Professional Documents
Culture Documents
I have wondered how the closed form for the sum of squares for the first n
natural numbers was derived. Given the formula for the sum,
Most textbooks would simply state those formulas, and prove them by
induction. Induction is, of course, a perfectly valid procedure; but it does
not give any insight into how on earth a person would ever have
dreamed the formula up in the first place, except by some lucky guess. …
Carl Friedrich Gauss used the following approach, apparently while still in
primary school, to derive a closed form for the sum of the first n positive
natural numbers, S = 1 + 2 + 3 + ⋯ + (n − 2) + (n − 1) + n. He first
reordered the sum as S ′ = n + (n − 1) + (n − 2) + ⋯ + 3 + 2 + 1, and
added S and S ′ , giving
S + S ′ = (1 + n) + (1 + n) + ⋯ + (1 + n) = n(1 + n). Since S = S ′ , we
get the sum as S = n(n + 1)/2. So, are there similarly simple derivations for
the sum of squares? As it turns out, there are several.
A simple derivation
Let S be the sum of squares of the first n natural numbers, such that
S = 12 + 22 + 32 + ⋯ + n2 .
Our aim is to derive a closed form formula for S in terms of n. We have
(x + 1)3 = (x + 1)(x2 + 2x + 1)
= x3 + 2x2 + x + x2 + 2x + 1
= x3 + 3x2 + 3x + 1.
23 + 33 + ⋯ + (n + 1)3 = 13 + 23 + ⋯ + n3 +
3S + 3n(n + 1)/2 + n
Some history
After studying further, I found out that there are several methods for deriving
this and similar results. In Concrete Mathematics by Graham, Knuth, and
Patashnik, Second Edition, pp. 41–46 [Addison-Wesley, December 2004],
several different approaches are discussed. Some involve simple
manipulations of sums, while others require integral calculus.
As the history goes, Archimedes derived the formula for the sum of squares a
long time ago. I have investigated the derivations and proofs in two principal
translations of Archimedes’ work: The Works of Archimedes by T. L. Heath
[Cambridge University Press, 1897], and Archimedes by E. J. Dijksterhuis
[Copenhagen: Ejnar Munksgaard, 1956].
to produce,
where a is the common difference. From this, Archimedes finally derives the
result Σni=1 i2 = n(n + 1)(2n + 1)/6 .
1. I have discussed the cited works in the order that I discovered them,
beginning with the bulletin message (http://mathforum.org/library/drmath
/view/56920.html) I read on 9 June 2010 Wednesday, 23:49↩
2. Did Gauss know about Archimedes’ work? In Dijksterhuis’ translation, a
diagrammatic argument is presented using two sequences of bars, each
representing a sequence of natural number, one ascending and the other
descending, which when combined produce a rectangle. It is easy to see
that this is a visual representation of Gauss’ derivation.↩
0 Comments Yaikhom
1 Login
Sort by Best
Recommend 6 ⤤ Share
Name
ALSO ON YAIKHOM