The round clay tablet YBC 7289 from around 1800 BC, with a square, its two diagonals, and cuneiform numbers giving the length of the diagonal
Unit 07 · All masterclasses · 70 min

Diophantine Equations

Whole-number solutions of equations: Pythagorean triples, descent, and Pell's equation
Image: Yale Peabody Museum of Natural History, Yale Babylonian Collection (CC0)

An equation becomes Diophantine when its solutions must be integers, and then even a simple one can take centuries to settle. This unit teaches the three classical methods: the parametrization of all solutions, shown on the Pythagorean triples and the rational points of the circle; Fermat's method of descent, with the full proof that x to the 4 plus y to the 4 equals z squared has no solution; and congruence obstructions such as the one that rules out y squared equals x cubed plus 7. It ends with Pell's equation, whose infinitely many solutions all grow out of the smallest one, and with a first look at the curves of degree three that current research is about.

Preview: the beginning of the unit

The equation \(x^2 + y^2 = z^2\) has infinitely many solutions in positive integers, and a clay tablet written around 1800 BC lists fifteen of them. The equation \(x^4 + y^4 = z^4\) has none, and the proof had to wait for the seventeenth century. Both are Diophantine equations: the unknowns are required to be integers, and that one restriction turns a question of school algebra into a question that can stay open for centuries. This unit teaches the three methods that have settled such questions since antiquity: to parametrize every solution (to give a formula that produces all of them), to descend from a supposed solution to a smaller one until nothing is left, and to reduce the equation modulo a well-chosen number. It ends at the equations of degree three, where the same questions are the subject of current research.

1 Equations in integers

Unit 2 called an equation Diophantine when its solutions are required to be integers, and it solved the linear case completely. The equation \(ax + by = c\) has integer solutions exactly when \(\gcd(a, b)\) divides \(c\), and once one solution is known, all the others come from it by adding multiples of \(b/\gcd(a, b)\) to \(x\) and subtracting the same multiples of \(a/\gcd(a, b)\) from \(y\). This unit is about equations of degree two and higher, and the first thing to settle is what it means to solve one.

Over the real numbers an equation such as \(x^2 + y^2 = z^2\) is not a problem at all, because any \(x\) and \(y\) can be completed by \(z = \sqrt{x^2 + y^2}\). Over the integers the same equation asks which right triangles have whole-number sides, and to solve a Diophantine equation means one of three things:

  1. to describe all integer solutions, usually by a formula with integer parameters that produces every solution exactly once (such a formula is called a parametrization);
  2. to prove that there is no solution at all, or none apart from obvious ones such as \(x = y = z = 0\);
  3. to find some solutions and to say whether there are finitely or infinitely many.

The methods of this unit match the three outcomes: a parametrization for \(x^2 + y^2 = z^2\) in Sections 2 and 3, descent and congruences for \(x^4 + y^4 = z^2\) and \(y^2 = x^3 + 7\) in Sections 4 and 5, and for Pell’s equation \(x^2 - dy^2 = 1\) in Section 6 a way to produce its infinitely many solutions from the smallest one.

No single method can replace these special arguments. In 1970 Yuri Matiyasevich (Steklov Institute of Mathematics, St Petersburg) proved that no algorithm can decide whether an arbitrary polynomial equation with integer coefficients has an integer solution, which settled the tenth of the 23 problems that David Hilbert (1862-1943) had posed in 1900. Every family of Diophantine equations therefore needs its own idea.

The full unit is part of the program

Want to see a complete unit first? Unit 3, Prime Numbers and Unique Factorization, is free to read.