Encyclopedia of Computer Science and Technology

296 Mauchly, John WilliamMoving from simple formulas to the manipulation of symbolicquantities (as in algebra), the Association for ComputingMachinery (ACM) classification system describes severalbroad areas of computer-aided mathematics. These includenumerical analysis (techniques for solving, linear, non-linear,and differential equations), discrete mathematics (combinatorialand graph theory), and probability and statistics.There are two general approaches to mathematicalsoftware. One is the creation of libraries of routines orprocedures that address particular kinds of problems. Aprogrammer who is creating software that must deal withparticular mathematical problems can link these routinesto the program, call the procedures with appropriate variablesor data, and return the results to the main programfor further processing (see procedures and functions).The language FORTRAN is still widely used for developingmathematics libraries, and there is a legacy of tens of thousandsof routines available. Modern systems have the abilityto link these procedures to programs written in more recentlanguages such as C.The advantage of using program libraries is that theydon’t require learning new programming techniques. Eachroutine can be treated as a “black box.” However, it is oftendesirable to work with traditional mathematical notation(what one might see on a blackboard in a calculus class,rather than typed into computer code). A stand-alone softwarepackage such as Mathcad, Matlab, or Mathematicacan automatically simplify or solve algebraic expressions orperform hundreds of traditional mathematical procedures.For statistical analysis, programs such as SPSS can apply allof the standard statistical tests to data and provide a largevariety of graphics.Further ReadingField, Andy. Discovering Statistics Using SPSS. 2nd ed. ThousandOaks, Calif.: SAGE Publications, 2005.Griffith, Arthur. SPSS for Dummies. Hoboken, N.J.: Wiley, 2007.Netlib Repository of Mathematical Software, Papers, and Databases.Available online. URL: http://www.netlib.org/. AccessedAugust 14, 2007.Press, William H., et al. Numerical Recipes. 3rd ed. New York:Cambridge University Press, 2007.Ruskeepaa, Heikki. Mathematica Navigator: Mathematics, Statistics,and Graphics. Burlington, Mass.: Elsevier AcademicPress, 2004.Wellin, Paul, Richard Gaylord, and Samuel Kamin. An Introductionto Programming with Mathematica. New York: CambridgeUniversity Press, 2005.Wolfram, Stephen. The Mathematica Book. 5th ed. Champaign, Ill.:Wolfram Media, 2003.Wolfram Mathematica Home Page. Available online. URL: http://www.wolfram.com/. Accessed August 14, 2007.Mauchly, John William(1907–1980)AmericanInventor, Computer ScientistJohn Mauchly was codesigner of the earliest full-scale digitalcomputer, ENIAC, and its first commercial successor,Univac (see also Eckert, J. Presper). His and Eckert’s workwent a long way toward establishing the viability of thecomputer industry in the early 1950s.Mauchly was born on August 30, 1907, in Cincinnati,Ohio. He attended the McKinley Technical High School inWashington, D.C., and then began his college studies atJohns Hopkins University, eventually changing his majorfrom engineering to physics. The spectral analysis problemshe tackled for his Ph.D. (awarded in 1932) and inpostgraduate work required a large amount of painstakingcalculation. So, too, did his later interest in weather prediction,which led him to design a mechanical computerfor harmonic analysis of weather data (see analog computer).He also learned about binary switching circuits(“flip-flops”) and experimented with building electroniccounters, which used vacuum tubes and were much fasterthan counters using electromagnetic relays.Mauchly taught physics at Ursinus College in Philadelphiafrom 1933 to 1941. On the eve of World War II, however,he went to the University of Pennsylvania’s MooreSchool of Engineering and took a course in military applicationsof electronics. He then joined the staff and beganworking on contracts to prepare artillery firing tables forthe military. Realizing how intensive the calculationswould be, in 1942 he wrote a memo proposing that anelectronic calculator be built to tackle the problem. Theproposal was rejected at first, but by 1943 table calculationby mechanical methods was falling even further behind.Herman Goldstine, who had been assigned by the AberdeenProving Ground to break the bottleneck, approved the calculatorproject.With Mauchly providing theoretical design work andJ. Presper Eckert heading the engineering effort, the ElectronicNumerical Integrator and Computer, better known asENIAC, was completed too late to influence the outcome ofthe war. However, when the machine was demonstrated inFebruary, 1946, it showed that a programmable electroniccomputer was not only about a thousand times faster thanan electromechanical calculator, it could be used as a general-purposeproblem-solver that could do much more thanexisting calculators.Mauchly and Eckert left the Moore School after a disputeabout who owned the patent for the computer work.They jointly founded what became known as the Eckert-Mauchly Computer Corporation, betting on Mauchly’s confidencethat there was sufficient demand for computers notonly for scientific or military use, but for business applicationsas well. By 1950, however, they were struggling to selland build their improved computer, Univac, while fulfillingexisting government contracts for a scaled-down versioncalled BINAC. In 1950, they sold their company to RemingtonRand, while continuing to work on Univac. In 1952,Univac stunned the world by correctly predicting the presidentialelection results on election night long before most ofthe votes had come in.Early on, Mauchly saw the need for a better way to writecomputer programs. Univac and other early computers hadbeen programmed through a mixture of rewiring, setting ofswitches, and entering numbers into registers. This made