George Marsaglia

From Wikipedia, the free encyclopedia
Jump to: navigation, search
George Marsaglia
Born (1924-03-12)March 12, 1924
Died February 15, 2011(2011-02-15) (aged 86)
Tallahassee, Florida
Nationality American
Fields Mathematics
Institutions Florida State University
Washington State University
Alma mater Ohio State University
Doctoral advisor Henry Mann

George Marsaglia (March 12, 1924 – February 15, 2011)1 was an American mathematician and computer scientist. He established the lattice structure of linear congruential generators in the paper "Random numbers fall mainly in the planes".2 This phenomenon is sometimes called the Marsaglia effect, and means that n-tuples with coordinates obtained from consecutive use of the generator will lie on a small number of equally spaced hyperplanes in n-dimensional space.3 He also developed the so-called "diehard tests", a series of tests to determine whether or not a sequence of numbers have the statistical properties that could be expected from a random sequence. In 1995 he published a CD-ROM of random numbers which included the diehard tests.4

He is also known for developing some of the most commonly used methods for generating random numbers and using them to produce random samples from various distributions. Some of the most widely used being the multiply-with-carry, subtract-with-borrow, Xorshift, KISS and Mother methods for random numbers, and the ziggurat algorithm for generating normally or other unimodally distributed random variables.

He was Professor Emeritus of Pure and Applied Mathematics and Computer Science at Washington State University and Professor Emeritus of Statistics at Florida State University.

Marsaglia died of a heart attack on February 15, 2011, in Tallahassee.

Family

Marsaglia had one son, John, with his first wife, Lee Ann Marsaglia. Until his death he was married to Doris Marsaglia. He had two grandchildren, Chris and Nicole Marsaglia, through their son John and his wife Michelle.

See also

References








Creative Commons License