations "salute " and " World ", reusable space system " Energy-buran ", automatic devices of scientific purpose " Luna ", " Venus ", " Mars "etc., participates in the development and implementation of international space projects. The world recognition has received works on algorithmic and software for vehicles of new type-stepping robots, automation of manual operations at assembling of products of mechanical engineering with the help of adaptive robots.

The Institute carried out calculations of unique in complexity and scope of tasks of dynamics explosion, protection against penetrating radiation, supersonic of aircraft flow, detailed neutron-physical calculation of nuclear reactor. In the early 1960s, long before such calculations in the United States, numerical experiments were carried out on the computer, which opened a new area of applied mathematics-computational electrodynamics.

The Institute is the ancestor of the use of electronic computer technology in the Soviet Union. Here was installed the first serial domestic computer and organized the first in the country structural unit, honoring pioneering work on the creation of software. In 1963, the translator Algol-60, one of the world's first translators with the full version of the language, was implemented, the IP-2 system was the first mini-os. The line of development of computer system maintenance in the future has received development in full-scale operating systems Dispak and OS IPM, optimizirujushhem the compiler from language Fortran, complex of programs grafor for graphic output of information, in works on Architecture and software of supercomputers, multimachine complexes and networks. An important place in the activities of the Institute is the work on system maintenance of Applied tasks: System of programming and debugging of equipment of reusable ship "Buran ", Real-time object management system, information-search Systems for various areas, systems of automation of design and production (CAD/CAM), instrumental systems of building of packages of application programs.

A lot of attention is paid to the Institute of qualified scientific personnel training. The Institute is a basic organization for universities such as MSU, MiFi and Mai. Here works the Department of Applied Mathematics MiPT.

Institute of Applied Mathematics. M.V. Keldysh is defined by the head organization on a number of leading directions of applied mathematics. The Institute's work has been widely recognized in our country and abroad. From it at various times Seymour and became leading in the directions such organizations of the Russian Academy of Sciences, as Computer Center, Institute of Space Researches, Institute of Mathematical Modelling.