History of Mathematical Programming Systems

The systematic development of practical computing methods for linear programming (LP) began in 1952 at the Rand Corporation in Santa Monica, under the direction of George B. Dantzig. The author worked intensively on this project there until late 1956, by which time great progress had been made on first-generation computers. The work continued at CEIR, Inc., in Washington for some years and later in many places by many individuals and firms. By the late 1960s, elaborate systems of programs known as mathematical programming systems (MPS) had become a standard part of the available software for a number of computers, notably the IBM 360, GE 635, CDC 6600, and Univac 1108. The major MPSs underwent significant updating and extension during the mid-1970s taking on their present and probably final form, at least for big mainframes. Work still continues, however, and substantial improvements are being made in speed, reliability, supporting data management and control systems, and application techniques. Development of quite powerful microcomputer systems is now underway.