Ivars Peterson's MathTrek
September 27, 2004
In 1769, while thinking about the problem now known as Fermat's last theorem, Leonhard Euler (17071783) proposed an intriguing variant.
A century earlier, Pierre de Fermat (16011665) had proved that no set of positive integers, a, b, and c, satisfies the equation a4 + b4 = c4 in the same way that numbers such as 3, 4, and 5 satisfy the more familiar equation x2 + y2 = z2. Euler conjectured that the equation a4 + b4 + c4 = d4 would also have no integer solutions.
Indeed, Euler went further. He proposed that for every integer greater than 2, the sum of n 1 nth powers of positive integers cannot itself be an nth power.
In 1966, L.J. Lander and T.R. Parkin disproved Euler's generalized conjecture when they found a counterexample for the case n = 5. The equation a5 + b5 + c5 + d5 = e5 is true when a = 27, b = 84, c = 110, d = 133, e = 144.
In 1986, Noam D. Elkies of Harvard University found the first counterexample that proved Euler's conjecture was wrong for the case n = 4. The equation is true when a = 2,682,440, b = 15,365,639, c = 18,796,760, and d = 20,615,673.
Elkies discovered the counterexample by a method that combined theoretical reasoning and a relatively short computer search. He converted the problem into an equivalent mathematical form that enabled him to pick out candidates likely to satisfy Euler's equation. A few mathematicians had tried a similar approach previously, but either they had given up too soon or they had been thinking in terms of proving rather than disproving the conjecture.
Once Elkies found the first counterexample, he was able to prove that there are infinitely many, each consisting of enormously large numbers. What Elkies didn't know at the time was whether his initial solution was the smallest one.
When Roger Frye, then at Thinking Machines in Cambridge, Mass., heard news of Elkies's achievement, he tackled the problem himself. Using hints supplied by Elkies to shorten the search, he wrote a computer program to look for smaller solutions.
Working at night in his spare time, Frye used Connection Machine computers to find the smallest positive integers that fit the equation. His exhaustive computer search showed that a = 95,800, b = 217,519, c = 414,560, and d = 422,481.
Subsequent computer searches turned up additional examples. In 2000, for d less than 2.1 x 107, the total stood at seven, with values for d of 422,481, 2,813,001, 8,707,481, 12,197,457, 16,003,017, 16,430,513, and 20,615,673.
That still leaves lots of problems and conjectures about sums of powers yet unsolved. For example, no one has found a counterexample to Euler's conjecture for any power greater than 5. Can a sixth power be expressed as the sum of five sixth powers?
Anyone with a computer interested in investigating "minimal equal sums of like powers" can join in the fun (see http://euler.free.fr/index.htm). Recently, such a computer search turned up a second solution for the case n = 5: 853595 = 852825 + 289695 + 31835 + 555.
And what about the sum of four fourth powers expressed as a fourth power? The list goes on and on.
Copyright © 2004 by Ivars Peterson
Bernstein, D.J. 2001. Enumerating solutions to p(p) + p(p) = p(p) + p(p). Mathematics of Computation 70(No. 233):389-394. Abstract available at http://www.ams.org/mcom/2001-70-233/S0025-5718-00-01219-9/home.html.
Elkies, N.D. 1988. On A4 + B4 + C4 = D4. Mathematics of Computation 51:825-835.
Frye, R.E. 1988. Finding 95,8004 + 217,5194 + 414,5604 = 422,4814 on the Connection Machine. In Supercomputing '88: Science and Applications (Proceedings), J.L. Martin and S.F. Lundstrom, eds. Silver Spring, Md.: IEEE Computer Society Press.
Lander, L.J., and T.R. Parkin. 1967. A counterexample to Euler's sum of powers conjecture. Mathematics of Computation 21:101-103.
Peterson, I. 1988. The curious power of large numbers. Science News 133(Jan. 30):70.
Information about a massive effort to compute minimal equal sums of like powers can be found at http://euler.free.fr/index.htm.
A collection of Ivars Peterson's early MathTrek articles, updated and illustrated, is now available as the Mathematical Association of America (MAA) book Mathematical Treks: From Surreal Numbers to Magic Circles. See http://www.maa.org/pubs/books/mtr.html.