Academician A.V.Kalyaev — his life and work
Anatoly Vasilevich Kalyaev was born on the 29-th of June 1922 in town Rtishchevo near Saratov (on Volga).
In 1940, at the age of 18, he entered Leningrad Polytechnic Institute (university). However, with the beginning of the Great Patriotic War (1941) he joined the 42-nd army as volunteer and during all three years of the siege of Leningrad he was on front line defending the town.
After the war Kalyaev returned to Polytechnic Institute to continue his learning. Still being a student he took part in research work under guidance of famous scientists — prof. Pavel Lazarevich Kalantarov and (later an academician) and Leonid Robertovich Neuman.
Kalyaev became interested in theory of electric circuits, theory of electromagnetic field, electrodynamics and problems of non-linear theory of electricity. At the same time, his attention was attracted by the beginning development of computing systems, including electromechanical ones, which were studied then at the Polytechnic Institute.
After receiving his diploma he stayed at the institute as a post-graduate and soon successfully defended dissertation thesis for «Candidate of sciences» degree. In 1954, according to the ministry's of higher education decision, he was given a position at the Institute of Radioengineering in town of Taganrog (South-West Russia - http://en.wikipedia.org/wiki/Taganrog), where he remained and worked more than 50 years.
Scientific and methodical/pedagogic level of the chair, headed at the institute by professor G.E. Pukhov, was very high. That made favorable conditions for -then beginning- researches in the field of analog computation systems. Kalyaev took most active part in them.
Since the first half of the 1950-s scientific works, aimed at creation of the first digital differential analyzers, were in progress in some western countries. Those computing devices were designed to solve differential equations of higher degree, some differential equations in partial derivatives, complex functional dependencies, modeling of complex dynamic systems, etc.
All that was connected with intensive development of rocket and space engineering, automatic control systems, nuclear power production, etc.
Same as in the West, the USSR science and industry were increasingly engaged in development of those problem oriented computing machines — digital differential analyzers, that integrated characteristic features of both digital and analog computers.
Than the Soviet researches on the digital differential analyzers were just beginning, and that made it easy for Kalyaev to quickly skip from purely analog machines to digital computers.
In the end of 1959, Kalyaev's department received an order from the Scientific Research Institute №3, of the USSR Ministry of Defense, for development of -the first soviet- digital differential analyzer for ballistic rocket trajectory calculations. The «Institute-3» itself was engaged in researches on rocket guidance systems, trajectory calculations, etc.
Following the order they designed, assembled and tested experimental computer-analyzer «Meteor-1» within less than two years.
Although generally it functioned well, rather slow performance of its -very simple- system of one place binary increments became the weak point, thus making the whole work less satisfactory.
For that reason Kalyaev himself developed theory of a new class of digital integrating computers able to process floating point numbers and multi-place increments of variables. Such computers could have both single- and parallel multiprocessor architecture.
In the first half of the 1960ies Kalyaev defined and implemented his idea of multiprocessor digital integrators with parallel architecture, e.g. in such as «Meteor-3» with 100 parallel processors and with integral performance - more than 3000000 ops, digital integrator «Meteor-4» with 40 parallel processors and performance of 4200000 ops., integrator «Omega» and a number of other similar devices.
The «Omega» was handed over to the Acoustic Institute of the USSR Academy of Sciences, where it solved systems of algebraic and differential equations, calculated complex functions, integrals and modeled dynamic objects and systems in real time mode.
In 1964, publishing house «Naukova Dumka» in Kiev printed Kalyaev's first monography «Introduction into Digital Integrators Theory». That was first detailed presentation of his theory of digital integrators, which formed scientific basis for the following development of the complex theory of digital integrating computers. That also became the most popular textbook among designers of such machines.
A few years later the USSR Ministry of Higher Education appointed Kalyaev as rector of the Taganrog Radioengineering Institute. As the scientific and educational level of the institute was surprisingly high the ministerial administration was going to further develop that powerful scientific centre of the South Russia. Kalyaev with his impressive organizing skills was the best person for such mission. He stayed in that position for more than 18 years.
In 1961 Kalyaev established Special design bureau «Mius» (modeling and control systems) at the institute. It was based on scientific departments of microelectronics and of Digital Integrators. Under his scientific guidance «Mius» performed many researches, designed special problem oriented computers and on-board computing devices. Their components were developed there as well.
In the beginning of the 1970-s a new group of microelectronic integrators was designed. That was development of his earlier ideas which partly were realized in his first integrators «Meteor», «Omega», etc. Among the new ones there was digital integrator «Don» with 50 parallel processors (assembled in 1970 for technical university AZNEFTEKHIM of Baku (Azerbaijan)), multiprocessor integrator «Taganrog» (1971) with 50 processors and performance up to 2 mil. Ops. The later was produced for a design bureau in Kharkov (Ukraine); many more devices were also created.
