Tom Rozwadowski.JPG

RT Rozwadowski (1960)

Tom Rozwadowski was born in Poland on 10 January 1938. His ancestors, both on the paternal and maternal side, belonged to well known families of the Polish nobility, who served with distinction in various capacities in past centuries. Tom spent the war years in occupied Poland with his mother and elder brother, whilst his father fought with the Polish forces in Europe. Tom's mother, having moved to Warsaw in 1943 with her two sons, endured with them the insurrection against the Germans that ended in the destruction of 80% of the town and the mass deportation of its inhabitants.

The family was reunited in 1945 in West Germany. They moved to England in 1947 and eventually settled in South Africa in 1948. Tom was educated in various colleges in the Cape Provinces, where he matriculated at the Marist Brothers St Joseph College in Rondebosch. In 1956, following his vocation to become a Marist Brother, he was sent to their Noviciate in New South Wales, Australia. He left the religious life in 1959 and returned to South Africa in 1960.

He obtained his BSc at the University of Cape Town in 1961 and then joined Leo Computer Bureaux in Johannesburg, where he worked until 1965. He obtained his B Sc Honours (Mathematics) at the University of South Africa in 1965, and in the same year he joined Control Data (Pty) Limited in Johannesburg. After completing his M Sc thesis at Wits University, he left for the USA to conduct research in the computer field at CDC Headquarters in Minneapolis. Before his untimely death, he was already preparing his PhD degree.

Tom was an all rounder, his interest ranging from philosophy to computers to operations research. He spoke five languages fluently. At school, he excelled in sport, winning several athletics cups, and he was a keen rugby player. He enjoyed tennis up to the end of his life.

In the operations research field, he was a specialist in optimisation techniques, whilst also having a good knowledge of inventory control, game theory, simulation and dynamic programming. Some of the work he did at Leo included the design and supervision of a full-scale PERT suite; the writing of a series of programs embodying several linear programs to optimise order batching and paper cutting for a large paper mill; a system for controlling the production of glass work; and a study on the use of computers for real time control of traffic. As a computer man he was quite capable of writing the best part of twelve volumes of a tender all by himself in a month!

He was the driving force behind the establishment of an operations research group in Johannesburg and a member of the group which founded the operations research society of South Africa.

He would tackle anything that came his way and would not give it up until some results had been achieved. His interest on how to measure the performance of computers prompted him to produce a paper (1966) in which he examined the question from a macroscopic point of view. Thus he gave criteria to assess the performance of CPU, of I/O processors, of the software and the computer throughput. However, in 1970 he reopened the question, asking himself what a computer actually does; what instruments would one use to measure the work done by a computer; what mathematical model would one use to assess the performance from the microscopic point of view? Tom used the concepts of information theory, such as uncertainty and entropy, and gave the clues to the performance indicator for a computer.

The same pattern of development took place with his interest in gold mining. In 1963, he developed a computer program for JCI that performed all the required calculations to evaluate the present value of any fully specified mining policy. In addition, in 1966, he helped to develop a computer-based management game of a gold mine to try out the various possible types of decisions encountered in mining operations.

In 1970 he had produced a massive thesis in which, using advanced operations research techniques, such as the decomposition method in linear programming as well as the method of the Lagrangian multipliers and dynamic programming, he had developed a mathematical model to provide optimum solutions to the long term mining problem, taking into account such variables as taxation and uncertainty. His results were such that where no capital expenditure was involved, the present value was maximised and, where capital expenditure was involved, the present value was a maximum subject to an adequate return on capital. He did not consider the question closed, and was suggesting the use of simulation techniques to try out short-term tactics in mining operations, coupled with further and more sophisticated usage of linear and dynamic programming for the long-term planning.

On Monday, 12 October 1970, the news was circulated at Control Data South Africa that a telex had arrived during the weekend from the USA giving the announcement of Tom's death. Later the details became available. Tom and his family had gone to spend the weekend a few miles outside Minneapolis, apparently to give a talk on South Africa in a Catholic Church at Annadale, Minnesota. They had been put up for the night in a cottage as guests of the church. The gas central heating had a leaking pipe and the next morning Tom, his wife and their two children were all found asphyxiated.

All those who knew Tom will remember him for never refusing a helping hand to anybody in need of it. When Tom turned his head to listen to the problems put to him, there was always a twinkle in his eyes; there was the serene smile of a man who was at peace with himself, and with the rest of the world.