Some Problems of Diophantus
Franz Lemmermeyer
December 21, 2003
It is believed that Diophantus worked around 250 AD; apart from this we
think that he lived for 84 years, since a puzzle given by by Metrodorus around
500 AD says
his boyhood lasted 1/6th of his life;
he married after 1/7th more;
his beard grew after 1/12th more,
and his son was born 5 years later;
the son lived to half his father’s age,
and the father died 4 years after the son.
This gives rise to a linear equation in Diophantus’ age x (much simpler than
anything Diophantus has done) with x = 84 as the solution.
Diophantus’ main claim to fame rests on his book “Arithmetika”, which
consists of 13 parts. Six of them were known since Fermat’s times, another
four have been discovered in Arabic translation. In these books, Diophantus
solves “indeterminate equations”: a determinate equation is an equation that
determines the solutions, like x2 − 2x + 3 = 0; an example of an indeterminate
equation is x2 + y 2 = 1, which has many rational solutions.
Diophantus invented algebraic notation: he had a symbol for one unknown x,
and other symbols for x2 , . . . , x6 . This bold step was unsurpassed until Viète
(1540–1603) improved upon it by using vowels for unknown and consonants
for known quantitites; Descartes later introduced the modern variant where
unknowns are denoted by x, y, z and known quantities by a, b, c . . . .
The Problems
One of the most famous problems that Diophantus treated was writing a square
as the sum of two squares (book II, problem 8):
To divide a given square into a sum of two squares.
This problem became important when Fermat, in his copy of Diophantus’ Arith-
metica edited by Bachet, noted that he had this wonderful proof that cubes can’t
be written as a sum of two cubes, fourth powers not as a sum of two fourth pow-
ers, and so on, but that the margin of this book was too small to contain it.
(This observation was later published by his son Samuel; Fermat publicly only
claimed to have proofs for exponents 3 and 4).
1
Diophantus’ Solution
Diophantus’ solution proceeds as follows: since his notation allows only one
unknown, he cannot treat the general case, so he starts by assuming that the
given square is 16; letting x2 denote one of the squares whose sum is 16, the
other square must be 16 − x2 . He writes this square in the form (2x − 4)2 , and
then has to solve 16 − x2 = (2x − 4)2 = 4x2 − 16x + 16, which gives 5x2 = 16x,
hence 5x = 16 (Diophantus has no problems with canceling x since he doesn’t
know 0). Now x = 16 12 256 144 400
5 , so 2x − 4 = 5 , and in fact 25 + 25 = 25 = 16.
Diophantus is aware of the fact that his method produces many more solu-
tions: he writes
I form the square from any number of x minus as many units as
there are in the side of 16.
The side of 16 is the square root of 16 (think of 16 as the area of square),
so in modern terms his statement means: instead of 2x − 4, the substitution
mx − 4 will produce a rational solution (of course m has to be chosen positive
and rational).
Solution with modern algebraic notation
Now let us solve the problem x2 + y 2 = a2 , where a is a given positive rational
number, using modern notation. Diophantus’ idea is to write y = mx − a (for
us, mx + a would work equally well), which gives a2 = x2 + m2 x2 − 2amx + a2 ,
2am
that is, x2 (1 + m2 ) = 2amx. Thus either x = 0 or x = 1+m 2 . Note that the
2am
substitution y = mx + a would have given x = − 1+m2 , which, for a and m
positive rational numbers, would produce negative values. Thus Diophantus
had good reasons for picking y = mx − a instead: he does not know negative
numbers.
Modern Geometric Interpretation
It has been discovered at the beginning of the 20th century (the earliest instance
I could find was a book of Klein; but there is a much more general theorem on
‘curves of genus 0’ containing all this and more by Hilbert and Hurwitz from
the 1890’s) that there is a geometric interpretation of the substitutions that
Diophantus (and his ‘modern students’, like Fermat, Euler, Lagrange, Cauchy,
Lucas, Sylvester etc.) was using.
Here’s how: consider the points (x, y) ∈ R × R satisfying x2 + y 2 = a2 ; this
circle has an obvious rational point, namely P = (0, −a). Now consider the lines
through this point P ; their equations are y = mx − a (with the exception of
the tangent at P , which has ‘slope ∞’ and is described by x = a). Intersecting
this line with the circle will give two points of intersection, one of them being
– of course – P . Let’s compute the second point: plugging the equation for the
line into x2 + y 2 = a2 and solving for x gives x2 (m2 + 1) = 2amx; the solution
x = 0 corresponds to the original point P = (0, −a), the x-coordinate of the
2
2m
second point of intersection is x = a 1+m 2. Plugging this into our equation
2
−1
y = mx − a gives y = a m m2 +1 . Since a is rational, this means that whenever
we plug in a rational number for m, these formulas gives us a rational point
satisfying x2 + y 2 = a2 . In fact, it is quite easy to see that conversely, every
rational point 6= P corresponds to some rational value of m.
This is a very neat 20th-century solution of Diophantus’ problem, and tech-
niques such as this one (actually, the techniques have become much mure sophis-
ticated during the last 100 years) are today subsumed under the title ‘arithmetic
geometry’.
Another Problem
Consider the following problem:
To add the same number to two given numbers so as to make each
of them a square.
Again, lack of algebraic notation forces Diophantus to explain the solution using
an example, and he picks the numbers 2 and 3 for this. Then he has to find a
number x such that 2 + x and 3 + x are squares. Using modern notation, let us
assume that 2+x = u2 and 3+x = v 2 ; then 1 = 3−2 = v 2 −u2 = (v −u)(v +u).
Now put v + u = 4; then v − u = 14 , and we find x = 97 64 , as well as 2 + x =
225 15 2 289 17 2
64 = ( 8 ) and 3 + x = 64 = ( 8 ) .
Observe that the equation 1 = v 2 − u2 describes a hyperbola; modern math-
ematicians could use the technique of sweeping lines to determine all solutions
of this (and related) diophantine equations.