In 1970 he published one more monograph «Theory of digital integrating machines and structures». It presented his complete theory of different classes of digital integrators, including the theory of extrapolational and interpolational integrators, parallel and sequential integrators, processing single and multiplace increments with fixed and floating point. There he studied numeric methods for Riemann-Stieltjes integrals and obtained precise formulae for numeric solving of Stieltjes integrals.
Kalyaev proved that those formulae for Stieltjes integrals, together with multiplace increments, floating points, with respect to quantization errors and development of parallel architecture in combination with electronic dynamic commutation, several times increase performance of the integrators without extension of hardware or enlargement of their size.
Already in that monograph he set forth the idea about designing uniform digital integrating structures based on -then just appearing- integrated microcircuitry. The progress of electronic microcircuitry opened way for creation of monocrystal digital integrating microprocessors and monocrystal multichannel switching devices, which enabled programming of communication channels that connected parallel digital integrating microprocessors.
His ideas became the starting point for practical development of uniform multiprocessor computing systems with programmable switching.
At the same time, the Taganrog Radio Institute made efforts to provide conditions for the further progress of microelectronics and computing. For that purpose it needed forming of appropriate scientific basis both for fundamental theoretical and experimental problem oriented researches and for practical applied development. Therefore, Kalyaev proposed to organize a new scientific research institute.
On the 29-th of December 1972 the Council of Ministers of the Russian Federation adopted decision to establish Scientific Research Institute of uniform microelectronic computing structures (SRI UMCS). Kalyaev was appointed as its director and held that position more than twenty years.
Later the institute was renamed into the Institute of Microprocessor Computation Systems. The main fields of its scientific activity were the following:
- development of theory and principles for creation of the multiprocessor computing systems with programmable architecture, intended for digital control and modeling;
- software development for multiprocessor digital systems;
- development of theory and principles for design of VLCI with programmed structure for multiprocessor computing systems;
- development of the theory and principles for designing adaptive neuronal computing and robotic systems.
More then 450 people worked at the institute, including dozens of doctors and candidates of sciences, and more then 250 engineers and technicians. About ten highly experienced leading scientists headed departments, its laboratories and research teams.
Alongside with his theoretical and experimental researches Kalyaev organized -and completed- numerous applied scientific-engineering and experimental design projects. As the result, about twenty universal and problem oriented computing systems were produced, and nearly half of them were the high performance massively parallel systems with programmable architecture.
In 1978 Kalyaev published his monograph «Uniform switching register oriented structures»,
which initiated researches in one more scientific field. In the book he described his theory of uniform switching structures developed for synthesis and programming of numerous, time-varying
communication channels, linking big number of parallel processors. The book also presented the world's first analysis of automata with programmable structures and switching. They subsequently formed the basis of multiprocessor computing systems with programmed architecture. A special chapter of the monograph was dedicated to the problems of switching in neuronal computing structures, architecture of 2D and 3D neuronal structures, and the methods of their adjustment and re-designing.
In 1984 Kalyaev was elected at the USSR Academy of Sciences meeting as its corresponding member.
His next monograph, «Multiprocessor Systems with Programmable Architecture» (1984) also played significant role in scientific progress of his institute.
In that work he set forth the concept and developed -resulting from it- theory of superhigh-performance multiprocessor computing systems with mass-parallelism, programmable architecture and structured organization of computations.
In those years they were unique systems with nothing similar in the world. They provided nearly peak performance at practically all tasks and their performance increased linearly according to the number of working processors. Their design essentially simplified efficient programming and organization of parallel computing. They were highly reliable with high liveness; their modular and extendible structure was well suited for efficient implementation of most modern microelectronics.
Kalyaev's idea of monocrystal microprocessors with programmable structure based on sets of enlarged operations and internal machine languages of high level, served as basic idea for multiprocessor computing systems. The theory of multichannel programmable internal structure was further developed by him and his pupils. Number of their publications appeared in the soviet and foreign scientific literature and periodicals.
In1986 Kalyaev's outstanding scientific contribution was awarded an honorable title «Hero of the Socialist Labour».
According to the USSR governmental directive, within the period from 1991 to 1993 the Taganrog institute was engaged as the leading organization in the scientific and engineering project «Super-macro-neuronal computer».
The project of universal multiprocessor computing system with programmable architecture was completed in 1993. For that system the institute designed special set of components, which included supertransputer, micromemory and microswitching system. All was based on one-micron microcircuit technology.
Simultaneously with the research and experimental design of multiprocessor parallel systems Kalyaev also performed researches on digital neuronal processor networks and neuronal computers with massive-parallelism and programmable structure.
