Grundy received his secondary education from
Malvern College, to which he had obtained a Major Scholarship in 1931, and from which he graduated in 1935. While there, he demonstrated his aptitude for mathematics by winning three prizes in that subject. After leaving school he entered
Clare College, Cambridge, on a Foundation Scholarship, where he read for the
Mathematical Tripos from 1936 to 1939, earning first class honours in Part II and a distinction in
Part III.
The work for which he is best known appeared in his first paper, Mathematics and Games, first published in the Cambridge University Mathematical Society's magazine, Eureka in 1939,[2] and reprinted by the same magazine in 1964.[3] The main results of this paper were discovered independently by Grundy and by
Roland Sprague, and had already been published by the latter in 1935.[4] The key idea is that of a function that assigns a
non-negative integer to each position of a class of combinatorial games, now called
impartial games, and which greatly assists in the identification of winning and losing positions, and of the winning moves from the former. The number assigned to a position by this function is called its Grundy value (or Grundy number), and the function itself is called the Sprague–Grundy function, in honour of its co-discoverers.[5] The procedures developed by Sprague and Grundy for using their function to analyse impartial games are collectively called Sprague–Grundy theory, and at least two different theorems concerning these procedures have been called Sprague–Grundy theorems.[6] The maximum number of
colors used by a
greedy coloring algorithm is called the
Grundy number, also after this work on games, as its definition has some formal similarities with the Sprague–Grundy theory.[7]
In 1939 Grundy began research in
algebraic geometry as a research student at the
University of Cambridge, eventually specialising in the
theory of ideals. In 1941 he won a
Smith's Prize for an essay entitled On the theory of
R-modules, and his first research paper in the area, A generalisation of additive ideal theory, was published in the following year.[8] In 1943 he was appointed to an assistant lectureship at the
University College of Hull, which he left in 1944. He was awarded a Ph.D. from the University of Cambridge in 1945.
Shortly after the end of World War II, Grundy moved away from the field of
algebra to take up work in
statistics. In 1947 he began formal training in the latter discipline at the
Rothamsted Experimental Station under a Ministry of Agriculture scholarship, graduating in 1949, when he then joined the permanent staff of the former organisation as an Experimental Officer. In 1951 he was promoted to Senior Experimental Officer. During his time at Rothamsted he performed most of his published statistical research, which included investigations of problems in the
design and analysis of experiments,
sampling, composition of animal populations, and fitting
truncated distributions.
From 1954 to 1958 Grundy worked as a statistician at the National Institute for Educational Research. During this period, he collaborated with
Michael Healy and D.H. Rees to extend
Frank Yates's work on
cost–benefit analysis of experimentation. The results of this collaboration were reported in an influential paper, Economic choice of the amount of experimentation, published in series B of the Journal of the Royal Statistical Society in 1956.[9] In 1958 Grundy moved to a position in the Biometry Unit at Oxford. However, he retired from this position after only one term, due to ill health.
Early in 1959 Grundy married Hilary Taylor, a former colleague from the National Institute of Educational Research. Although his health then greatly improved throughout 1959, he was unfortunately killed in an accident in November of that year.
List of Grundy's papers
With the exception of the final item, this list is taken from Smith's obituary (
1960). The first item is missing from Goddard's (
1960) list, which is otherwise the same as Smith's.
"A general technique for the analysis of experiments with incorrectly treated plots", Journal of the Royal Statistical Society, Series B, 13: 272–83, 1951
Grundy, P. M. (1951), "The expected frequencies in a sample of an animal population in which the abundances of species are log-normally distributed (Part I)", Biometrika, 38 (3–4): 427–34,
doi:
10.1093/biomet/38.3-4.427
"Selection without replacement from within strata with probability proportional to size", Journal of the Royal Statistical Society, Series B, 15: 253–61, 1953 (with
F. Yates)
"A method of sampling with probability exactly proportional to size", Journal of the Royal Statistical Society, Series B, 16: 236–8, 1954
"Economic choice of the amount of experimentation", Journal of the Royal Statistical Society, Series B, 18: 32–49, 1956 [11] (with D.H. Rees and
M.J.R. Healy)
"Fiducial distributions and prior distributions: an example in which the former cannot be associated with the latter", Journal of the Royal Statistical Society, Series B, 18: 217–21, 1956
^Except where otherwise indicated by alternative citations, the sources for the material in this article are the obituaries by Goddard (
1960) and Smith (
1960).
^Almost any comprehensive treatment of combinatorial game theory will cover Sprague and Grundy's results in some form. Examples are Berlekamp et al. (
1984), Conway (
1991), Siegel (
2013), and Smith (
2015).
^The theorem given that name by Smith (
2015,
p.340) is one which was in fact proved by Sprague and Grundy. The one given that name by Siegel (
2013,
478) and
by Wikipedia, however, relies on some later developments. Although it is an almost trivial consequence of Sprague and Grundy's results, which are also encapsulated in its statement and proof, it was not even formulated, let alone proved, by either of them.
^Erdős, Paul; Hedetniemi, Stephen T.; Laskar, Renu C.; Prins, Geert C. E. (2003), "On the equality of the partial Grundy and upper ochromatic numbers of graphs", Discrete Mathematics, 272 (1): 53–64,
doi:
10.1016/S0012-365X(03)00184-5,
MR2019200.