Interesting Math

Sums of Squares of Consecutive Integers

A couple issues ago, I posed the following question:

We know 3² + 4² = 5²; notice 3, 4, 5 is consecutive and there is 1 more term on the left. Can we extend this idea? In particular, does there exist n + 1 consecutive positive integers whose squares sum to the sum of squares of the next n consecutive integers?

This will be a more technical article than usual, but it is such an interesting problem I decided to go over it. If you tried this out, you might have discovered that it certainly holds for n = 2 as well, since 10² + 11² + 12² = 13² + 14² . To solve this problem in general, we want to solve for a such that

Include graphic InterestingMath0.eps here

where I have rewritten the sum on the right to make it the difference between "full" sums starting from 0. Rearranging and expanding the squares, this is simply

Include graphic InterestingMath1.eps here

You may recall (or rederive) the formulae for sums of consecutive integers or consecutive squares:

Include graphic InterestingMath2.eps here

After substituting and simplifying, we end up with

Include graphic InterestingMath3.eps here

which is a qudratic in a. Solving, we find two solutions: a = -n and a = n(2n + 1). The first solution makes sense, since we will have -n, -n+1, ... 0 as the first n + 1 consecutive integers, which trivially has sum of squares equal to the sum of squares of the next n consecutive integers. Since I wanted positive integers, the second solution will answer the problem. This is quite an interesting result: for any n, there exist n + 1 consecutive positive integers whose squares sum to the sum of squares of the next n consecutive integers!

Can we extend this result to higher powers? For example, 3³ + 4³ + 5³ = 6³. Then in general, does there exist n + 2 consecutive positive integers whose cubes sum to the sum of cubes of the next n consecutive integers? Unfortunately, this is not true. The result fails for the next value of n, n = 2. Indeed, suppose there is an integer a with

Include graphic InterestingMath4.eps here

Then this equation holds in every modulus, in particular, in mod 4. However, the first two terms cancel in mod 4, leaving the sum of two consecutive cubes is 0 in mod 4, which never happens (the sum of consecutive powers in mod 4 is either 1 or -1, for any positive power). Oh well, we tried. Maybe there's another generalization to be found — but that's an article for another time.

Vince's problem of the issue

Let S be the set of integers n such that some factorization of n into factors a₁,...,a_n satisfy a₁ + ... + a_n = 2009. What is the maximum element of S?

Vince Chan
v2chan@math.uwaterloo.ca