Thus, in 1977 an experimental digital neuronal ensemble was designed, manufactured and fully tested. It had ten parallel neuronal processors able to recombine their structures adjusting them to realization of formal-logic functions and for dynamic modeling of neurons. Besides, it possessed flexible switching system enabling programming of its architecture.
In 1987 a pilot universal modeling complex was produced for neurocybernetic research. It consisted of a personal computer, neuronal processor with programmable structure to be used as accelerating co-processor and of a software set that supported modeling of various neuro-processor complexes and networks with number of neuronal components up to 32x104 and number of neuron synaptic links up to 2.5x106. That complex enabled modeling of numerous neuro-paradigms and had performance up to 106 CUPS (commutations of internodal links per second).
In 1989 the world's first LSI - Large Scale Integrated-circuit, on basic matrix chrystal, was designed for digital neuroprocessor with programmable structure. Now the neuroprocessor structure could be programmed to realize various types of neurons, among them dynamic, adaptive at in- and outputs, formal logical, etc.
In 1990 an experimental microassembly of digital neuroprocessor ensemble was designed and produced. In its one microassembly unit there was 6 unpacked LSIs of digital neuroprocessors and programmable switch, what permitted programming of various types of neuroprocessor ensembles in the microassembly.
That microassembly was displayed at several international exhibitions and was given an award.
In 1992 the institute designed the world's first parallel multi-neuroprocessor digital neurocomputer with programmable architecture. It was performed as combined neurocomputer system with personal computer PC AT and parallel accelerating co-processor, that consisted of 15 parallel digital neuroprocessors with programmable structure and programmable switching system.
In 1990 Kalyaev and his colleagues published monograph «Uniform Control Structures of Adoptive Robots», where they analyzed problems of design and manufacturing of neuronal based control systems for adaptive robots of the third generation with elements of artificial intelligence. The robots were moving in dynamically changing environment with numerous obstacles. The authors presented characteristics of produced robots and described algorithms for planning of behavior of both single robots and their groups.
In 1990 the institute produced experimental set of VLSI circuitry with uniform neuronal structure for planning of trajectory of robot movements in complex environment with obstacles. One VLSI integrated 128 parallel primitive/basic neuronal processors.
In 1992 an on-board multiprocessor computing system with increased survivability was designed for control of three-axis stabilized platform «Argus». It was a part of space research program «Mars-94». The installation with platform should be mounted on the space station studying the surface of Mars. The first functioning equipment for the system was manufactured in 1993-1994. It was highly prized by the specialists of the space research center.
In 1997-2003 main efforts of Kalyaev and his scientific school were concentrated on creation of hard- and software means of support for programming systems, for adjustment and re-configuration of the architecture of multiprocessor computing systems with massive parallelism. They also solved problems of reaching extra-high performance in those systems and of achieving linear character of the performance growth at the level of the best samples of foreign supercomputers and to bypass the difficulties connected to lag of domestic microelectronic technology in case of creation of the super computer.
At the same time scientists of the institute continued researches and development of system software for multiprocessor computing systems with massive parallelism and programmable architecture. They also developed technologies to make possible translation of the existing sequential programs into parallel ones. Number of interesting solutions of many scientific problems have been obtained during those fundamental researches.
In May 2000 at the General Assembly of the Academy of Sciences of the Russian Federation Kalyaev was elected its academician.
In 2003 Kalyaev's scientific team participated in joint Russian and Byelorussian program «Development, serial production and implementation of the family of supercomputers -- high-performance computing systems with parallel architecture, and the following creation of hard-and software complexes on their basis». In course of this program a new minisupercopmputer was developed. It was based on modular extendible multiprocessor system with programmable architecture and structural-procedure-oriented organization of calculations. Appropriate software was also developed.
In 2003 Kalyaev and his colleagues published two monographs: «Modular extendible multiprocessor systems with structural procedure-oriented organization of calculations» and «Performance evaluation in multiprocessor computing systems with massive parallelism».
During the 50 years of his scientific and pedagogical career Kalyaev -both himself and with collaborators- completed over 200 projects. He was a scientific tutor and supervisor of 18 scientists for their doctor's dissertations (Doctor of Technical sciences degree) and of 59 candidates of sciences. He had 380 scientific publications — among them 13 monographs, and also 69 patents for his inventions.
Kalyaev's scientific merits were awarded with the «Lenin order» (the highest civil award), with three orders «Red Banner of Labour», also order of «Patriotic War», order of «People's Friendship» and 16 medals.
Academician Anatoly Kalyaev died on the 10-th of March 2004 in town of Taganrog and is buried at the city cemetery.
Published in museum, 24.09.2013