Born in Cluj, Romania, in 1922 to a Jewish family of Hungarian origin, Egon Balas became involved in underground activities during the Second World War. He was captured and tortured by the Germans, but managed to escape. After the war, he was appointed to a diplomatic post in London and held positions of responsibility in the emerging communist Romania. But he was imprisoned once again for holding unorthodox views, repeatedly interrogated by the dreaded Securitate, held in solitary confinement for more than two years, then released and expelled from the Communist Party. He did not begin his career as a mathematician until the age of 37. This fascinating odyssey is recounted in his autobiography, La Liberté et rien d’autre, published in French by L’Harmattan in 2003.
After his release from prison, Egon Balas was assigned to the Water and Forestry Institute in Bucharest, where Romania's logging operations were planned. To develop suitable logistical tools, he had to teach himself mathematics and operations research from whatever books he could obtain. Peter Hammer (1936–2006), who would also become very well known in operations research, worked at the institute during the same period. To plan timber transport, Balas and Hammer created new tools based on network flow theory and linear programming (Hammer was then publishing under the name Ivanescu).
A pioneer of integer optimization ----------------------------------------------------
In 1962, Egon Balas faced a complex problem. An entire network of access roads had to be built in one area of the forest to reach remote plots. The task was to decide which plots to log and which access roads to build. These decisions were closely interconnected, giving rise to logical implications: if section A of a road is built, section B must also be built so that A can be reached. Egon Balas formulated the problem as a linear program with binary (0, 1) variables. For example, if xA = 1 represents the construction of section A and xB = 1 represents the construction of section B, the constraint xAxB, for variables xA, xB taking the values 0 or 1, represents the logical implication described above. Similarly, to express the condition "at least one of sections A, B and C must be built" as a linear constraint in 0–1 variables, we would write xA + xB + xC ≥ 1.