Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Brouwer 1975 note number unique subgraphs graph j
main_theorem: Brouwer's theorem that the largest number f(n) of unique subgraphs of a graph on n vertices satisfies log_2 f(n) = n^2/2 - n log_2 n + O(n), the upper bound from the count of unlabelled graphs and the lower bound from an explicit construction.
Brouwer, A. E., Note: ``On the number of unique subgraphs of a graph'' (J. Combinatorial Theory Ser. B {\bf 13} (1972), 112-115) by R. C. Entringer and P. Erdős. J. Combinatorial Theory Ser. B {\bf 18} (1975), 184-185. DOI 10.1016/0095-8956(75)90047-7.
The copy read for this card is the Mathematisch Centrum report ZN 56/73 (December 1973) rather than the two-page JCTB note of 1975: the same author and the same result, with pages numbered 1 and 2; the page references below are the report's. As Brouwer reports (p. 1), Entringer and Erdős call a subgraph H of a graph G "unique if H is not isomorphic to any other subgraph of G", and they proved that f(n), the largest number of unique subgraphs of a graph on n vertices, satisfies f(n) > 2^{n^2/2 - cn^{3/2}} for c > 3/sqrt(2) and n sufficiently large. The abstract says the note answers a question of Entringer and Erdős concerning the number of unique subgraphs, without restating it. The main theorem (p. 1, unnumbered) determines log_2 f(n) up to O(n): log_2 f(n) = n^2/2 - n log_2 n + O(n). The upper bound follows from the number of nonisomorphic graphs on n vertices, which the note quotes from Harary and Palmer as 2^{n(n-1)/2}/n! times a factor tending to 1 (p. 1). The lower bound is a construction (pp. 1--2): with m = ceil(log_2 n) and N = n - m - 2, a complete graph on N vertices, a rigid tree on m vertices and one pendant edge are joined so that every graph between the construction and its copy with the N-clique emptied is a unique subgraph, giving 2^{N(N-1)/2} = 2^{n^2/2 - n log_2 n + O(n)} unique subgraphs. No notice is printed in the copy read (two title pages, the colophon page and the two text pages), and the repository's record for the report (https://ir.cwi.nl/pub/7446) states no copyright or license term; the term is unstated.
Read status: claims checked for the main theorem, its definitions and the construction, read clause by clause on the page images of pp. 1--2; the uniqueness argument was followed. The enumeration formula is cited, not proved. Nothing here is independently reviewed. Result page: main_theorem.
Source: https://ir.cwi.nl/pub/7446.
Bears on. #426: the problem asks whether some graph on n vertices has >> 2^{n(n-1)/2}/n! unique subgraphs. The main theorem (p. 1) puts log_2 f(n) within O(n) of the logarithm of that quantity; an O(n) error in the exponent hides factors 2^{O(n)} either way, so the note does not decide the question.
Results.
- Main theorem
(p. 1, unnumbered; proof pp. 1--2): the largest number f(n) of unique
subgraphs of a graph on n vertices satisfies log_2 f(n) = n^2/2 - n log_2 n
- O(n).
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.