Encyclopedia of Computer Science and Technology

498 von Neumann, JohnJohn von Neumann developed automata theory as well as fundamentalconcepts of computer architecture such as storing programsin memory along with the data. He also did seminal work in logic,quantum physics, simulation, and game theory. (SPL / PhotoResearchers, Inc.)would then serve as privatdozcent, or lecturer, at Berlin andthe University of Hamburg.During the mid-1920s, two competing mathematicaldescriptions of the behavior of atomic particles were beingoffered by Erwin Schrödinger’s wave equations and WernerHeisneberg’s matrix approach. Von Neumann showed thatthe two theories were mathematically equivalent. His 1932book, The Mathematical Foundations of Quantum Mechanics,remains a standard textbook to this day. Von Neumann alsodeveloped a new form of algebra where “rings of operators”could be used to describe the kind of dimensional spaceencountered in quantum mechanics.Meanwhile, von Neumann had become interested in themathematics of games, and developed the discipline thatwould later be called game theory. His “minimax theorem”described a class of two-person games in which both playerscould minimize their maximum risk by following aspecific strategy.Computation and Computer ArchitectureIn 1930, von Neumann immigrated to the United States,where he would become a naturalized citizen and spend therest of his career. He was made a Fellow at the new Institutefor Advanced Study at Princeton at its founding in 1933,and would serve in various capacities there and as a consultantfor the U.S. government.In the late 1930s, interest had begun to turn to the constructionof programmable calculators or computers (seeChurch, Alonzo and Turing, Alan). Just before and duringWorld War II, von Neumann worked on a variety ofproblems in ballistics, aerodynamics, and later, the designof nuclear weapons. All of these problems cried out formachine assistance, and von Neumann became acquaintedboth with British research in calculators and the massiveHarvard Mark I programmable calculator (see Aiken,Howard).A little later, von Neumann learned that two engineerswere working on a new kind of machine: an electronic digitalcomputer called ENIAC that used vacuum tubes for itsswitching and memory, making it about a thousand timesfaster than the Mark I. Although the first version of ENIAChad already been built by the time von Neumann came onboard, he served as a consultant to the project at the Universityof Pennsylvania’s Moore School.The earliest computers (such as the Mark I) read instructionsfrom cards or tape, discarding each instruction as itwas performed. This meant, for example, that to program aloop, an actual loop of tape would have to be mounted andcontrolled so that instructions could be repeated. The electronicENIAC was too fast for tape readers to keep up, so ithad to be programmed by setting thousands of switches tostore instructions and constant values. This tedious proceduremeant that it wasn’t practicable to use the machine foranything other than massive problems that would run formany days.In his 1945 “First Draft of a Report on the EDVAC” andhis more comprehensive 1946 “Preliminary Discussion ofthe Logical Design of an Electronic Computing Instrument,”von Neumann established the basic architecture and designprinciples of the modern electronic digital computer.Von Neumann declared that in future computers themachine’s internal memory would be used to store constantdata and all instructions. With programs in memory, loopingor other decision making can be accomplished simplyby “jumping” from one memory location to another. Computerswould have two forms of memory: relatively fastmemory for holding instructions, and a slower form of storagethat could hold large amounts of data and the results ofprocessing. (In today’s PCs these functions are provided bythe random access memory [RAM] and hard drive respectively.)The storage of programs in memory also meant thata program could treat its own instructions like data andchange them in response to changing conditions.In general, von Neumann took the hybrid design ofENIAC and conceived of a design that would be all-electronicin its internal operations and store data in the mostnatural form possible for an electronic machine—binary,with 1 and 0 representing the on and off switching statesand, in memory, two possible “marks” indicated by magnetism,voltage levels, or some other phenomenon. The logicaldesign would be consistent and largely independent of thevagaries of hardware.