Squared triangular number

Nicomachus, at the end of Chapter 20 of his Introduction to Arithmetic, pointed out that if one writes a list of the odd numbers, the first is the cube of 1, the sum of the next two is the cube of 2, the sum of the next three is the cube of 3, and so on. He does not go further than this, but from this it follows that the sum of the first n cubes equals the sum of the first ${\displaystyle n(n+1)/2}$ odd numbers, that is, the odd numbers from 1 to ${\displaystyle n(n+1)-1}$. The average of these numbers is obviously ${\displaystyle n(n+1)/2}$, and there are ${\displaystyle n(n+1)/2}$ of them, so their sum is ${\displaystyle [n(n+1)/2]^{2}.}$

Many early mathematicians have studied and provided proofs of Nicomachus's theorem. Stroeker (1995) claims that "every student of number theory surely must have marveled at this miraculous fact". Pengelley (2002) finds references to the identity not only in the works of Nicomachus in what is now Jordan in the first century CE, but also in those of Aryabhata in India in the fifth century, and in those of Al-Karaji circa 1000 in Persia. Bressoud (2004) mentions several additional early mathematical works on this formula, by Al-Qabisi (tenth century Arabia), Gersonides (circa 1300 France), and Nilakantha Somayaji (circa 1500 India); he reproduces Nilakantha's visual proof.

The sequence of squared triangular numbers is

0, 1, 9, 36, 100, 225, 441, 784, 1296, 2025, 3025, 4356, 6084, 8281, ... (sequence A000537 in the OEIS).

These numbers can be viewed as figurate numbers, a four-dimensional hyperpyramidal generalization of the triangular numbers and square pyramidal numbers.

As Stein (1971) observes, these numbers also count the number of rectangles with horizontal and vertical sides formed in an n × n grid. For instance, the points of a 4 × 4 grid (or a square made up of three smaller squares on a side) can form 36 different rectangles. The number of squares in a square grid is similarly counted by the square pyramidal numbers.

The identity also admits a natural probabilistic interpretation as follows. Let X, Y, Z, W be four integer numbers independently and uniformly chosen at random between 1 and n. Then, the probability that W be the largest of the four numbers is equal to the probability that both Y is at least as large as X and W is at least as large as Z, that is, P({max(X,Y,Z) ≤ W}) = P({X ≤ Y} ∩ {Z ≤ W}). These probabilities are respectively the left and right sides of the Nichomacus identity, normalized to make probabilities by dividing both sides by n⁴.

Charles Wheatstone (1854) gives a particularly simple derivation, by expanding each cube in the sum into a set of consecutive odd numbers. He begins by giving the identity

${\displaystyle n^{3}=\underbrace {\left(n^{2}-n+1\right)+\left(n^{2}-n+1+2\right)+\left(n^{2}-n+1+4\right)+\cdots +\left(n^{2}+n-1\right)} _{n{\text{ consecutive odd numbers}}}.}$

That identity is related to triangular numbers ${\displaystyle T_{n}}$ in the following way:

${\displaystyle n^{3}=\sum _{k=T_{n-1}+1}^{T_{n}}(2k-1),}$

and thus the summands forming ${\displaystyle n^{3}}$ start off just after those forming all previous values ${\displaystyle 1^{3}}$ up to ${\displaystyle (n-1)^{3}}$. Applying this property, along with another well-known identity:

${\displaystyle n^{2}=\sum _{k=1}^{n}(2k-1),}$

we obtain the following derivation:

${\displaystyle {\begin{aligned}\sum _{k=1}^{n}k^{3}&=1+8+27+64+\cdots +n^{3}\\&=\underbrace {1} _{1^{3}}+\underbrace {3+5} _{2^{3}}+\underbrace {7+9+11} _{3^{3}}+\underbrace {13+15+17+19} _{4^{3}}+\cdots +\underbrace {\left(n^{2}-n+1\right)+\cdots +\left(n^{2}+n-1\right)} _{n^{3}}\\&=\underbrace {\underbrace {\underbrace {\underbrace {1} _{1^{2}}+3} _{2^{2}}+5} _{3^{2}}+\cdots +\left(n^{2}+n-1\right)} _{\left({\frac {n^{2}+n}{2}}\right)^{2}}\\&=(1+2+\cdots +n)^{2}\\&={\bigg (}\sum _{k=1}^{n}k{\bigg )}^{2}.\end{aligned}}}$

Row (1893) obtains another proof by summing the numbers in a square multiplication table in two different ways. The sum of the ${\displaystyle i}$th row is ${\displaystyle i}$ times a triangular number, from which it follows that the sum of all the rows is the square of a triangular number. Alternatively, one can decompose the table into a sequence of nested gnomons, each consisting of the products in which the larger of the two terms is some fixed value. The sum within each gmonon is a cube, so the sum of the whole table is a sum of cubes.

Visual demonstration that the square of a triangular number equals a sum of cubes.

In the more recent mathematical literature, Edmonds (1957) provides a proof using summation by parts. Stein (1971) uses the rectangle-counting interpretation of these numbers to form a geometric proof of the identity (see also Benjamin, Quinn & Wurtz 2006); he observes that it may also be proved easily (but uninformatively) by induction, and states that Toeplitz (1963) provides "an interesting old Arabic proof". Kanim (2004) provides a purely visual proof, Benjamin & Orrison (2002) provide two additional proofs, and Nelsen (1993) gives seven geometric proofs.

A similar result to Nicomachus's theorem holds for all power sums, namely that odd power sums (sums of odd powers) are a polynomial in triangular numbers. These are called Faulhaber polynomials, of which the sum of cubes is the simplest and most elegant example. However, in no other case is one power sum a square of another (Edmonds 1957).

Stroeker (1995) studies more general conditions under which the sum of a consecutive sequence of cubes forms a square. Garrett & Hummel (2004) and Warnaar (2004) study polynomial analogues of the square triangular number formula, in which series of polynomials add to the square of another polynomial.

• Weisstein, Eric W. "Nicomachus's theorem". MathWorld.
• A visual proof of Nicomachus's theorem