Chapter 7: Mechanisms in Programs and Nature

Section 8: The Problem of Satisfying Constraints

History [of combinatorial optimization]

Work on combinatorial optimization started in earnest in the late 1950s, but by the time NP completeness was discovered in 1971 (see page 1143) it had become clear that finding exact solutions would be very difficult. Approximate methods tended to be constructed for specific problems. But in the early 1980s, simulated annealing was suggested by Scott Kirkpatrick and others as one of the first potentially general approaches. And starting in the mid-1980s, extensive work was done on biologically motivated so-called genetic algorithms, which had been advocated by John Holland since the 1960s. Progress in combinatorial optimization is however often difficult to recognize, because there are almost no general results, and results that are quoted are often sensitive to details of the problems studied and the computer implementations used.

From Stephen Wolfram: A New Kind of Science [citation]