wrote a book called Arithmetic, consisiting of problems and theor solutions which can today be formulated as polynomial equations in one or more variables with integer coefficients; all the solutions to these equations were meant to be positive integers (or sometimes, rational numbers). (It is important to note that these equations were not algebraicalgebraic notation and symbolic methods were not to be invented for some 700 yearsand that there was a long history of similar problem solving known to Babylonian mathematics.) From the tradition of Diophantus, we call any equation whose solutions are meant to be integers a Diophantine equation. A very important class of Diophantine equations are of linear type, and the simplest nontrivial equation of this type has the form Given integers a, b, c, find integers x, y so that ax + by = c. Our discussion of how the gcd (a, b) has the form ax + by for a suitable choice of x, y suggests that c must be related to (a, b). This is indeed the case.
Brahmaguptas Theorem Given integers a, b, c,
the Diophantine equation (*)
ax + by = c
has no solutions unless (a, b)|c, in which case
solutions to (*) are identical with solutions to a b c x+ y= . (a,b) (b,c) (a,b)
(**)
If x = x0, y = y0 is any one solution to (**), then all
solutions have the form
x = x0 +
b a k, y = y 0 k for any integer k. (a,b) (a,b)
Proof If (*) has a solution, then since (a, b) divides
the left side of the equation, it must divide the right: (a, b)|c. When this condition holds, dividing through (*) by (a, b) yields (**), so both equations have the same set of solutions. Putting a b c a = , b = , c = , (a,b) (a,b) (a,b) note that since a and b are relatively prime, we can, by the extended Euclidean algorithm, find integers X, Y so that aX + bY = 1. It follows that x0 =cX , y0 = cY satisfy (**). Thus, (**) is solvable.
If x = x0, y = y0 is any one solution to (**) and
x = x1, y = y1 is another, then subtracting these two equations gives the relation a(x 0 - x1 ) = -b( y0 - y1 ). And since a and b have no common factors, we conclude that a | (y 0 - y1 ), b | (x0 - x1 ). It follows that y1 = y0 - ka , and so also x1 = x0 + kb . //
What is more, in light of the extended Euclidean
algorithm, the theorem suggests how to compute
the solution setto the Diophantine equation ax +by = c. We illustrate with an example of a problem from Euler:
A farmer lays out the sum of 1770 crowns in purchasing horses
and oxen. He pays 31 crowns for each horse and 21 crowns for each ox. How many horses and oxen did the farmer buy?
With x = number of horses and y = number of oxen,
the problem becomes one of solving the Diophantine equation 31x + 21y = 1770. Since (31,21) = 1, we compute, by the extended Euclidean algorithm: 31 1 0 21 0 1 1 10 1 -1 2 1 -2 3 whence 1 = -231 + 3 21 and by multiplying through by 1770 we get 1770 = -3540 31+ 5310 21.
Therefore, x0 = 3540, y0 = 5310 is one solution.
The entire solution set has the form x = -3540 + 21k, y = 5310 - 31k where k is any integer. Fig 2.6.1 (p. 49) for an indication of the See geometric interpretation of this solution. Since the equation is a Diophantine one, all solutions points of the integer lattice in the plane that lie on the line 31x +21y =1770. It is also sometimes desirable to restrict attention to positive integer solutions only. (Diophantus does this in the Arithmetic.) From the general solution, it is a simple addition of inequality conditions that resolves this extra requirement. In the example we worked out above, this proceeds thusly: -3540 + 21k 0 21k 3540 k 168.57...
5310 - 31k 0 5310 31k 171.29... k
Consequently, since k must be an integer, k = 169,
170, or 171, leading to precisely three solutions: